JPH07261838A - Performance monitoring method and performance monitoring device for plant equipment or system - Google Patents

Abstract

(57) [Summary] [Purpose] Performance monitoring method and performance of plant equipment or system that can collect and process plant operation data effectively by computer and accurately monitor long-term performance of plant equipment or system. Provide a monitoring device. According to a first aspect of the present invention, there is provided a performance monitoring method for plant equipment or a system, wherein a signal relating to the performance of the equipment or the system is scanned at a time interval smaller than a response time of the control system, and a detection value is preset. Calculate the average value, root mean square value, maximum value, and minimum value statistic for each storage time of multiple stages, save the calculation processing data, and deviate from the saved data the predetermined fluctuation at normal time. It is characterized in that the extracted signal is extracted, and whether or not the performance of the device or the system is normal is determined by corresponding to the performance determination formula specific to the device or the system provided in advance.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to performance monitoring of plant equipment or system, and more particularly to performance monitoring of plant equipment or system that collects and monitors data and performs diagnostic support in plant operation data taken into a computer. A method and a performance monitoring device.

[0002]

2. Description of the Related Art Trouble due to breakdown of various devices and control systems in a large-scale plant has been conventionally detected by an alarm device provided to protect the devices or systems. However, with only the alarm of trouble occurrence,
With equipment that causes such troubles,
It is impossible to detect the state of wear and deterioration, and these in the process of gradually advancing over a long period of time.

Therefore, long-term data monitoring is required for preventive maintenance for operating the plant constantly in a stable state, and in order to carry out this accurately, long-term appropriate data collection and processing and Evaluation is important.

[0004]

For the long-term data collection of the state and function of each component in a plant, recording data on a chart paper has been performed. However, on the chart paper, qualitative analysis of signals is performed. It is desirable to collect data by a computer in order to perform accurate monitoring, since it is only possible to evaluate it. However, if the collection interval is reduced to improve the monitoring accuracy, the amount of data will increase and the processing will be difficult even by a computer, and if the collection interval is increased, the collection data will become coarse and the monitoring accuracy will deteriorate. .

When a boiling water nuclear power plant is taken as an example of a large-scale plant, the process computer used to operate the nuclear power plant does not have a function of storing data. You have to rely on chart paper. However, it was difficult to find abnormal signs of plant equipment or system on the chart paper, analyze their tendency, and make an accurate judgment.

In addition, other monitoring systems and data collection devices usually collect data for a short period before and after the occurrence of a defect for the purpose of investigating the cause of the defect after the device trip or reactor scrum. Therefore, there has been a demand for development related to long-term data collection with high monitoring accuracy by a computer.

An object of the present invention is to monitor the performance of a plant device or system by collecting and processing the plant operation data effectively by a computer and accurately monitoring the long-term condition of the plant device or system. A method and a performance monitoring device are provided.

[0008]

In order to achieve the above object, a method of monitoring the performance of a plant device or system according to the invention of claim 1 provides a signal relating to the performance of the plant device or system to a control system of the plant device or system. Scan at a time interval smaller than the response time of, and calculate the statistical value of the average value, root mean square value, maximum value, and minimum value of the detection values obtained by the signal for each of a plurality of preset storage times , Save the calculation processing data, extract a signal that deviates from a predetermined normal fluctuation from the saved data, and make the extracted signal correspond to a pre-equipped device or system-specific performance judgment formula It is characterized by determining whether or not the performance of the device or system is normal.

According to a second aspect of the present invention, there is provided a performance monitoring apparatus for plant equipment or a system, wherein data collection means for inputting a signal relating to the performance of the plant equipment or system and performing sequential statistical calculation processing, and calculation in the data collection means. Storage means for storing the processed data for each preset cycle; abnormal signal extraction means for extracting a signal varying from the data stored for each cycle by the storage means by deviating from a preset threshold value; It is characterized by comprising a diagnosis support means for calculating whether or not the equipment or system is normal by calculating the data extracted by the abnormal signal extraction means by a performance determination formula for each equipment or system provided in advance.

[0010]

According to the present invention, the signal from the plant equipment or system is scanned at a time interval smaller than the response time of the control system of the plant equipment or system, and the detection value obtained by the signal is preset. In addition, the statistics of the average value, root mean square value, maximum value, and minimum value are calculated and saved for each of the multiple stages of storage time. By automatically extracting a signal that deviates from a predetermined normal fluctuation from the stored calculation data, and making the extracted signal correspond to a device or system-specific performance judgment formula provided in advance, the device or Determine whether the system performance is normal.

According to a second aspect of the invention, the data collecting means inputs a signal relating to the performance of the plant equipment or system,
Sequential statistical calculation processing is performed and output to the storage means. The saving means saves the calculation processing data from the data collecting means at preset intervals. The data stored by the storage means for each cycle extracts a signal that fluctuates outside the normal fluctuation range due to a threshold value in the abnormal signal extraction means, and a device provided in advance in the diagnosis support means. Alternatively, a performance judgment formula for each system is used to determine whether the device or system is normal.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to the drawings by taking a nuclear power plant as an example. This performance monitoring method, as shown in the schematic diagram of FIG.
The data collection step 1 and the storage step 2, the abnormal signal extraction step 3 and the diagnosis support step 4 are performed. In the data collection step 1, the input analog data 5 is scanned at a preset interval Δt. It should be noted that this interval Δt is fine enough to capture the characteristics of the data sufficiently, for example, it is generally set to an interval sufficiently faster than the response speed of the control system of the plant.

In the next storage step 2, the cycle ΔT = 1 hour is used to see the output increase characteristic at startup, and the cycle ΔT = 1 day is used to see the characteristic through one operation cycle at the rated output. As described above, the data storage period ΔT is determined according to the phenomenon to be monitored, and the data sampled in the data collection step 1 is statistically processed for each period ΔT to obtain an average value and a root mean square value (Root Mean). Squere Valu
e, hereinafter referred to as RMS value), and obtains and saves the maximum and minimum values during the period ΔT and their times.

In the abnormal signal extraction step 3, the data obtained in the storage step 2 for all the signals is screened by applying a threshold value to screen out normal signals, and the abnormal signals are fluctuated outside the normal range. Extract what is present. Further, in the diagnostic support step 4, the signal data screened in the abnormal signal extraction step 3 is used to calculate a performance determination formula for each device or system corresponding to the signal data, for example, as shown in a correlation diagram. It is displayed on a display unit (not shown) together with the determination result.

A performance monitoring apparatus for performing such a performance monitoring method is, as an embodiment, as shown in the block diagram of FIG.
The data collection means 1A and the storage means 2A, and the abnormal signal extraction (screening) means 3A and the diagnosis support means 4A. For example, the data collection means 1A and the storage means 2A are online minicomputers, and the abnormal signal extraction means 3A and the diagnosis support means. 4A is built on the workstation. The analog data 5a taken in from the sensor 5 is amplified by the amplifier 6, further finely sampled through the A / D converter 7, and output to the data collecting means 1A.

The data collecting means 1A comprises a data fetching section 8, a statistical calculation processing section 9 and a statistical processing result storing section 10. The data 7a is fetched online in the data fetching section 8 and is transferred to the storing means 2A. Signals are sent and received, and the average value, RMS value, maximum and minimum values are successively calculated.

The storing means 2A is the data collecting means 1A.
The calculation processing result, the periodic signal 11 preset by the user, and the data of the data collecting means 1A are exchanged to store and store the storage period, the storage period data 12, and the storage period determination unit.
13. The storage period determination unit 14 and the processed data file 15 are provided, and when the storage cycle preset by the user comes, the calculation result at that time is stored as a value at that time. The data of the processed data file 15 accumulated in this way is transmitted to the abnormal signal extracting means 3A via the Ethernet 16 which is a high speed communication line forming a local area network.

The abnormal signal extraction means 3A comprises the storage means 2
RMS data reading section 17 for inputting data from A via Ethernet 16, threshold value data 19 for inputting threshold value signal 18 preset by the user, threshold value comparison processing section 20, and comparison result storage section 21 and extracted signal data 22. In this abnormal signal extraction means 3A, in order to find a signal that deviates from the normal fluctuation range in the data from the storage means 2A, the RMS data among the data thus obtained is previously detected. Apply the set threshold and extract those that exceed this threshold.

The diagnostic support means 4A includes a signal name comparison unit 23, a coincidence determination unit 24, a diagnostic item priority determination unit 25, a diagnostic item selection unit 26, a normal / abnormal determination processing unit 27, a diagnostic result storage unit 28,
Diagnosis table 29, reference data and past case data 30,
A diagnostic result display section 31, a diagnostic result and detailed information 32 are provided, and a printer and a CRT 33 as display means are provided. In the diagnosis support means 4A, the data extracted by the abnormal signal extraction means 3A is processed by the signal name comparison section 26 and the normal / abnormality determination processing section.
It is configured to input to 30 to select a device, system, or item to be diagnosed, determine whether it is normal or abnormal, and display the diagnostic result to the user via the printer and CRT 33 which are display means. Has been done.

Next, the function of the main means of the performance monitoring apparatus will be described in detail with respect to the operation of the above configuration.
The data collection means 1A and the storage means 2A are shown in the block diagram of FIG. 3 and the procedure flow chart of FIG. First, if necessary, the user gives the periodic signal 11 the storage cycle of the storage means 2A and the period data 12, and the storage means 2 is stored as shown in FIG.
In A, the data storage period ΔT is set. For example, areas are prepared on the disks 34, 35, 36, 37 for each time constant ΔT.

Taking a nuclear power plant as an example, the time constant ΔT, which is a data storage period, is 1 min for disk 34, 10 min for disk 35, 1 hour for disk 36, and disk 37.
Then, set 1day, and set that each can save 500 points each. As a result, the data length of each disc is
Disk 34 is 500 min, disk 35 is 5000 mi
n, disk 36 500 hours, disk 37 500
It becomes day.

Here, in the data collecting means 1A, the sampling interval Δt = 100 msec is set, and the data scanned as shown in FIG. 4 is sequentially subjected to statistical calculation processing, and the average value and the RMS value are calculated at each time point for each cycle ΔT. Save the value, maximum value, minimum value, and the time in each area, and overwrite when the number of data points exceeds 500 points.

If the data is stored in the storage means 2A in this manner, the monitoring accuracy is sufficiently maintained and the maximum length is 5
Data for 00 days will be collected together. In a nuclear power plant, one operating cycle is about one year,
As a result, one cycle can be sufficiently covered. The average value Y used as the statistic here
_n is defined by the following equation (1), the RMS value σ _n is defined by the equations (2) and (3), and the weight W _n is obtained by the equation (4).

Y _n = Y _n-1 + W _n × (X _n −Y _n-1 ) ... (1) σ ² _n = √σ ² _n-1 + W _n × ((X _n −Y _n-1 ) ² −σ ² _n-1 ) (2) σ _n = √σ ² _n (3) W _n = MAX (1 / _n , t / ΔT) (4)

Here, X _n is sample data, W _n is weight, σ _n is RMS value, _n is number of data, t is sampling interval, and ΔT is data storage period. Next, in the screening in the abnormal signal extraction means 3A, as shown in the procedure flow chart of FIG. 5, the data accumulated in the storage means 2A is read and compared with a preset threshold value, and this value is exceeded. Register signals one by one.

Here, the threshold value is an upper limit value and a lower limit value that define a range regarded as normal with respect to the signal data.
It was given to 22. In addition, in the example of FIG. 5, the RMS value is used among the four statistic values, because the reason is that the sensitivity is highest in extracting the fluctuation when monitoring the long-term trend.

An example of the pattern of abnormal data extracted by this screening is shown in the characteristic diagram of FIG. Screening by RMS value is applied to each of these three patterns to prove its effectiveness. FIG. 6 (a) shows the pattern of average value drift over time, and this curve
For anomalies of the type that gradually change in characteristics over a long period, such as 38, it is difficult to detect changes in a short period of time. Also,
Since it may not be detected from the fluctuation of the data, it is effective to monitor the average value in the long term.

FIG. 6B shows a pattern in which abnormal data suddenly appears. Since this abnormal data instantaneously appears like abnormal peaks 39a and 39b on the curve 39, it is difficult to detect the average value or the RMS value. Therefore, the abnormal data can be detected by detecting the abnormal peaks 39a, 39b that exceed the thresholds 40a, 40b set in advance and monitoring the maximum value and the minimum value.

FIG. 6C shows an abnormal fluctuation pattern. In such a case, the width of the fluctuation gradually or abruptly changes as shown by the curve 41, and becomes constant on average. Therefore, it is effective to monitor the RMS value. In this way, the three main abnormal phenomena are:
Since this combination is also possible, detection can be performed by monitoring the average value, RMS value, maximum value, and minimum value.

In the diagnosis support means 4A, the performance of the abnormal signal extracted by the abnormal signal extraction means 3A is evaluated by examining the performance determination method specific to each device or system. For example, a diagnostic table showing the correlation between process signals is prepared, a correlation formula of the extracted abnormal signal and the process signal related thereto is calculated, and this is illustrated and the performance is evaluated. Taking a nuclear power plant as an example, if the extracted signal is used as the valve opening, the correlation equation with the control signal can be obtained from this diagnostic table, and the performance can be determined by plotting it and comparing it with the normal condition. You can

As shown in the procedure flow chart of FIG. 7, a diagnostic table is prepared in advance in which performance determining methods unique to each device or system, such as correlation equations between process signals, are arranged. For each item, a group of signals required for performance determination is compared with the signal extracted by the screening of the abnormal signal extracting means 3A, and the priority of the diagnostic item is determined and presented based on the degree of coincidence.

The user selects an item based on the result,
By referring to the item selected by the diagnosis support means 4A, the data is processed according to the method and compared with the case of the reference data, and the result is displayed on the CRT 33.
Further, this result is stored in the database together with detailed information as a diagnosis result.

The screen diagram of FIG. 8 shows the diagnosis result of the valve as a display example. The flow of the diagnosis is as shown in screen A and screen B of FIG.
With respect to the data extracted by the screening of the time-series data as described above, the correlation display is performed in the form of the screen D of FIG. 8 according to the diagnosis table of the screen C of FIG.

Finally, the screens A, B, and C of FIG. 8 are presented to the user as a reference together with the determination of normality and abnormality. Here, the screen A of FIG. 8 shows the average value of the signal in a time series by the curve 42, the RMS value by the curve 43, the maximum value by the curve 44, and the minimum value by the curve 45. On the screen B of FIG. 8, the average, RMS, maximum and minimum values of the signal are calculated, respectively, and the maximum, minimum, average and RMS values are calculated. By displaying the value and the minimum value at the same time, the comparison on the CRT screen is easy and the abnormality can be easily detected.

The screen C of FIG. 8 is a diagnostic table, and signal correlation or the like is used as an index for judging the performance of each device or system. Therefore, a list of signals corresponding to the X axis and the Y axis is provided for each item. These are organized, for example, the valve performance of diagnostic items46,
A control signal 47 of a signal corresponding to the X axis and a valve opening degree 48 of a signal corresponding to the Y axis are displayed.

The screen D of FIG. 8 is an example in which the correlation display is performed based on the screen C of FIG. 8. The white circle in the screen is the signal data 49, and the straight line is the fitting curve 50 (or straight line) of the reference data.
Is. In this example, the control signal 47 is plotted on the X axis and the valve opening 48 is plotted on the Y axis according to the diagnostic table on screen C of FIG.
Since the fitting curve 50 (a straight line in this case) of the reference data which is obtained in advance is displayed at the same time, it is possible to easily judge whether or not the fitting is normal by the degree of deviation from the signal data reference by comparing these.

[0037]

As described above, according to the present invention, since the data sampled finely is stored while being processed, there is an advantage that the monitoring accuracy can be maintained high and the data amount can be small. And it can be monitored in various periods depending on the system.

Further, all signal data are processed into four statistical quantities, and a series of operations up to performance judgment are automatically performed to efficiently monitor the performance of plant equipment and system. Since it is possible to detect the abnormality of the performance of the plant equipment and the system before the trouble, and to support the behavior prediction and the cause investigation after that, there is an effect that the reliability of the plant operation is improved.

[Brief description of drawings]

FIG. 1 is a schematic diagram showing an outline of a processing flow according to an embodiment of the present invention.

FIG. 2 is a block configuration diagram of a performance monitoring device according to an embodiment of the present invention.

FIG. 3 is a block configuration diagram of a storage unit according to an embodiment of the present invention.

FIG. 4 is a procedure flow chart of a data collecting unit and a storing unit according to an embodiment of the present invention.

FIG. 5 is a procedure flow chart of the abnormal signal extracting means of the embodiment according to the present invention.

FIG. 6 is a characteristic diagram of an example of an abnormal data pattern according to an embodiment of the present invention, (a) shows a drift of an average value, (b) shows a sudden abnormality, and (c) shows a fluctuation abnormality.

FIG. 7 is a procedure flow chart of the diagnostic monitoring means of the embodiment according to the present invention.

FIG. 8 is a display screen diagram of an embodiment according to the present invention.

[Explanation of symbols]

1 ... Data collection step, 1A ... Data collection means, 2 ...
Storage step, 2A ... Storage means, 3 ... Abnormal signal extraction step, 3A ... Abnormal signal extraction means, 4 ... Diagnostic monitoring step, 4A ... Diagnostic monitoring means, 5 ... Sensor, 6 ... Amplifier, 7
... A / D converter, 8 ... Data acquisition section, 9 ... Statistical calculation processing section, 10 ... Statistical processing result storage section, 11 ... Period signal, 12 ... Storage cycle and storage period data, 13 ... Storage cycle determination section, 14 … Retention period judgment part, 15… Processed data file, 16… Et
Hernet, 17 ... RMS data reading unit, 18 ... Threshold signal, 19 ... Threshold data, 20 ... Threshold comparison processing unit, 21 ... Comparison result storage unit, 22 ... Extracted signal data, 23 ... Signal name comparison Section, 24 ... coincidence determination section, 25 ... priority determination section, 26
... diagnosis item selection section, 27 ... normality and abnormality determination processing section, 28 ... diagnosis result storage section, 29 ... diagnosis table, 30 ... reference data and past case data, 31 ... diagnosis result display section, 32 ... diagnosis result Detailed information, 33 ... CRT, 34-37 ... Disc, 38, 39,
41 to 45 ... Curve, 39a, 39b ... Abnormal peak, 40a, 40b ...
Threshold, 46 ... Valve performance, 47 ... X-axis control signal, 48 ... Valve opening, 49 ... Signal data, 50 ... Fitting curve.

Claims (2)

[Claims] 1. A signal relating to the performance of a plant device or a system is scanned at a time interval smaller than a response time of a control system of the plant device or the system, and a detection value obtained by the signal is stored in a plurality of preset stages. Calculate the statistic of the mean value, root mean square value, maximum value, and minimum value for each time, save the calculation processing data, and extract the signal that deviates from the predetermined normal time fluctuation from this saved data. A method for monitoring the performance of a plant device or a grid, which comprises extracting the extracted signal and determining whether or not the performance of the device or the grid is normal by making the extracted signal correspond to a performance judgment formula unique to the device or the grid. . 2. A data collection means for inputting a signal relating to the performance of plant equipment or a system to perform a sequential statistical calculation process, a storage means for storing the calculation process data in the data collection means in a preset cycle, and the storage. An abnormal signal extracting means for extracting a signal that fluctuates by deviating from a preset threshold value from data stored for each cycle by means, and a device or system provided with the data extracted by the abnormal signal extracting means in advance 2. A performance monitoring device for plant equipment or a system, comprising: a diagnosis assisting means for calculating whether or not the equipment or the system is normal by calculating with the performance determination formula of FIG.