CN120278705A - Method and system for analyzing equipment state data applied to automatic electric - Google Patents

Abstract

The present invention provides a method and system for analyzing equipment status data applied to automated electrical equipment, the method comprising: obtaining a historical operation data set of a target equipment, constructing a multi-level status analysis model based on the historical operation data set, the multi-level status analysis model comprising at least one first analysis level and at least one second analysis level, extracting key operation parameters from the real-time operation data stream of the target equipment through an operation parameter screening strategy in the first analysis level, and generating parameter screening results according to the key operation parameters; inputting the parameter screening results into a status evaluation strategy in the second analysis level, combining the dynamic threshold range corresponding to the key operation parameters, generating a status evaluation index of the target equipment; outputting the real-time operation status level of the target equipment according to the comparison result of the status evaluation index and the preset status threshold. The present invention can realize efficient monitoring and accurate early warning of the health status of the equipment.

Description

Method and system for analyzing equipment state data applied to automatic electric

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a method and a system for analyzing equipment status data applied to automated electric equipment.

Background

State data analysis of automated electrical equipment is one of the cores in the industrial operation and maintenance field, aiming at identifying potential faults and optimizing maintenance strategies by monitoring equipment operating parameters in real time. In the prior art, the state of equipment is usually subjected to binary evaluation directly based on a preset static threshold, for example, after key parameters are screened through fixed rules or manual experience, the selected parameters are compared with a constant threshold to judge the state grade of the equipment, or a general machine learning model is adopted to carry out indiscriminate training on all the parameters to generate a state classification result, however, the static threshold cannot be adapted to the dynamic change characteristics of the parameters of the equipment in different operation stages and under load conditions, so that normal fluctuation is misjudged to be an abnormal state or true abnormality is missed, secondly, the coupling degree of the parameter screening and the state evaluation links is too high, a hierarchical processing mechanism is not established, so that the evaluation precision of redundant parameters is interfered, the response speed of the model is limited by high dimensional processing pressure, moreover, the traditional method lacks consideration of time sequence relevance and weight dynamic allocation among the parameters, key feature change in the degradation process of the equipment cannot be accurately captured, so that the maintenance decision lacks look-ahead performance and precision, the problem of excessive maintenance or maintenance hysteresis is caused under complex working conditions, the operation efficiency of the equipment is seriously affected, and the operation cost is increased.

Disclosure of Invention

In view of the above, the present invention provides a method and a system for analyzing equipment status data applied to automated electric equipment. The technical scheme of the invention is realized as follows:

The invention provides a device state data analysis method applied to automated electric, which comprises the steps of obtaining a historical operation data set of target devices, wherein the historical operation data set comprises a historical data sequence of various device operation parameters and corresponding device state labels, constructing a multi-level state analysis model based on the historical operation data set, wherein the multi-level state analysis model comprises at least one first analysis level and at least one second analysis level, each first analysis level corresponds to an operation parameter screening strategy, each second analysis level corresponds to a state evaluation strategy, the second analysis level is located behind the first analysis level, extracting key operation parameters from a real-time operation data stream of the target devices through the operation parameter screening strategy in the first analysis level, generating parameter screening results according to the key operation parameters, inputting the parameter screening results into the state evaluation strategy in the second analysis level, combining a dynamic threshold range corresponding to the key operation parameters, generating a state evaluation index of the target devices, comparing the state index with a preset state threshold, and outputting the state evaluation index of the target devices in real-time according to the state evaluation results.

In another aspect, the invention provides a data analysis system comprising a memory and a processor, the memory storing a computer program executable on the processor, the processor implementing the steps of the method described above when the program is executed.

The method for analyzing the state data of the equipment applied to the automated electric system, provided by the invention, screens key operation parameters and generates parameter screening results through a first analysis level in a multi-level state analysis model, generates a state evaluation index of target equipment by combining a state evaluation strategy based on a dynamic threshold range in a second analysis level, finally outputs a real-time operation state level by comparing with a preset state threshold, can effectively solve the problems of low calculation efficiency and high misjudgment rate caused by poor static threshold adaptability and parameter redundancy in the state analysis of the traditional equipment, wherein the first analysis level dynamically extracts the key operation parameters and distributes weights through parameter correlation screening conditions and stability indexes, ensures that the parameter screening results input to the second analysis level are focused on core data which has obvious influence on the state of the equipment, reduces irrelevant parameter interference, and the second analysis level dynamically adjusts a threshold boundary by combining a historical operation data distribution interval and a real-time operation stage, enables the state evaluation index to accurately reflect the degradation trend of the equipment under different operation environments, realizes closed loop logic chains from the parameter screening to the state evaluation through sequential dependence and a data cooperation mechanism of the first analysis level and the second analysis level, improves the reliability and the reliability, simultaneously can accurately and accurately control the failure rate of the equipment by detecting the failure level by the dynamic threshold, reduces the failure level and the failure level, and reduces the real-time maintenance level by the failure level, and the failure level detection, the method realizes high-efficiency monitoring and accurate early warning of the health state of the equipment in a complex industrial scene.

Drawings

Fig. 1 is a schematic implementation flow chart of a method for analyzing equipment status data applied to an automated electric system according to an embodiment of the present invention.

Fig. 2 is a schematic diagram of a hardware entity of a data analysis system according to an embodiment of the present invention.

Detailed Description

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is to be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments and can be combined with one another without conflict. The term "first/second/third" is merely to distinguish similar objects and does not represent a particular ordering of objects, it being understood that the "first/second/third" may be interchanged with a particular order or precedence, as allowed, to enable embodiments of the invention described herein to be implemented in other than those illustrated or described herein.

Embodiments of the present invention provide a method for analyzing device state data for use in automated electrical applications, which may be performed by a processor of a data analysis system. The data analysis system may refer to a server, a notebook computer, a tablet computer, a desktop computer, or other devices with data processing capability.

Fig. 1 is a schematic implementation flow chart of a method for analyzing equipment status data applied to an automated electric system according to an embodiment of the present invention, as shown in fig. 1, the method includes the following steps:

Step S100, a historical operation data set of the target equipment is obtained, wherein the historical operation data set comprises historical data sequences of various equipment operation parameters and corresponding equipment state labels.

The operating parameters of the device refer to various physical quantities or indicators, such as voltage, current, temperature, rotational speed, etc., that are capable of reflecting the operating state of the device during operation of the device. The historical data sequence is a collection of data in which the operating parameters of the device were recorded in chronological order over a period of time. The device status tag classifies and identifies the operating status of the device at each point in time, such as normal, abnormal, faulty, etc.

In the embodiment of the invention, the acquisition of the historical operation data set of the target device is the basis for carrying out subsequent device state analysis. The device operating parameters may be collected in real time by various sensors mounted on the device and stored in a database. Over time, this data forms a historical operating dataset. For example, for an electric motor, current, temperature and rotation speed data during operation can be collected by installing a current sensor, a temperature sensor and a rotation speed sensor, and corresponding equipment state labels such as normal operation, overload operation, faults and the like are marked for the data of each time point.

Step 200, constructing a multi-level state analysis model based on the historical operation data set, wherein the multi-level state analysis model comprises at least one first analysis level and at least one second analysis level, each first analysis level corresponds to an operation parameter screening strategy, each second analysis level corresponds to a state evaluation strategy, and the second analysis level is located behind the first analysis level.

The operating parameter screening strategy refers to a method of screening a number of plant operating parameters for key operating parameters that have an impact on plant condition analysis. The state evaluation strategy is a method for evaluating and judging the current state of the equipment according to the screened key operation parameters.

The first analysis level can remove parameters with smaller influence on the state of equipment through an operation parameter screening strategy, so that the complexity of data processing is reduced, and the analysis efficiency is improved. The second analysis level utilizes a state evaluation strategy to evaluate the state of the equipment by combining the key operation parameters to obtain a more accurate state evaluation result.

The first analysis level may use a correlation analysis method to screen the key operation parameters, calculate a degree of correlation between each operation parameter and the device status label, and select a parameter with a higher degree of correlation as the key operation parameter. The second analysis level may employ any feasible machine learning model, such as a support vector machine, neural network, etc., to classify and evaluate device states based on key operating parameters.

And step 300, extracting key operation parameters from the real-time operation data stream of the target equipment through an operation parameter screening strategy in the first analysis level, and generating a parameter screening result according to the key operation parameters.

The real-time operation data stream refers to device operation parameter data generated by the target device in real time in the current operation process. The key operation parameters are parameters which are screened from the real-time operation data stream and have important significance for equipment state analysis. The parameter screening results are related information including key operating parameters, such as real-time measured values of parameters, parameter weights, and the like.

In the step, the real-time operation data stream is processed through the operation parameter screening strategy, so that the key operation parameters can be extracted rapidly and accurately. The key operation parameters can reflect the current operation state of the equipment more effectively, and provide important basis for subsequent state evaluation.

For example, for a transformer in an electrical power system, the real-time operating data stream contains a number of parameters, such as voltage, current, oil temperature, winding temperature, etc. Through the operation parameter screening strategy of the first analysis level, the association degree between the voltage and the oil temperature and the state of the transformer is found to be higher, so that the voltage and the oil temperature are extracted as key operation parameters, and a parameter screening result is generated according to the real-time measured values and the related calculation of the voltage and the oil temperature.

As an embodiment, the step S300 may specifically include the steps of:

step S310, intercepting an operation parameter sequence in a current time window from a real-time operation data stream, wherein the operation parameter sequence comprises real-time measurement values of a plurality of equipment operation parameters.

A time window refers to a range of time periods in which data is selected in a real-time operational data stream. The sequence of operating parameters is a sequence of time-sequentially ordered real-time measurements of a plurality of plant operating parameters acquired within the time window.

In practical applications, since the real-time operation data stream is continuously generated, it is necessary to intercept data within a suitable time window for easy analysis and processing. For example, the time window may be set to 1 minute, and real-time measurements of the operating parameters of the device, such as voltage, current, temperature, etc., within the last 1 minute may be taken from the real-time operating data stream to form an operating parameter sequence.

Step S320, calculating a relevance coefficient of each of the plurality of equipment operation parameters and the equipment state label in the historical operation data set based on the parameter relevance screening conditions in the first analysis level.

The parameter correlation screening conditions are rules or criteria set in advance for evaluating the correlation between the operating parameters and the equipment status tags. The association coefficient is a numerical value used for measuring the association degree between each operation parameter and the equipment state label, and the larger the association coefficient is, the closer the association between the operation parameter and the equipment state is.

In this step, the degree of influence of each operating parameter on the state of the apparatus can be quantitatively evaluated by calculating the correlation coefficient. For example, the relevance coefficient may be calculated using pearson correlation coefficient, mutual information, or the like. For example, for an industrial robot, the operation parameters include joint angle, motor current, load torque, etc., and by calculating the correlation coefficient between these parameters and the device status label (such as normal, fault, etc.) in the historical operation data set, it can be determined which parameters are more important for the status evaluation of the robot.

As an embodiment, the step S320 may specifically include the following steps:

Step S321, extracting data distribution characteristics of each equipment operation parameter under different equipment state labels from the historical operation data set, wherein the data distribution characteristics comprise parameter mean values, variances and distribution form indexes.

The parameter average value refers to the average value of all historical data of the running parameters of the equipment under a certain equipment state label. The variance is a statistic that measures the degree of dispersion of data and reflects the fluctuation of the data relative to the mean. The distribution morphology index is used to describe the distribution shape of the data, such as skewness, kurtosis, and the like.

In this step, extracting the data distribution characteristics helps to understand the change rule of each device operation parameter under different device states. For example, in a normal operation state, the mean value of the spindle rotation speed parameter of a numerical control machine tool may be stabilized near a set value, and the variance is small, while in a fault state, the mean value may be significantly changed, and the variance may be increased. By analyzing the data distribution characteristics, an important basis can be provided for the subsequent calculation of the relevance coefficient.

Step S322, calculating a distinguishing degree value of each equipment operation parameter among different equipment state labels according to the data distribution characteristics, wherein the distinguishing degree value is determined based on the combination of parameter mean value difference and variance ratio.

The differentiation value is an indicator that measures the ability of one device operating parameter to differentiate between different device status labels. The parameter mean value difference reflects the difference between the mean values of the parameters under different equipment states, and the variance ratio takes the discrete degree of the data into consideration. By combining the parameter mean difference with the variance ratio, a more comprehensive discrimination value can be obtained.

In actual calculation, the discrimination value may be calculated using the formula discrimination value = (mean 1-mean 2)/(variance 1+ variance 2), where mean 1 and mean 2 are the mean of the parameter under two different device state labels, respectively, and variance 1 and variance 2 are the corresponding variances, respectively. For example, as for a flow rate parameter of one water pump, in a normal operation state and a failure operation state, by calculating a discrimination value thereof, it can be judged whether the parameter can effectively discriminate the two states.

Step S323, obtaining a parameter change trend curve of each equipment operation parameter in the historical operation data set, and extracting the number of peak points and the positions of trend inflection points in the parameter change trend curve.

The parameter trend curve is a curve formed by connecting historical data of operating parameters of equipment in time sequence, and reflects the trend of the parameters along with time. The peak point is the point of the local maximum in the curve, and the trend inflection point is the point where the slope of the curve changes.

Extracting the number of peak points and the positions of the trend inflection points helps to find abnormal changes in the operating parameters of the device. For example, for a wind power generator power parameter, a sudden increase in the number of peak points in the parameter trend curve or an abnormal shift in the trend inflection point position may mean that the generator is malfunctioning or is affected by the external environment.

Step S324, generating a relevance coefficient by combining the distinguishing value, the peak point number and the trend inflection point position, wherein the relevance coefficient is positively correlated with the distinguishing value and the peak point number and negatively correlated with the dispersion of the trend inflection point position.

The generation of the association coefficient comprehensively considers a plurality of factors such as a discrimination value, the number of peak points, the position of a trend inflection point and the like. The larger the discrimination value, the stronger the discrimination capability of the parameter between different device states, and the larger the influence on the device state, so that the correlation coefficient is positively correlated with the discrimination value. The greater the number of peak points, the more dramatic the change that may be indicative of the parameter, and the more closely associated with the device state, the more closely associated the association coefficient is with the number of peak points. The smaller the dispersion of the trend inflection point position, the more stable the variation trend of the parameter is, and the higher the degree of association with the equipment state is, so that the degree of association coefficient is inversely related to the dispersion of the trend inflection point position. In practical calculation, a weighted summation method can be adopted to generate a relevance coefficient, namely the relevance coefficient=a×differential value+b×peak point number-c×dispersion of trend inflection point positions, wherein a, b and c are preset weight coefficients.

And step S325, carrying out normalization processing on the relevancy coefficient, and carrying out dynamic matching verification on the normalized relevancy coefficient and time sequence change of the equipment state label in the historical operation data set.

The normalization process converts the correlation coefficient to a preset interval, such as [0, 1], so as to facilitate comparison and analysis between different parameters. And the dynamic matching verification is to compare the normalized relevancy coefficient with the time sequence change of the equipment state label in the historical operation data set and check whether the relevancy coefficient can accurately reflect the change of the equipment state.

In this step, the normalization process may employ a common normalization method, such as min-max normalization, Z-score normalization, and the like. Through dynamic matching verification, the accuracy and the reliability of the relevance coefficient can be verified. For example, if the normalized relevance coefficient can be adjusted correspondingly in time when the equipment state changes, the relevance coefficient can reflect the change of the equipment state better, and the accuracy and the reliability are higher.

And step S330, screening candidate operation parameters with the association coefficient larger than a preset association threshold value from the operation parameters of the plurality of devices, and determining a parameter stability index according to the fluctuation amplitude of the real-time measured value of the candidate operation parameters.

The preset association threshold is a preset critical value, and is used for judging whether the association degree of an operation parameter and the equipment state is strong enough or not. The candidate operation parameters are operation parameters with the association coefficient larger than a preset association threshold. The parameter stability index is an index for measuring the fluctuation condition of the real-time measured value of the candidate operation parameter, and the smaller the fluctuation amplitude is, the higher the parameter stability is.

In the step, the data volume of subsequent processing can be reduced and the analysis efficiency can be improved by screening candidate operation parameters. And determining parameter stability indexes according to the fluctuation amplitude of the real-time measured value of the candidate operating parameters, so that the stability of the parameters can be known, and an important basis is provided for subsequent parameter sequencing and key operating parameter selection. For example, for operating parameters collected by a plurality of sensors in an automated production line, the parameter stability index of the operating parameters is determined by screening candidate operating parameters with a correlation coefficient greater than a preset correlation threshold, and then calculating the fluctuation amplitude of real-time measured values of the candidate operating parameters.

Step S340, sorting the candidate operation parameters according to the parameter stability index, and selecting the preset number of candidate operation parameters sorted in front as key operation parameters.

The preset number is the preset number of key operation parameters to be selected. By sorting the candidate operating parameters according to the parameter stability index, the parameter with higher stability can be preferentially selected as the key operating parameter.

In practical application, the parameter with higher stability can reflect the running state of the equipment more accurately, and errors caused by parameter fluctuation are reduced. For example, for a plurality of monitoring parameters in a power system, after ranking candidate operating parameters according to parameter stability indicators, selecting the top 5 candidate operating parameters as key operating parameters for subsequent equipment state analysis.

And step 350, generating a parameter screening result comprising parameter weight distribution based on the real-time measured value of the key operation parameter, wherein the parameter weight is dynamically adjusted according to the association coefficient and the parameter stability index.

The parameter weight refers to the proportion of each key operation parameter in the parameter screening result, and is used for reflecting the importance degree of the parameter on the equipment state analysis. Dynamic adjustment refers to adjusting the parameter weight in real time according to the change of the association coefficient and the parameter stability index.

In this step, by generating a parameter screening result including parameter weight assignment, the influence of each key operation parameter on the device state can be more accurately comprehensively considered. The parameter weight is dynamically adjusted according to the association coefficient and the parameter stability index, so that the parameter screening result can be more in line with the actual running condition of the equipment. For example, for a key operating parameter in an industrial control system, the parameter weights are dynamically adjusted according to their relevance coefficients and parameter stability indicators, and parameter screening results containing different weights are generated for subsequent state evaluation.

As an embodiment, the step S350 may specifically include the steps of:

Step S351, calculating the real-time offset according to the real-time measured value of the key operation parameter and the historical average value of the corresponding parameter in the historical operation data set.

The real-time offset is the difference between the real-time measurement of the critical operating parameter and the mean of that parameter in the historical operating dataset. It reflects the degree of deviation of the current parameter value from the historical average level.

In this step, the calculation of the real-time offset helps to find abnormal changes in the critical operating parameters. For example, for a temperature parameter of an air conditioning system having a historical average of 25 ℃, the current real-time measurement is 28 ℃, the real-time offset is 3 ℃. By monitoring the real-time offset in real time, whether abnormal fluctuation occurs in the temperature parameter can be timely found.

And step S352, determining the real-time abnormal probability of each key operation parameter based on the real-time offset and the parameter stability index.

The real-time anomaly probability refers to the magnitude of the probability of anomaly of a certain key operation parameter at the current moment. The method is obtained by comprehensive calculation according to the real-time offset and the parameter stability index.

In actual computation, a probability model may be employed to determine the real-time anomaly probability. For example, when the real-time offset is larger and the parameter stability index is lower, the possibility of abnormality of the parameter is larger, and the real-time abnormality probability is correspondingly improved. For a speed parameter of an elevator, if the real-time offset exceeds a preset range and the stability of the parameter is poor, the real-time abnormal probability of the speed parameter is increased.

And step S353, generating a dynamic weight distribution proportion according to the real-time anomaly probability and the relevance coefficient, wherein the dynamic weight distribution proportion is positively correlated with the product of the real-time anomaly probability and the relevance coefficient.

The dynamic weight distribution proportion refers to the specific weight proportion of each key operation parameter in the parameter screening result. The real-time anomaly probability and the correlation coefficient are positively correlated, namely, the larger the real-time anomaly probability and the correlation coefficient are, the larger the dynamic weight distribution proportion is.

In the step, the weight of each key operation parameter can be distributed more reasonably by generating the dynamic weight distribution proportion, so that the parameter screening result reflects the operation state of the equipment more accurately. For example, for a plurality of key operating parameters in a sewage treatment system, dynamic weight distribution ratios are generated according to their real-time anomaly probabilities and relevance coefficients, so that parameters with higher anomaly probabilities and higher relevance to equipment states have higher weights in the parameter screening results.

Step S354, carrying out weighted fusion on the real-time measured value of the key operation parameter and the dynamic weight distribution proportion to generate a parameter screening result containing the weighted parameter.

Weighted fusion refers to multiplying the real-time measurement of the key operating parameter by the corresponding dynamic weight distribution ratio, and then adding the products to obtain a comprehensive result. The parameter screening results including the weighting parameters can more fully reflect the impact of each key operating parameter on the device status.

In practical application, the effect of the key operation parameters with larger weight can be highlighted through weighted fusion, and meanwhile, the influence of the parameters with smaller weight is reduced. For example, for key operation parameters collected by a plurality of sensors in an intelligent home system, the real-time measurement values and the dynamic weight distribution proportion of the key operation parameters are subjected to weighted fusion, and a parameter screening result containing the weighted parameters is generated for subsequent equipment state evaluation.

Step S355, generating a parameter screening code according to the weighted parameter values in the parameter screening result, wherein the parameter screening code is used for identifying the priority order of different key operation parameters in the second analysis level.

Parameter screening encoding is a way to encode key operating parameters by determining the order of priority of different key operating parameters in a second analysis level based on weighted parameter values in the parameter screening results.

In this step, generating parameter screening codes helps to more efficiently process key operating parameters in the second analysis level. For example, for a plurality of critical operating parameters in an automated warehouse system, parameter screening codes are generated according to their weighted parameter values, so that the parameters can be sequentially processed in a priority order in a second analysis level, thereby improving the efficiency and accuracy of state evaluation.

And step 400, inputting the parameter screening result into a state evaluation strategy in the second analysis level, and generating a state evaluation index of the target equipment by combining a dynamic threshold range corresponding to the key operation parameter.

The state evaluation strategy is a method for evaluating and judging the operation state of the device according to the input parameter screening result. The dynamic threshold range refers to a reasonable value range of key operation parameters in different equipment operation stages and conditions, and the key operation parameters can be adaptively adjusted along with the change of factors such as equipment operation time, environment and the like. The state evaluation index is a set of indexes for comprehensively reflecting the current running state of the target equipment, such as an abnormality level, an abnormality duration, and the like.

In the step, the parameter screening result is input into a state evaluation strategy, and is analyzed in combination with a dynamic threshold range, so that the running state of the target equipment can be evaluated more accurately. For example, for a wind generating set, the parameter screening result is input into a state evaluation strategy, and meanwhile, the state evaluation index of the wind generating set is generated by combining the dynamic threshold range of key operation parameters such as generator temperature, rotation speed and the like, such as whether abnormality exists or not, the severity of the abnormality and the like.

As an embodiment, the step S400 may specifically include the steps of:

Step S410, analyzing the real-time measured value of the key operation parameter and the corresponding parameter weight distribution proportion from the parameter screening result.

The parameter screening result comprises real-time measurement values of key operation parameters and weight distribution proportion corresponding to each parameter. The information is parsed to enable accurate subsequent state assessment using the data.

In the step, the real-time measured value of the key operation parameter and the parameter weight distribution proportion can be extracted by analyzing the parameter screening result. For example, for a parameter screening result of a chemical production device, real-time measured values of key operation parameters such as temperature, pressure and the like and corresponding parameter weight distribution ratios thereof are analyzed, and basic data is provided for subsequent state evaluation.

Step S420, determining a dynamic threshold range of the key operation parameters according to parameter distribution intervals under different equipment state labels in the historical operation data set, and adaptively adjusting the dynamic threshold range along with the operation time of the equipment.

The parameter distribution interval refers to the value range of the key operation parameters under different equipment state labels. The dynamic threshold range is determined according to the parameter distribution interval and is adaptively adjusted along with the change of the running time of the equipment so as to adapt to the dynamic change of the running state of the equipment.

In the step, the reasonable value range of the key operation parameters in different states can be determined by analyzing the parameter distribution intervals under different equipment state labels in the historical operation data set. As the device's run time increases, the device's performance and operating state may change, and thus the dynamic threshold range may need to be adjusted accordingly. For example, for a battery parameter of an electric vehicle, according to parameter distribution intervals in different charging states and driving states in a historical operation data set, a dynamic threshold range of key operation parameters such as battery voltage and current is determined, and self-adaptive adjustment is performed along with the increase of the service time and the charge and discharge times of the battery.

In one embodiment, in step S420, determining the dynamic threshold range of the key operation parameter may specifically include the following steps:

Step S421, extracting the historical maximum and minimum values of the key operation parameters in different equipment operation stages from the historical operation data set.

Historical maximum and minimum values refer to maximum and minimum values reached by critical operating parameters at different plant operating stages in the historical operating dataset. They reflect the range of values of the parameter during historical operation.

In this step, the historical maximum and minimum values of the key operating parameters are extracted to help understand the range of variation of the parameters at different operating stages. For example, for a temperature parameter of a steel smelting furnace, historical maximum and minimum values at different operating stages such as melting stage, refining stage and the like are extracted from a historical operating data set to provide a reference for the subsequent determination of dynamic threshold ranges.

Step S422, calculating an initial threshold range according to the historical maximum value and the historical minimum value, and determining the current operation stage based on the real-time operation time of the target equipment.

The initial threshold range is a preliminary threshold range calculated from the historical maximum and minimum values, which provides a basis for subsequent dynamic corrections. The current operation stage is determined according to the real-time operation time of the target equipment and the operation rule of the equipment, and different operation stages can correspond to different parameter value ranges.

In actual calculation, the initial threshold range may take the interval between the historical maximum and minimum values. For example, for a water level parameter of an industrial boiler, an initial threshold range is calculated from the historical maximum and minimum values as [ water level minimum, water level maximum ]. And then judging different operation stages such as a current start-up stage, a normal operation stage, a shutdown stage and the like according to the real-time operation time of the boiler.

Step S423, obtaining a sliding average value and a sliding variance of the real-time measured value of the key operation parameter in the current operation stage.

A running average refers to the average of real-time measurements of critical operating parameters over a sliding window during the current operating phase. The sliding variance is a statistic that measures the degree of dispersion of these real-time measurements within the sliding window.

In this step, the sliding average and sliding variance are obtained to help understand the real-time variation of the key operating parameters during the current operating phase. For example, for a feed rate parameter of a numerically controlled machine tool, the stability and fluctuation of the parameter can be determined by calculating its running average and running variance over the current run phase.

Step S424, dynamically correcting the initial threshold range according to the sliding average value and the sliding variance, and generating a dynamic threshold range containing the correction offset.

The correction offset amount is an amount that requires adjustment of the upper limit value and the lower limit value of the threshold value range when dynamically correcting the initial threshold value range. By dynamically modifying the initial threshold range according to the sliding average and the sliding variance, the dynamic threshold range can be made to more conform to the current operating state of the device.

In this step, the process of dynamic correction may adjust the upper and lower limit values of the initial threshold range according to the variation of the sliding average and the sliding variance. For example, if the sliding average value changes greatly, the average level of the specification parameter changes, and the threshold range needs to be adjusted accordingly, and if the sliding variance increases, the fluctuation of the specification parameter increases, and the threshold range needs to be adjusted appropriately.

As an embodiment, step S424 may specifically include the following steps:

Step S4241, calculating the difference between the sliding average value and the median value of the initial threshold range as the mean shift amount.

The mean shift refers to the difference between the running average and the median of the initial threshold range, reflecting the degree of deviation of the average level of the current parameter from the initial threshold range.

In this step, the mean shift is calculated to help determine the direction and magnitude of the initial threshold range that needs to be adjusted. For example, for a refrigerant system pressure parameter, the difference between its running average and the median of the initial threshold range is calculated, and if the difference is positive, indicating that the average level of the current pressure is above the median of the initial threshold range, a corresponding up-regulation of the threshold range is required.

Step S4242, determining a variance correction coefficient according to the ratio of the sliding variance to the historical variance.

The variance modification coefficient is a coefficient calculated from the ratio of the sliding variance to the historical variance, and is used to adjust the width of the threshold range to accommodate the variation in the parameter fluctuation.

In this step, by comparing the magnitude of the sliding variance with the magnitude of the history variance, it can be determined whether the fluctuation of the parameter has changed. If the sliding variance is larger than the historical variance, the variance correction coefficient needs to be increased to enlarge the threshold range, otherwise, if the sliding variance is smaller than the historical variance, the variance correction coefficient needs to be decreased to reduce the threshold range.

Step S4243, multiplying the mean shift amount by the variance correction coefficient to generate dynamic correction amount.

The dynamic correction amount is a value calculated from the mean shift amount and the variance correction coefficient, and is used to adjust the upper limit value and the lower limit value of the initial threshold range.

In the step, the mean shift is multiplied by the variance correction coefficient, so that the deviation degree of the average level of the parameter and the change of the fluctuation condition can be comprehensively considered, and a reasonable dynamic correction amount can be obtained. For example, for an oil temperature parameter of a power transformer, the calculated mean shift and variance correction coefficients are multiplied to obtain a dynamic correction amount for adjusting the dynamic threshold range of the oil temperature.

Step S4244, adding the dynamic correction amount to the upper limit value and the lower limit value of the initial threshold range respectively, and generating a corrected dynamic threshold range.

In this step, the dynamic correction amount is added to the upper limit value and the lower limit value of the initial threshold range, respectively, so that the corrected dynamic threshold range can be obtained. For example, for a speed parameter of an elevator, the initial threshold range is [ lower speed limit, upper speed limit ], and the dynamic correction amount is Δv, and the corrected dynamic threshold range is [ lower speed limit+Δv, upper speed limit+Δv ].

And step S4245, performing boundary constraint processing on the modified dynamic threshold range so that the upper limit value of the dynamic threshold range does not exceed the preset percentage of the history maximum value and the lower limit value is not lower than the preset percentage of the history minimum value.

The boundary constraint process is to ensure that the dynamic threshold range does not exceed a reasonable range. The preset percentage is a preset proportional value for limiting the upper limit value and the lower limit value of the dynamic threshold range.

In this step, the rationality and reliability of the dynamic threshold range can be ensured by the boundary constraint processing. For example, for a temperature parameter of an automated production line, a preset percentage of a history maximum value is set to 110%, a preset percentage of a history minimum value is set to 90%, and a boundary constraint process is performed on the modified dynamic threshold range so that the upper limit value does not exceed 110% of the history maximum value and the lower limit value is not lower than 90% of the history minimum value.

Step S425, matching and verifying the dynamic threshold range and the historical change rate of the key operation parameters so that the change rate of the dynamic threshold range does not exceed the preset rate limit.

The historical rate of change refers to the rate of change of the key operating parameter during historical operation. The preset rate limit is a preset threshold value for limiting the rate of change of the dynamic threshold range.

In the step, the dynamic threshold range and the historical change rate of the key operation parameters are matched and verified, so that the dynamic threshold range can be prevented from being changed too severely, and the stability and reliability of the dynamic threshold range are ensured. For example, for a pressure parameter of a chemical reactor, the rate of change of the dynamic threshold range is compared with the historical rate of change of the pressure parameter to ensure that the rate of change of the dynamic threshold range does not exceed a preset rate limit.

And S430, calculating the deviation degree of the real-time measured value and the upper limit value and the lower limit value of the dynamic threshold range, and generating a comprehensive deviation score by combining the parameter weight distribution proportion.

The degree of deviation refers to the absolute value of the difference between the real-time measurement of the critical operating parameter and the upper or lower limit of the dynamic threshold range. The comprehensive deviation score is a score value obtained by comprehensive calculation according to the deviation degree and the parameter weight distribution proportion and is used for reflecting the abnormality degree of the key operation parameters.

In this step, by calculating the integrated deviation score, the anomalies of the critical operating parameters can be more fully evaluated. For example, for a generator voltage parameter, the degree of deviation of its actual measured value from the upper and lower values of the dynamic threshold range is calculated, and then the combined deviation score is generated in combination with the weight distribution ratio of the parameter. If the integrated deviation score is higher, it is indicated that the degree of abnormality of the voltage parameter is greater.

Step S440, determining the state abnormality level corresponding to the key operation parameter according to the comparison result of the comprehensive deviation score and the preset deviation threshold.

The preset deviation threshold is a preset critical value for judging whether the abnormality degree of the key operation parameter reaches a set level. The status abnormality levels are different levels such as mild abnormality, moderate abnormality, severe abnormality, and the like, which are classified according to the comparison result of the comprehensive deviation score and the preset deviation threshold.

In the step, the state abnormality level corresponding to the key operation parameter can be accurately determined by comparing the comprehensive deviation score with a preset deviation threshold. For example, for a flow parameter of a water pump, if the integrated deviation score is greater than a preset deviation threshold and the deviation degree is greater, determining that the state abnormality level corresponding to the flow parameter is heavy abnormality.

And S450, integrating the state anomaly grades of all the key operation parameters to generate an overall state evaluation index of the target equipment, wherein the overall state evaluation index comprises anomaly grade distribution and anomaly duration.

The abnormal level distribution refers to the distribution of state abnormal levels corresponding to each key operation parameter, and reflects the abnormal degree of the target equipment in different aspects. The abnormal duration refers to a duration in which the target device is in an abnormal state.

In this step, the overall state evaluation index of the target device can be obtained by integrating the state anomaly levels of all the key operation parameters. For example, for a production line of an automated factory, the status anomaly levels of each key operating parameter (such as temperature, pressure, speed, etc.) are integrated to generate an overall status assessment index for the production line, including anomaly level distribution (such as which of the mild anomaly parameters, which of the moderate anomaly parameters, etc.) and anomaly duration, so as to fully understand the operating status of the production line.

And S500, outputting the real-time running state grade of the target equipment according to the comparison result of the state evaluation index and the preset state threshold value.

The preset state threshold is a preset set of critical values for classifying the state evaluation index into different levels. The real-time running state grade is determined according to the comparison result of the state evaluation index and the preset state threshold value, and reflects the current running state of the target equipment, such as normal, warning, fault and the like.

In this step, the real-time operation state level of the target device can be accurately output by comparing the state evaluation index with the preset state threshold. For example, for a certain wind driven generator of a wind power plant, comparing its state evaluation index (such as abnormal level distribution, abnormal duration, etc.) with a preset state threshold, if the state evaluation index is within a normal range, outputting a real-time operation state level as normal, if the warning threshold is exceeded but the fault threshold is not reached, outputting a real-time operation state level as warning, and if the fault threshold is exceeded, outputting a real-time operation state level as fault.

As an implementation manner, the method provided by the embodiment of the invention may further include:

Step S210, adding a third analysis level in the multi-level state analysis model, wherein the third analysis level is behind the second analysis level.

The third analysis level is an analysis level added in the multi-level state analysis model and is used for further analyzing and predicting the state of the device. After the third analysis level is added to the second analysis level, a further analysis may be performed using the state evaluation index output by the second analysis level.

In this step, adding a third analysis level may enhance the functionality of the multi-level state analysis model, enabling it to provide more comprehensive device state information. For example, for a server device of a large data center, a third analysis level is added on the basis of the original first analysis level and the second analysis level, so that future states of the server can be predicted, and potential fault hidden dangers can be found in advance.

And S220, predicting a future state evolution path of the target device according to the historical change trend of the state evaluation index through a state prediction strategy in the third analysis level.

The state prediction policy is a method for predicting a future state of the target device according to a historical variation trend of the state evaluation index. The future state evolution path refers to a state change process that the target device may undergo over a period of time in the future.

In the step, the future state of the target equipment can be predicted by using the historical data through a state prediction strategy, and a reference is provided for maintenance and management of the equipment. For example, for a bridge structural health monitoring system, according to a state prediction strategy in a third analysis level, according to the historical change trend of bridge state evaluation indexes, predicting a structural safety state evolution path of the bridge in a period of time in the future, and taking corresponding maintenance measures in advance.

As an embodiment, the step S220 may specifically include the steps of:

Step S221, extracting time series data from the state evaluation index, wherein the time series data comprise the state evaluation index values of the target equipment at different time points and the real-time measured values of the corresponding key operation parameters.

The time series data is a set of data arranged in time sequence, which reflects the state evaluation index value of the target device and the time-dependent change of the real-time measurement value of the key operation parameter.

In this step, time-series data is extracted so that the state prediction can be performed using these data later. For example, for a state evaluation index of an aeroengine, the state evaluation index values of different time points in the past period and the corresponding real-time measured values of key operation parameters (such as temperature, pressure and the like) are extracted to form time series data.

Step S222, performing multi-scale decomposition on the time series data through a state prediction strategy in a third analysis level to generate decomposed time series data comprising a long-term trend component, a periodic fluctuation component and a short-term noise component.

Multiscale decomposition is a method of decomposing time series data into different scale components. The long-term trend component reflects the long-term variation trend of the time series data, the periodic fluctuation component reflects the periodic variation characteristic of the data, and the short-term noise component is a random fluctuation part in the data.

In the step, different characteristics in the time series data can be separated through multi-scale decomposition, so that the subsequent analysis and processing of different components are facilitated. For example, for a load prediction problem of an electric power system, the time series data of the load is subjected to multi-scale decomposition to obtain a long-term trend component, a periodic fluctuation component and a short-term noise component, and these components are respectively analyzed and predicted, so that the accuracy of the prediction can be improved.

Step S223, the overall degradation direction of the target equipment is identified based on the long-term trend component, and the operation state change period of the target equipment is extracted by combining the periodic fluctuation component.

The overall degradation direction refers to a direction in which the target device gradually decreases during long-term operation. The operation state change period refers to a time interval at which the operation state of the target device is periodically changed according to a rule.

In the step, the overall degradation condition of the target equipment can be known by analyzing the long-term trend component, and the change period of the running state of the equipment can be extracted by combining the periodic fluctuation component. For example, vibration monitoring data of a mechanical device is analyzed to find out that the vibration amplitude of the device is gradually increased by analyzing a long-term trend component, so that the device has a trend of overall degradation, and the periodic fluctuation component is combined to extract the change period of the vibration of the device, so as to know the periodic change rule of the running state of the device.

Step S224, a predictor model set is generated according to the overall degradation direction and the running state change period, wherein the predictor model set comprises a plurality of predictor models matched with different degradation stages and period fluctuation characteristics.

A set of predictive sub-models is a set of sub-models for predicting the future state of a target device, each designed for a different degradation phase and periodic wave characteristics.

In the step, a predictor model set is generated according to the overall degradation direction and the running state change period, so that the accuracy and pertinence of state prediction can be improved. For example, for a battery life prediction problem of an electric vehicle, a plurality of prediction sub-models including different degradation phases and cycle fluctuation feature matching, such as an early degradation phase sub-model, a middle degradation phase sub-model, and the like, are generated according to the overall degradation direction and charge-discharge cycle fluctuation feature of the battery.

Step S225, inputting the decomposed time series data into each predictor model in the predictor model set, generating future state prediction results under different confidence degrees, and performing cross-validation on the future state prediction results.

Cross-validation is a method for evaluating the accuracy and reliability of predictive models by dividing the data set into multiple subsets, and using different subsets in turn for training and validation.

In the step, the decomposed time series data is input into each predictor model, future state prediction results under different confidence degrees are generated, and cross verification is carried out, so that the most accurate and reliable prediction results can be screened out. For example, for quality prediction problems in an industrial process, decomposed time series data is input to each predictor model in a set of predictor models to generate product quality prediction results with different confidence levels, and then the accuracy of the results is evaluated through cross-validation to select the optimal prediction result.

And step S226, screening out a target predictor model with the minimum deviation from the latest state evaluation index value in the real-time operation data stream of the target equipment according to the cross-validation result.

The target prediction sub-model is a sub-model which is screened from a prediction sub-model set and is most suitable for the state prediction of the current target equipment, and the deviation between the target prediction sub-model and the latest state evaluation index value in the real-time operation data flow of the target equipment is the smallest.

In the step, the accuracy of state prediction can be improved by screening the target predictor model. For example, for a performance prediction problem of a communication base station, a target prediction sub-model with the smallest deviation from the latest state evaluation index value in the base station real-time operation data stream is screened out according to the cross-validation result, and is used for subsequent performance prediction of the base station.

And step S227, carrying out trend correction on short-term noise components in the time sequence data through the target predictor model, and generating a future state evolution path containing the corrected noise influence.

The trend correction is to process the short-term noise component and remove random fluctuation in the short-term noise component so as to enable the short-term noise component to conform to the overall change trend. The future state evolution path containing the modified noise effect is a path of change of the future state of the target device obtained after taking the modified effect of the short-term noise component into account.

In the step, the future state evolution path can be smoother and more accurate by carrying out trend correction on the short-term noise component. For example, for a weather forecast problem, the short-term noise component in the weather data is subjected to trend correction through the target prediction sub-model, a future weather change path containing the corrected noise influence is generated, and the accuracy of the weather forecast is improved.

And step S228, performing association matching on the future state evolution path and a dynamic threshold range of key operation parameters in the real-time operation data stream, and identifying an abnormal prediction interval exceeding the dynamic threshold range in the future state evolution path.

Correlation matching refers to comparing a future state evolution path with a dynamic threshold range of a key operation parameter, and finding out a part exceeding the dynamic threshold range in the future state evolution path. The abnormal prediction interval refers to a time period exceeding a dynamic threshold range in the future state evolution path.

In the step, the time period of possible abnormality of the target equipment can be found in advance through association matching, so that a basis is provided for preventive maintenance. For example, for a chemical production device, the future state evolution path is associated and matched with the dynamic threshold range of key operation parameters such as temperature, pressure and the like, an abnormal prediction interval in which the temperature or pressure in the future state evolution path exceeds the dynamic threshold range is identified, and measures are taken in advance to avoid accidents.

Step S229, based on the starting time point and the ending time point of the abnormal prediction interval and the corresponding key operation parameter types, generating preventive maintenance suggestions comprising maintenance trigger conditions and maintenance parameter ranges.

The maintenance trigger condition refers to a condition under which equipment maintenance is required, and is generally related to the starting time point of the abnormality prediction interval and the abnormal condition of the key operation parameters. The maintenance parameter range refers to a reasonable value range of parameters which need to be adjusted when equipment maintenance is performed.

In the step, preventive maintenance suggestions are generated according to the starting time point and the ending time point of the abnormal prediction interval and the corresponding key operation parameter types, so that maintenance work of equipment can be planned in advance, and the reliability and the service life of the equipment are improved. For example, for a power transformer, based on the starting time point of the abnormality prediction interval and the corresponding abnormality of the temperature parameter, preventive maintenance advice including a maintenance trigger condition (e.g., maintenance performed when the temperature exceeds a certain threshold) and a maintenance parameter range (e.g., a temperature range for replacing cooling oil) is generated.

Step S230, extracting key turning points in the future state evolution path, and generating preventive maintenance suggestions by combining real-time measured values of key operation parameters.

The critical turning point is a point in the future state evolution path where the state change is more severe, and it usually indicates an important change in the state of the device. The preventive maintenance advice is generated by combining the real-time measured value of the key operation parameter, so that the maintenance advice is more in line with the actual operation condition of the equipment.

In this step, the critical turning points are extracted and preventive maintenance suggestions are generated in combination with real-time measurement values, so that maintenance work of the equipment can be more accurately arranged, and unnecessary maintenance and faults are avoided. For example, for an aeroengine, the critical turning points in the future state evolution path are extracted, and in combination with real-time measurement values of the critical operation parameters such as the temperature, the pressure and the like of the engine, targeted preventive maintenance suggestions are generated, such as overhauling and debugging of the engine before the critical turning points.

As an embodiment, the step S230 may specifically include the following steps:

Step S231, a section of which the trend change rate of the predicted state curve exceeds a preset rate threshold is identified from the future state evolution path, and a starting time point and a terminating time point of the section are extracted as initial turning points.

The predicted state curve is a graphical representation of the future state evolution path reflecting the change in future state of the target device over time. The trend change rate refers to the slope of the predicted state curve at a point, which indicates how fast or slow the state changes. The preset rate threshold is a preset critical value, and is used for judging whether the trend change rate is too high. The initial turning point refers to a starting time point and an ending time point of a section of the predicted state curve, wherein the trend change rate of the section exceeds a preset rate threshold.

In this step, by identifying the interval section where the trend change rate of the predicted state curve exceeds the preset rate threshold, a portion with more intense state change in the future state evolution path can be found, and the starting time point and the ending time point of the portion can be extracted as initial turning points. For example, for an elevator operation state prediction curve, a section of the curve in which the trend change rate exceeds a preset rate threshold is identified, and the starting time point and the ending time point of the section are extracted as initial turning points, which may be predictive of important changes in the elevator operation state.

Step S232, calculating trend association degree between adjacent initial turning points based on fluctuation amplitude and fluctuation direction of the prediction state curve between the initial turning points, and screening out the initial turning points with the trend association degree smaller than a preset association threshold as candidate turning points.

The trend correlation is an index for measuring the similarity of the predicted state curves between adjacent initial turning points, and is calculated according to the fluctuation amplitude and the fluctuation direction. The preset association threshold is a preset critical value, and is used for judging whether the trend association degree is small enough. The candidate turning points are initial turning points with trend correlation less than a preset correlation threshold, and state changes between the turning points may be more independent and important.

In this step, the initial turning points with the trend correlation degree smaller than the preset correlation threshold are selected as candidate turning points by calculating the trend correlation degree between the adjacent initial turning points, so that the range of the key turning points can be further narrowed. For example, for a motion state prediction curve of an industrial robot, calculating a trend correlation degree between adjacent initial turning points, and screening out initial turning points with the trend correlation degree smaller than a preset correlation threshold as candidate turning points, wherein the candidate turning points may correspond to important changes of the motion state of the robot.

Step S233, according to the distribution density of the candidate turning points in the future state evolution path and the time interval of the adjacent candidate turning points, the second derivative change characteristics of the predicted state curve are fused, and the candidate turning points meeting the density condition and the time interval condition are determined as key turning points.

The distribution density refers to the distribution density of the candidate turning points in the future state evolution path. The time interval refers to the time distance between adjacent candidate turning points. The second derivative change characteristic reflects the curvature change condition of the predicted state curve. The density condition and the time interval condition are preset standards for screening out real key turning points.

In the step, the distribution density, the time interval and the second derivative change characteristic of the predicted state curve of the candidate turning points are comprehensively considered, and the candidate turning points meeting the conditions are determined as key turning points, so that the most important state change points in the future state evolution path can be more accurately found. For example, for a power prediction curve of a wind turbine generator, according to the distribution density of candidate turning points and the time interval between adjacent candidate turning points, in combination with the second derivative variation characteristics of the curve, candidate turning points meeting the density condition and the time interval condition are determined as key turning points, and these key turning points may correspond to the fault occurrence points or the performance variation points of the wind turbine generator.

Step S234, obtaining real-time measurement values of the key operation parameters in the target time window corresponding to the key turning points, and extracting abnormal times and abnormal duration time of the real-time measurement values exceeding the dynamic threshold range in the target time window.

The target time window refers to a time period centered around the critical turning point for obtaining real-time measurements of the critical operating parameters. The abnormal times refer to the number of times that the real-time measurement of the critical operating parameter exceeds the dynamic threshold range within the target time window. The anomaly duration refers to the total duration that the critical operating parameter is in an anomaly state.

In the step, the real-time measured value of the key operation parameter in the target time window corresponding to the key turning point is obtained, the abnormal times and the abnormal duration are extracted, and the abnormal condition of the key operation parameter near the key turning point can be known. For example, for a temperature parameter of a chemical reaction kettle, acquiring a real-time temperature measurement value in a target time window corresponding to a key turning point, and extracting abnormal times and abnormal duration of the temperature exceeding a dynamic threshold range in the time window, wherein the information can provide basis for judging the running state of the reaction kettle and taking corresponding maintenance measures.

And S235, determining the abnormal state grade of the key operation parameter at the key turning point according to the abnormal times and the abnormal duration, and generating a maintenance type identifier by combining the trend change rate of the key turning point.

The status anomaly level is a different level classified according to the anomaly number and anomaly duration, such as mild anomaly, moderate anomaly, severe anomaly, etc. The maintenance type identifier is a code for identifying a type of maintenance to be performed, which is generated based on the state anomaly level of the key operation parameter and the trend change rate of the key turning point.

In the step, the state abnormality grade is determined according to the abnormality times and the abnormality duration, and the maintenance type identifier is generated by combining the trend change rate, so that clear guidance can be provided for subsequent maintenance work. For example, for an oil temperature parameter of a power transformer, the state anomaly level is determined to be heavy anomaly according to the anomaly times and anomaly duration of the oil temperature near the critical turning point, and a maintenance type identifier, such as "emergency maintenance", is generated in combination with the trend change rate of the critical turning point, which indicates that the transformer needs to be immediately overhauled.

Step S236, the maintenance action sequences in the preset maintenance strategy library are matched based on the maintenance type identifier, and the maintenance action sequences are prioritized according to the state abnormality level.

The preset maintenance strategy library is a database containing maintenance action sequences corresponding to various maintenance types. A maintenance action sequence refers to a series of maintenance actions that need to be performed in order to accomplish a certain maintenance type. The priority ranking is to rank the maintenance action sequences according to the state anomaly level, so that the maintenance action sequences with higher state anomaly levels have higher priorities.

In the step, the maintenance action sequences in the preset maintenance strategy library are matched based on the maintenance type identifier, and the priority ranking is carried out according to the state abnormality level, so that proper maintenance measures can be timely adopted when the equipment is abnormal. For example, for equipment failure of an automated production line, corresponding maintenance action sequences, such as replacement of parts, adjustment parameters, etc., are matched from a maintenance policy library according to the maintenance type identifier, and then the maintenance action sequences are prioritized according to the state anomaly level, and the maintenance action sequences with the state anomaly level are preferentially executed.

Step S237, generating preventive maintenance suggestions comprising maintenance types, maintenance time windows and maintenance action execution sequences according to the time distribution characteristics of the key turning points and the priority ordering of the maintenance action sequences.

The time distribution feature refers to the distribution of key turning points on a time axis, such as interval time, occurrence frequency, and the like. The maintenance time window refers to an appropriate period of time for performing maintenance work. The maintenance action execution sequence refers to the execution sequence of each maintenance action in the maintenance action sequence.

In the step, preventive maintenance suggestions are generated according to the time distribution characteristics of key turning points and the priority ordering of maintenance action sequences, so that maintenance work of equipment can be reasonably arranged, and the maintenance efficiency and the reliability of the equipment are improved. For example, for a server device in a large data center, preventive maintenance advice is generated that includes a maintenance type (e.g., periodic inspection, trouble shooting, etc.), a maintenance time window (e.g., weekend idle time), and a maintenance action execution order (e.g., checking hardware first, updating software second, etc.) based on the time distribution characteristics of the critical turning points and the prioritization of the maintenance action sequences.

As an implementation manner, after step S500, the method provided by the embodiment of the present invention may further include:

And S600, setting a feedback optimization mechanism in the multi-level state analysis model, and carrying out iterative updating on the operation parameter screening strategy and the state evaluation strategy by the feedback optimization mechanism according to the output result of the real-time operation state level.

The feedback optimization mechanism is a mechanism for optimizing the multi-level state analysis model, and adjusts the operation parameter screening strategy and the state evaluation strategy according to the output result of the real-time operation state level so as to improve the accuracy and the adaptability of the model.

In the step, a feedback optimization mechanism is arranged, so that a multi-level state analysis model can be continuously learned and improved, and the method is suitable for the change of the running state of equipment. For example, for a device state analysis model of a smart grid, an operation parameter screening strategy is adjusted according to the output result of a real-time operation state level through a feedback optimization mechanism, a more appropriate key operation parameter is selected, and a state evaluation strategy is updated, so that the accuracy of state evaluation is improved.

And step S700, obtaining a matching error between the real-time running state grade and the actual equipment maintenance record, and calculating model optimization weight according to the matching error.

The matching error refers to the degree of difference between the real-time running state level and the actual equipment maintenance record. The model optimization weight is a weight value obtained by calculation according to the matching error and is used for adjusting the updating degree of the operation parameter screening strategy and the state evaluation strategy.

In the step, the matching error is obtained, the model optimization weight is calculated, the inaccuracy degree of the model can be quantized, and a basis is provided for model optimization. For example, for a state analysis model of an industrial plant, by comparing the real-time operating state level with the actual plant maintenance records, calculating the matching error, then calculating the model optimization weight according to the matching error, and when the matching error is large, increasing the model optimization weight, thereby enhancing the updating of the operating parameter screening strategy and the state evaluation strategy.

Step S800, adjusting parameter relevance screening conditions in the first analysis level and dynamic threshold range generation rules in the second analysis level based on the model optimization weights.

The parameter relevance screening conditions are conditions in the first analysis level for screening the key operating parameters, and the dynamic threshold range generation rules are rules in the second analysis level for determining the dynamic threshold range of the key operating parameters. These conditions and rules are adjusted based on model optimization weights, which can enable the multi-level state analysis model to more accurately screen key operating parameters and determine dynamic threshold ranges.

In the step, parameter relevance screening conditions and dynamic threshold range generation rules are adjusted according to model optimization weights, so that the performance of the model can be gradually improved. For example, for a state analysis model of an elevator installation, the parameter correlation screening conditions in the first analysis level are adjusted according to the model optimization weights, the operation parameters with more correlation are selected as key operation parameters, and the dynamic threshold range generation rules in the second analysis level are adjusted so that the dynamic threshold range is more consistent with the actual operation condition of the elevator.

And step 900, applying the adjusted parameter correlation screening conditions and dynamic threshold range generation rules to the newly received real-time operation data stream to generate an updated multi-level state analysis model.

In this step, the adjusted parameter correlation screening conditions and dynamic threshold range generation rules are applied to the newly received real-time operational data stream, and an updated multi-level state analysis model is generated through the processing and analysis of the new data. For example, for a device state analysis model of a sewage treatment plant, the adjusted parameter correlation screening conditions and dynamic threshold range generation rules are applied to the newly received real-time operation data stream, and the parameters such as sewage flow, water quality and the like are subjected to rescreening and threshold determination to generate an updated multi-level state analysis model.

And step S1000, performing backtracking verification on the historical operation data set through the updated multi-level state analysis model so that the prediction accuracy of the updated multi-level state analysis model is not lower than a preset accuracy threshold.

The backtracking verification refers to that the updated multi-level state analysis model is used for analyzing and predicting the historical operation data set, and the prediction result is compared with the actual equipment state label so as to verify the accuracy of the model. The preset accuracy threshold is a preset critical value, and is used for measuring whether the prediction accuracy of the model meets the requirement.

In the step, the performance of the updated multi-level state analysis model can be improved through backtracking verification. For example, for a state analysis model of an aerospace device, backtracking verification is performed on the historical operation data set through the updated model, whether the prediction accuracy of the model is not lower than a preset accuracy threshold is checked, and if the prediction accuracy is not lower than the preset accuracy threshold, parameters and rules of the model are continuously adjusted until the requirements are met.

As an implementation manner, the method provided by the embodiment of the invention may further include:

Step S1100, dynamically adjusting the boundary value range of the preset state threshold based on the historical distribution characteristics of the real-time running state level.

The historical distribution characteristics of the real-time running state grades refer to the characteristics of the occurrence frequency, the distribution interval and the like of different real-time running state grades in a past period of time. The dynamic adjustment of the boundary value range of the preset state threshold refers to adjusting the upper limit value and the lower limit value of the preset state threshold according to the historical distribution characteristics so as to enable the preset state threshold to better conform to the actual running condition of the equipment.

In the step, the boundary value range of the preset state threshold is dynamically adjusted based on the historical distribution characteristics of the real-time running state level, so that the accuracy of state evaluation can be improved. For example, for the state analysis of a solar photovoltaic power generation system, according to the historical distribution characteristics of the real-time operation state levels, the frequency of occurrence of a certain state level is found to be too high or too low, which means that the boundary value range of the preset state threshold value may not be reasonable, and the boundary value range of the preset state threshold value is dynamically adjusted at this time, so that the state evaluation more accurately reflects the actual operation state of the device.

Step 1200, obtaining a plurality of real-time running state grades output by the target device in a preset time period, and counting state evaluation index value distribution intervals corresponding to the grades.

The preset time period is a preset time range and is used for acquiring the real-time running state grade of the target equipment. The state evaluation index value distribution interval refers to the value range of the state evaluation index under each real-time running state level.

In the step, a plurality of real-time running state grades of the target equipment in a preset time period are obtained, and state evaluation index value distribution intervals corresponding to the grades are counted, so that the distribution rule of the state evaluation indexes under different state grades can be known. For example, for the equipment state analysis of an intelligent building, a plurality of real-time operation state grades output by the building equipment in one month are obtained, the distribution interval of state evaluation index values such as temperature, humidity and the like corresponding to each grade is counted, and data support is provided for subsequent adjustment of a preset state threshold value.

Step S1300, matching the state evaluation index value distribution interval with the occurrence frequency of the parameter abnormal event under the same equipment state label in the historical operation data set, and determining a sensitive threshold interval needing to be preferentially adjusted in the preset state threshold.

The occurrence frequency of the parameter abnormal event refers to the number of times of parameter abnormality under the same equipment state label in the historical operation data set. The sensitive threshold interval is an interval with higher correlation with occurrence frequency of parameter abnormal events in the preset state threshold, and the adjustment of the interval may have a larger influence on the accuracy of state evaluation.

In the step, the state evaluation index value distribution interval is matched with the occurrence frequency of the parameter abnormal event, the sensitive threshold interval is determined, and the preset state threshold can be adjusted in a targeted manner. For example, for the state analysis of an industrial robot, the state evaluation index value distribution interval is matched with the occurrence frequency of the parameter abnormal event under the same equipment state label in the historical operation data set, and the occurrence frequency of the parameter abnormal event in a certain interval of a certain state evaluation index is found to be higher, wherein the interval is a sensitive threshold interval, and the preset state threshold of the interval needs to be adjusted preferentially.

And step 1400, generating threshold offset correction coefficients for different real-time running state levels according to the fluctuation characteristics of the state evaluation index values in the sensitive threshold interval.

The fluctuation feature refers to the features such as the change amplitude, the change frequency and the like of the state evaluation index value in the sensitive threshold value interval. The threshold offset correction coefficient is a coefficient for adjusting a preset state threshold boundary value, and is generated according to the fluctuation characteristic of a state evaluation index value in a sensitive threshold interval, and different real-time running state grades can correspond to different threshold offset correction coefficients.

In this step, the threshold offset correction coefficient is generated according to the fluctuation characteristic of the state evaluation index value in the sensitive threshold interval, so that the preset state threshold can be adjusted more accurately. For example, for equipment state analysis of an electric power system, threshold offset correction coefficients for different real-time running state levels are generated according to fluctuation characteristics of state evaluation index values such as voltage, current and the like in a sensitive threshold interval, and when the fluctuation of the state evaluation index values is large, the threshold offset correction coefficients are increased so as to enlarge the boundary value range of a preset state threshold.

And S1500, performing superposition operation on the threshold offset correction coefficient and the current boundary value of the preset state threshold to generate the dynamically adjusted preset state threshold.

In the step, the superposition operation is carried out on the threshold offset correction coefficient and the current boundary value of the preset state threshold value, so as to obtain the preset state threshold value after dynamic adjustment. For example, for the device state analysis of a communication base station, the current boundary value of the preset state threshold is [ threshold lower limit, threshold upper limit ], the threshold offset correction coefficient is delta, and the preset state threshold after dynamic adjustment is [ threshold lower limit+delta, threshold upper limit+delta ].

And step 1600, reclassifying the real-time running state level input subsequently according to the dynamically adjusted preset state threshold value, and outputting the updated real-time running state level and a corresponding threshold value adjustment log.

Reclassification refers to reclassifying and determining the real-time running state level of the subsequent input according to the dynamically adjusted preset state threshold. The threshold adjustment log is a log file for recording the preset state threshold adjustment process and result, and contains information such as adjustment time, adjustment parameters, adjustment amplitude and the like.

In the step, the real-time running state grade is reclassified according to the preset state threshold after dynamic adjustment, so that the state evaluation can more accurately reflect the actual running state of the equipment. Meanwhile, the threshold adjustment log is output, so that the subsequent review and analysis of the threshold adjustment process can be facilitated. For example, for the equipment state analysis of an automatic production line, the real-time running state level input subsequently is reclassified according to the preset state threshold value after dynamic adjustment, and the updated real-time running state level and the corresponding threshold value adjustment log are output so as to discover the change of the equipment state and evaluate the effect of threshold value adjustment in time.

Step 1700, performing association verification on the threshold adjustment log and the operation maintenance record of the target device, and screening out abnormal threshold intervals which are not matched with the actual device fault event in the preset state threshold after dynamic adjustment.

The correlation verification refers to comparing the threshold adjustment log with the operation maintenance record of the target equipment, and checking whether the dynamically adjusted preset state threshold is matched with the actual equipment fault event. The abnormal threshold interval refers to an interval which is not matched with an actual equipment fault event in a preset state threshold after dynamic adjustment, and the interval may cause inaccuracy of state evaluation.

In this step, the abnormal threshold interval is screened out through association verification, so that the preset state threshold can be further optimized. For example, for the equipment state analysis of a wind power plant, performing association verification on a threshold adjustment log and an operation maintenance record of the wind power generator, and finding that a preset state threshold after dynamic adjustment of a certain state evaluation index is not matched with an actual equipment fault event in a certain section, wherein the section is an abnormal threshold section, and the threshold of the section needs to be adjusted again.

Step S1800, based on the adjusting time node of the abnormal threshold interval and the corresponding state evaluation index value, calculating logic of the reverse correction threshold offset correction coefficient generates a preset state threshold after secondary correction.

The calculation logic of the reverse correction threshold offset correction coefficient refers to adjusting and improving the method for generating the threshold offset correction coefficient according to the adjusting time node of the abnormal threshold interval and the corresponding state evaluation index value. The preset state threshold after the secondary correction is a more accurate preset state threshold obtained after calculation logic of the reverse correction threshold offset correction coefficient.

In this step, based on the adjustment time node of the abnormal threshold interval and the corresponding state evaluation index value, the calculation logic of the threshold offset correction coefficient is reversely corrected to generate the preset state threshold after the secondary correction, so that the accuracy and reliability of the preset state threshold can be improved. For example, for the state analysis of a chemical production device, according to the adjustment time node of the abnormal threshold interval and the corresponding state evaluation index values of temperature, pressure and the like, the calculation logic of the correction coefficient of the reverse correction threshold offset is used for generating a preset state threshold after the secondary correction, so that the state evaluation is more in accordance with the actual running condition of the device.

And S1900, performing secondary reclassification on the real-time running state grade through the secondarily corrected preset state threshold value, and outputting the finally calibrated real-time running state grade and the calibrated parameter set.

The secondary reclassification refers to reclassifying and determining the real-time running state level according to the preset state threshold after the secondary correction. The calibration parameter set is a set containing the preset state threshold after the secondary correction and other relevant calibration parameters, and records the final calibration result.

In the step, the final calibrated real-time running state grade and the calibration parameter set are output through secondary reclassification, so that the accuracy of state evaluation can be ensured to reach a higher level. For example, for a server device of a large data center, the real-time running state level is reclassified by using the preset state threshold after the secondary correction, the final calibrated real-time running state level such as normal, early warning, failure, etc. is output, and the preset state threshold after the secondary correction and related calibration parameters (such as adjustment coefficient, boundary value, etc.) are sorted into a calibration parameter set. The method not only can provide accurate equipment state information for operation and maintenance personnel of the data center, but also can provide reliable basis for subsequent equipment management and maintenance decision.

As an implementation manner, the method provided by the embodiment of the invention may further include:

and step S2000, generating a device maintenance trigger instruction sequence according to the finally calibrated real-time running state level.

A device maintenance trigger instruction sequence is a series of instruction sets for triggering a device maintenance operation. Different final calibrated real-time operating state levels correspond to different maintenance requirements, so that corresponding maintenance trigger instructions can be generated according to the levels. For example, when the real-time running state is rated as "normal", the instruction may be generated as regular inspection, when the real-time running state is rated as "early warning", the instruction may be to perform preliminary inspection and parameter adjustment of the equipment, and when the real-time running state is rated as "fault", the instruction may be to immediately stop for maintenance, etc. In practical application, for a mechanical device on an automation production line, according to the final calibrated real-time operation state level, the system generates an instruction sequence including specific maintenance operation and execution sequence, for example, power-off operation of the device is performed first, and then fault detection is performed.

In step S2100, the extracting device maintains a real-time running state class duration corresponding to each instruction in the trigger instruction sequence and a time interval between adjacent instructions.

The real-time running state grade duration refers to the duration of a certain real-time running state grade of the equipment, and the time interval between adjacent instructions refers to the time difference between the execution of two adjacent instructions in the equipment maintenance trigger instruction sequence. Extracting this information helps to properly schedule maintenance work and sequence. For example, in an electric power system, for a maintenance trigger instruction sequence of a generator, the duration of a real-time running state class corresponding to each instruction is extracted, so that the running stability of the generator in different states can be judged, and the time interval between adjacent instructions is extracted, so that a maintenance plan can be optimized, and the excessive concentration or overlong interval of maintenance work can be avoided.

The extraction of the real-time operational status class duration may be performed by analyzing historical data of the device status monitoring system. The device status monitoring system records the real-time operation status levels of the device at different time points, and the duration of each status level can be obtained by sorting and counting the records. For example, for an intelligent building elevator installation, the status monitoring system would record in real time the operating status (normal, malfunctioning, etc.) of the elevator, and by analyzing these records, the duration of time that the elevator is in a malfunctioning state can be determined in order to evaluate the severity and scope of the malfunction.

Step S2200, generating a preliminary maintenance strategy scheme based on the fact that the duration time interval is matched with the maintenance action execution sequence in the preset maintenance strategy library.

The preset maintenance policy library is a database containing various equipment maintenance policies and action execution sequences, and is classified and stored according to the types, states and maintenance requirements of the equipment. And searching a maintenance action execution sequence matched with the extracted real-time running state grade duration and the time interval between adjacent instructions in a preset maintenance strategy library, so as to generate a preliminary maintenance strategy scheme. For example, for a wind generating set, according to the duration of the real-time running state level and the time interval between adjacent instructions, a corresponding maintenance strategy is found in a preset maintenance strategy library, for example, when the equipment is in an early-warning state and the duration is long, a maintenance action sequence of performing equipment inspection and then component debugging is executed, and a preliminary maintenance strategy scheme is generated.

Step S2300, performing conflict detection on an execution time window of each maintenance action in the preliminary maintenance strategy scheme and a production schedule of the target equipment, and screening out maintenance actions to be adjusted with time conflicts.

The execution time window refers to a time range in which each maintenance action can be executed, and the production schedule of the target device specifies production tasks and operation schedules of the device in different time periods. The conflict detection can avoid the conflict between maintenance work and production tasks, and ensure the normal operation of the equipment and the smooth execution of the production plan. For example, for a line device of an automobile manufacturing plant, the execution time window of a certain maintenance action in the preliminary maintenance strategy scheme overlaps with the production peak period of the production line, and this time conflict can be found through conflict detection, and the maintenance action is marked as a maintenance action to be adjusted.

Step 2400, reallocating the execution priority of the maintenance action according to the conflict type of the maintenance action to be regulated and the emergency degree of the real-time operation state level.

The conflict type can be classified into time conflict, resource conflict and the like, and the emergency degree of the real-time running state level reflects the severity of the equipment problem. Reassigning the execution priority of the maintenance actions based on these factors can ensure that maintenance work can be performed efficiently and orderly. For example, for a substation device of an electric power system, a certain maintenance action to be adjusted needs to be rearranged due to time conflict, meanwhile, the real-time running state of the device is in a "fault" level, the emergency degree is high, and then the execution priority of the maintenance action should be correspondingly improved, and the execution is preferentially arranged.

And step S2500, reorganizing the execution sequence of the maintenance actions in the preliminary maintenance strategy scheme based on the reassigned priority, and generating an optimized dynamic maintenance strategy.

According to the redistributed maintenance action execution priority, the maintenance action execution sequence in the primary maintenance strategy scheme is adjusted and reorganized, so that maintenance work can be more reasonably and efficiently carried out. For example, for an automatic warehouse system device, after the execution priority of the maintenance action is reassigned, the maintenance action with high emergency degree is arranged in advance, and related maintenance actions are reasonably combined to generate an optimized dynamic maintenance strategy.

Step S2600, splitting the optimized dynamic maintenance strategy into a plurality of maintenance subtasks which can be independently executed, and distributing corresponding state evaluation index monitoring conditions for each maintenance subtask.

The optimized dynamic maintenance strategy is split into a plurality of maintenance subtasks which can be independently executed, so that maintenance personnel can operate and manage the maintenance subtasks conveniently. And corresponding state evaluation index monitoring conditions are allocated to each maintenance subtask, so that the execution condition of the maintenance subtask and the state change of the equipment can be monitored in real time. For example, for maintenance of a ship power system, the optimized dynamic maintenance strategy is split into a plurality of maintenance subtasks such as engine maintenance, oil circuit inspection, electrical system maintenance and the like, and corresponding state evaluation index monitoring conditions such as engine rotation speed, oil temperature, voltage and the like are allocated to each subtask.

Step S2700, whether the monitoring condition of the real-time monitoring state evaluation index meets a preset maintenance task activation threshold value or not, and triggering an execution instruction of a corresponding maintenance subtask when the monitoring condition meets the preset maintenance task activation threshold value.

The preset maintenance task activation threshold is a preset critical value, and is used for judging whether a certain maintenance subtask needs to be executed. The state is monitored in real time to evaluate the index monitoring condition, and when the index value reaches or exceeds a preset maintenance task activation threshold value, an execution instruction corresponding to the maintenance subtask is triggered, so that the maintenance work can be timely performed. For example, for maintenance of an air conditioning system, the refrigerating efficiency of the air conditioner is monitored in real time, and when the refrigerating efficiency is lower than a preset maintenance task activation threshold, an execution instruction of a maintenance subtask for cleaning and debugging the air conditioner is triggered.

Step S2800, continuously collecting key operation parameter variation data of the target device during the execution of the maintenance subtask, and generating a maintenance effect evaluation index.

In the execution process of the maintenance subtask, key operation parameter change data of the target equipment, such as temperature, pressure, rotating speed and the like, are continuously collected, and maintenance effect evaluation indexes are generated through analysis and processing of the data. These indicators may reflect the performance of maintenance subtasks and the status improvement of the device. For example, for a maintenance subtask of a machine tool, key operation parameter change data such as machining precision and vibration frequency of the machine tool are continuously collected in the maintenance process, and maintenance effect evaluation indexes such as machining precision improvement rate and vibration frequency reduction rate are generated.

Step S2900, according to the difference value between the maintenance effect evaluation index and the expected maintenance target, the trigger condition and the execution sequence of the unexecuted maintenance subtasks are adjusted.

The difference value between the maintenance effect evaluation index and the expected maintenance target reflects the difference between the execution effect of the maintenance subtask and the expected maintenance target. According to the difference value, the trigger conditions and the execution sequence of the unexecuted maintenance subtasks are adjusted, so that the maintenance work can more accord with the actual requirements of the equipment. For example, for maintenance of an industrial robot, after a part of maintenance subtasks is completed, it is found that a difference exists between a maintenance effect evaluation index and an expected maintenance target, for example, the motion accuracy improvement amplitude of the robot does not reach an expected value, at this time, the triggering condition of the maintenance subtasks which are not executed may be adjusted, for example, the execution threshold of some maintenance subtasks is lowered, or the execution sequence is adjusted, so that the maintenance subtasks which are more helpful for improving the motion accuracy are preferentially executed.

And step S3000, updating the real-time running state grade based on the latest state evaluation index value after all maintenance subtasks are executed, and feeding back to a dynamic adjustment process of a preset state threshold value.

After all maintenance subtasks are executed, the real-time operation state grade of the equipment is re-estimated according to the latest state estimation index values, such as the performance parameters, the operation stability and the like of the equipment, which are acquired. And feeding the updated real-time running state level back to a dynamic adjustment process of the preset state threshold value, further optimizing the preset state threshold value, and improving the accuracy of state evaluation. For example, for a wind generating set, after all maintenance subtasks are completed, state evaluation index values such as power output, vibration condition and the like of the set are collected, the real-time running state level of the wind generating set is reevaluated, for example, the wind generating set is updated from an early warning state to a normal state, and the updated information is fed back to a dynamic adjustment module of a preset state threshold value to adjust the related threshold value.

By continuously and circularly executing the steps, namely updating the real-time running state grade according to the latest state evaluation index value and feeding back to the dynamic adjustment process of the preset state threshold value, the multi-level state analysis model can be continuously optimized and improved, the running state change of the equipment can be better adapted, the accuracy and reliability of the equipment state analysis can be improved, and more effective support can be provided for the maintenance and management of the equipment. For example, in an equipment management system of a large manufacturing enterprise, through continuous feedback and adjustment, the fault early warning accuracy of equipment is continuously improved, the maintenance cost of the equipment is continuously reduced, and the production efficiency is improved.

Fig. 2 is a schematic diagram of a hardware entity of a data analysis system according to an embodiment of the present invention, and as shown in fig. 2, the hardware entity of the data analysis system 1000 includes a processor 1001 and a memory 1002, where the memory 1002 stores a computer program that can run on the processor 1001, and the processor 1001 implements steps in the method of any of the above embodiments when executing the program.

The memory 1002 stores a computer program executable on a processor, and the memory 1002 is configured to store instructions and applications executable by the processor 1001, and may also cache data (e.g., image data, audio data, voice communication data, and video communication data) to be processed or already processed by each module in the processor 1001 and the data analysis system 1000, which may be implemented by a FLASH memory (FLASH) or a random access memory (Random Access Memory, RAM).

The processor 1001 performs the steps of any one of the above-described device state data analysis methods applied to the automated electric system when executing the program. The processor 1001 generally controls the overall operation of the data analysis system 1000.

The foregoing is merely an embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily think about changes or substitutions within the technical scope of the present invention, and the changes and substitutions are intended to be covered by the scope of the present invention.

Claims (10)

1. A method for analyzing device state data applied to automated electrical equipment, the method comprising:

Acquiring a historical operation data set of target equipment, wherein the historical operation data set comprises historical data sequences of various equipment operation parameters and corresponding equipment state labels;

Constructing a multi-level state analysis model based on the historical operating dataset, the multi-level state analysis model comprising at least one first analysis level and at least one second analysis level, wherein each first analysis level corresponds to an operating parameter screening policy, each second analysis level corresponds to a state assessment policy, and the second analysis level is subsequent to the first analysis level;

extracting key operation parameters from the real-time operation data stream of the target equipment through an operation parameter screening strategy in the first analysis level, and generating a parameter screening result according to the key operation parameters;

Inputting the parameter screening result into a state evaluation strategy in the second analysis level, and generating a state evaluation index of the target equipment by combining a dynamic threshold range corresponding to the key operation parameter;

and outputting the real-time running state grade of the target equipment according to the comparison result of the state evaluation index and a preset state threshold value.

2. The method of claim 1, wherein the extracting key operating parameters from the real-time operating data stream of the target device by the operating parameter screening policy in the first analysis level and generating a parameter screening result according to the key operating parameters comprises:

Intercepting an operation parameter sequence in a current time window from the real-time operation data stream, wherein the operation parameter sequence comprises real-time measurement values of a plurality of equipment operation parameters;

Calculating a relevance coefficient of each of the plurality of device operating parameters to the device state tags in the historical operating dataset based on parameter relevance screening conditions in the first analysis hierarchy;

Screening candidate operation parameters with the association coefficient larger than a preset association threshold value from the plurality of equipment operation parameters, and determining a parameter stability index according to the fluctuation amplitude of the real-time measured value of the candidate operation parameters;

Sorting the candidate operation parameters according to the parameter stability index, and selecting a preset number of candidate operation parameters which are sorted in front as the key operation parameters;

And generating a parameter screening result comprising parameter weight distribution based on the real-time measured value of the key operation parameter, wherein the parameter weight is dynamically adjusted according to the relevancy coefficient and the parameter stability index.

3. The method of claim 2, wherein the calculating a relevance coefficient for each of the plurality of device operating parameters to the device state label in the historical operating dataset based on parameter relevance screening conditions in the first analysis level comprises:

Extracting data distribution characteristics of each equipment operation parameter under different equipment state labels from the historical operation data set, wherein the data distribution characteristics comprise parameter mean values, variances and distribution form indexes;

Calculating a distinguishing degree value of each equipment operation parameter among different equipment state labels according to the data distribution characteristics, wherein the distinguishing degree value is determined based on the combination of parameter mean value difference and variance ratio;

acquiring a parameter change trend curve of each equipment operation parameter in the historical operation data set, and extracting the number of peak points and trend inflection points in the parameter change trend curve;

Generating the relevance coefficient by combining the discrimination value, the peak value number and the trend inflection point position, wherein the relevance coefficient is in positive correlation with the discrimination value and the peak value number and in negative correlation with the dispersion of the trend inflection point position;

And carrying out normalization processing on the relevancy coefficient, and carrying out dynamic matching verification on the normalized relevancy coefficient and the time sequence change of the equipment state label in the historical operation data set.

4. A method according to claim 3, wherein said generating a parameter screening result comprising parameter weight assignment based on real-time measurements of said critical operating parameters comprises:

calculating a real-time offset according to the real-time measured value of the key operation parameter and the historical average value of the corresponding parameter in the historical operation data set;

Determining real-time anomaly probabilities for each critical operating parameter based on the real-time offset and the parameter stability indicator;

Generating a dynamic weight distribution proportion according to the real-time abnormal probability and the relevance coefficient, wherein the dynamic weight distribution proportion is positively correlated with the product of the real-time abnormal probability and the relevance coefficient;

Carrying out weighted fusion on the real-time measured value of the key operation parameter and the dynamic weight distribution proportion to generate a parameter screening result containing weighted parameters;

and generating a parameter screening code according to the weighted parameter values in the parameter screening result, wherein the parameter screening code is used for identifying the priority order of different key operation parameters in the second analysis level.

5. The method according to claim 1, wherein the inputting the parameter screening result into the state evaluation policy in the second analysis level, in combination with the dynamic threshold range corresponding to the key operation parameter, generates the state evaluation index of the target device, includes:

Analyzing the real-time measured value of the key operation parameter and the corresponding parameter weight distribution proportion from the parameter screening result;

Determining a dynamic threshold range of the key operation parameters according to parameter distribution intervals under different equipment state labels in the historical operation data set, wherein the dynamic threshold range is adaptively adjusted along with the operation time of equipment;

calculating the deviation degree of the real-time measured value and the upper limit value and the lower limit value of the dynamic threshold range, and generating a comprehensive deviation score by combining the parameter weight distribution proportion;

determining a state abnormality level corresponding to the key operation parameter according to a comparison result of the comprehensive deviation score and a preset deviation threshold;

And integrating the state anomaly levels of all the key operation parameters to generate an overall state evaluation index of the target equipment, wherein the overall state evaluation index comprises anomaly level distribution and anomaly duration.

6. The method of claim 5, wherein said determining a dynamic threshold range of said critical operating parameter comprises:

Extracting historical maximum values and minimum values of the key operation parameters in different equipment operation stages from the historical operation data set;

Calculating an initial threshold range according to the historical maximum value and the historical minimum value, and determining a current operation stage based on the real-time operation time of the target equipment;

acquiring a sliding average value and a sliding variance of real-time measured values of the key operation parameters in the current operation stage;

Dynamically correcting the initial threshold range according to the sliding average value and the sliding variance to generate a dynamic threshold range containing correction offset;

Matching and verifying the dynamic threshold range and the historical change rate of the key operation parameter so that the change rate of the dynamic threshold range does not exceed a preset rate limit;

Wherein the dynamically correcting the initial threshold range according to the sliding average value and the sliding variance to generate a dynamic threshold range including a correction offset includes:

calculating a difference between the sliding average value and a median value of the initial threshold range as a mean shift;

determining a variance modification coefficient according to the ratio of the sliding variance to the historical variance;

multiplying the mean shift amount by a variance correction coefficient to generate a dynamic correction amount;

Respectively adding the dynamic correction amount to the upper limit value and the lower limit value of the initial threshold range to generate a corrected dynamic threshold range;

And carrying out boundary constraint processing on the modified dynamic threshold range so that the upper limit value of the dynamic threshold range does not exceed the preset percentage of the history maximum value and the lower limit value is not lower than the preset percentage of the history minimum value.

7. The method according to claim 1, wherein the method further comprises:

adding a third analysis level in the multi-level state analysis model, the third analysis level being subsequent to the second analysis level;

predicting a future state evolution path of the target device according to the historical change trend of the state evaluation index through a state prediction strategy in the third analysis level;

Extracting key turning points in the future state evolution path, and generating preventive maintenance suggestions by combining real-time measured values of the key operation parameters;

and adjusting the response priority of the parameter screening strategy and the state evaluation strategy in the multi-level state analysis model according to the matching result of the preventive maintenance suggestion and the real-time running state level.

8. The method of claim 7, wherein predicting, by the state prediction strategy in the third analysis level, the future state evolution path of the target device according to the historical trend of the state evaluation index comprises:

Extracting time sequence data from the state evaluation index, wherein the time sequence data comprises state evaluation index values of the target equipment at different time points and real-time measured values of corresponding key operation parameters;

Performing multi-scale decomposition on the time series data through a state prediction strategy in the third analysis level to generate decomposed time series data comprising a long-term trend component, a periodic fluctuation component and a short-term noise component;

Identifying the overall degradation direction of the target device based on the long-term trend component, and extracting the running state change period of the target device by combining the periodic fluctuation component;

Generating a predictor model set according to the overall degradation direction and the running state change period, wherein the predictor model set comprises a plurality of predictor models matched with different degradation stages and periodic fluctuation characteristics;

Inputting the decomposed time series data into each predictor model in the predictor model set, generating future state prediction results under different confidence degrees, and performing cross verification on the future state prediction results;

Screening a target predictor model with minimum deviation from a latest state evaluation index value in a real-time operation data stream of the target equipment according to the cross verification result;

Carrying out trend correction on short-term noise components in the time sequence data through the target prediction sub-model, and generating a future state evolution path containing corrected noise influence;

Performing association matching on the future state evolution path and a dynamic threshold range of key operation parameters in the real-time operation data stream, and identifying an abnormal prediction interval exceeding the dynamic threshold range in the future state evolution path;

Based on the starting time point and the ending time point of the abnormal prediction interval and the corresponding key operation parameter types, preventive maintenance suggestions containing maintenance trigger conditions and maintenance parameter ranges are generated.

9. The method according to claim 1, wherein the method further comprises:

Setting a feedback optimization mechanism in the multi-level state analysis model, wherein the feedback optimization mechanism carries out iterative updating on the operation parameter screening strategy and the state evaluation strategy according to the output result of the real-time operation state level;

Obtaining a matching error between the real-time running state grade and an actual equipment maintenance record, and calculating model optimization weight according to the matching error;

Adjusting parameter relevance screening conditions in the first analysis level and dynamic threshold range generation rules in the second analysis level based on the model optimization weights;

applying the adjusted parameter correlation screening conditions and dynamic threshold range generation rules to the newly received real-time operation data stream to generate an updated multi-level state analysis model;

And performing backtracking verification on the historical operation data set through the updated multi-level state analysis model so that the prediction accuracy of the updated multi-level state analysis model is not lower than a preset accuracy threshold.

10. A data analysis system comprising a memory and a processor, the memory storing a computer program executable on the processor, wherein the processor performs the steps of the method of any one of claims 1 to 9 when the program is executed.