WO2006016440A1 - データ処理方法及びデータ処理装置、並びに診断方法及び診断装置 - Google Patents
データ処理方法及びデータ処理装置、並びに診断方法及び診断装置 Download PDFInfo
- Publication number
- WO2006016440A1 WO2006016440A1 PCT/JP2005/008172 JP2005008172W WO2006016440A1 WO 2006016440 A1 WO2006016440 A1 WO 2006016440A1 JP 2005008172 W JP2005008172 W JP 2005008172W WO 2006016440 A1 WO2006016440 A1 WO 2006016440A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- operation mode
- self
- map
- abnormal
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
- G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
- G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
-
- E—FIXED CONSTRUCTIONS
- E02—HYDRAULIC ENGINEERING; FOUNDATIONS; SOIL SHIFTING
- E02F—DREDGING; SOIL-SHIFTING
- E02F9/00—Component parts of dredgers or soil-shifting machines, not restricted to one of the kinds covered by groups E02F3/00 - E02F7/00
- E02F9/26—Indicating devices
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0208—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system
- G05B23/021—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the configuration of the monitoring system adopting a different treatment of each operating region or a different mode of the monitored system, e.g. transient modes; different operating configurations of monitored system
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B23/00—Testing or monitoring of control systems or parts thereof
- G05B23/02—Electric testing or monitoring
- G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
- G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
- G05B23/0283—Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL]
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0499—Feedforward networks
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/082—Learning methods modifying the architecture, e.g. adding, deleting or silencing nodes or connections
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
Definitions
- Data processing method, data processing apparatus, diagnostic method, and diagnostic apparatus are Data processing method, data processing apparatus, diagnostic method, and diagnostic apparatus
- the present invention relates to a data processing method, a data processing apparatus, and a data processing method suitable for determining an abnormal state that occurs in a target machine that can operate in a plurality of operation modes, particularly a working machine such as a hydraulic excavator.
- the present invention relates to a diagnostic method and a diagnostic apparatus.
- Predictive maintenance refers to predicting deterioration and remaining life by diagnosing soundness by inference based on understanding of the load during operation'environmental information, past maintenance history database, failure physics, etc.
- Patent Document 1 relates to an abnormality diagnosis device for a work vehicle such as a construction machine, and a pressure sensor for detecting a discharge pressure of a hydraulic pump and a rotation speed of an engine on a vehicle body of the work machine (hydraulic excavator). It is equipped with an engine speed sensor, an oil temperature sensor that detects the oil temperature in the hydraulic circuit, and a communication device that wirelessly transmits detection information of each sensor force to the network control station.
- an engine speed sensor an oil temperature sensor that detects the oil temperature in the hydraulic circuit
- a communication device that wirelessly transmits detection information of each sensor force to the network control station.
- Patent Document 2 relates to an abnormality detection device for fixed machine equipment such as a Notch plant or continuous plant, and normally collects normal data when the target plant is in a normal state.
- a feature map that represents the distance relationship between output units based on the features of normal data extracted using a self-organizing map Is stored as a normal state model, and a technique for detecting an abnormality of the target plant based on the normal state model and unknown input data (input vector) is disclosed.
- the normal state model is obtained by converting multi-dimensional data into a visualized two-dimensional map as shown in FIG. 20, for example.
- Patent Document 1 Japanese Patent Laid-Open No. 2002-323013
- Patent Document 2 Japanese Patent Laid-Open No. 11-338848
- the vehicle body moves forward or backward or turns.
- detection elements such as the operating pressure that controls the packet, the operating pressure of the bucket cylinder that controls the packet, the operating pressure of the stick cylinder that controls the stick, and the operating pressure of the boom cylinder that controls the boom .
- the work machine When the work machine performs a series of work, the work machine performs the work by combining various operation operations (that is, operation modes). For example, when loading accumulated sediment on a truck vessel (loading platform) using a work machine, it can be roughly divided into the following four modes of operation.
- operation modes For example, when loading accumulated sediment on a truck vessel (loading platform) using a work machine, it can be roughly divided into the following four modes of operation.
- each parameter value changes according to each of such operation modes, but accurate abnormality diagnosis may not be performed even if each parameter value is analyzed separately.
- the current driving operation may not apply to any of the above four operating modes. In this case, it is recognized that the current operation is an unknown operation mode, or that some abnormality has occurred.
- the above-described diagnostic apparatus is not limited to application to machinery such as the work machine, and the operation (or fluctuation) is not limited to a plurality of operation modes. It can be applied to many diagnostic objects (subjects) that can be classified by the mode (or fluctuating mode).
- a self-organizing map is formed as a separate separation model for each of the operation modes using the data detection step and the normal data sets of the plurality ⁇ a detected in the normal data detection step.
- the value of each parameter in the normal data set detected in the organization map formation step and the normal data detection step is set to a plurality (E) of different values with the same number of fluctuation values as the number of parameters.
- An abnormal data creation step for creating a virtual abnormal data set at the time of abnormality of the target object by the number of the fluctuation vectors (E) in the one normal data set by increasing / decreasing by the variation vector;
- the abnormal data generation step from the self-organization map for each operation mode formed in the self-organization map formation step.
- a self-woven and weaving map with the highest similarity to the created abnormal data set is obtained for each abnormal data set, and for each variation vector, the operation mode for each operation mode is determined. It is characterized by comprising an abnormal operation mode ratio calculation step for calculating an abnormal operation mode ratio vector representing a ratio.
- the diagnosis method of the present invention as set forth in claim 2 uses the correspondence relationship between each variation vector obtained by the data processing method according to claim 1 and each abnormal operation mode ratio vector. It is characterized by the diagnosis of!
- the diagnosis method of the present invention described in claim 3 is the method according to claim 2, comprising the values of the n parameters (P 1, P 2,..., P 3) during actual operation of the object. During actual operation
- the self-organization map with the highest degree of similarity to the data set is obtained for each of the above-mentioned actual operation data sets, and the actual operation mode ratio vector representing the ratio of each operation mode to the above all operation modes is calculated.
- the data processing device of the present invention is configured such that n parameter values (p 1, P 2,..., P, which fluctuate according to the operation of an object that can operate in a plurality of operation modes. )
- Detecting means for detecting the set (d [P 1, P 2,..., P]) for each operation mode, and i 1 2 n
- Self-organizing map forming means for forming a self-organizing map as an individual separation model, and the self-organizing map for each operation mode formed by the self-organizing map forming means.
- An operation mode that selects the self-organization map that has the highest similarity to the input data set and represents the ratio of each operation mode to the above-mentioned all operation modes for the above-mentioned multiple data sets.
- An operation mode ratio calculation means for calculating a ratio vector, and the value of each parameter of the data set detected by the detection means during the preliminary operation in a normal state of the target object, In other words, by increasing or decreasing by a plurality of different (E) variation vectors, a virtual abnormality data set at the time of abnormality of the object is converted into the above one data set.
- the abnormal mode data creation means for creating only the number of fluctuation vectors (E) is included in the sett, and the operation mode ratio calculation means, when the abnormal data set is input by the abnormal data creation means,
- the self-organization map having the highest similarity to the abnormal data set is selected from the above-mentioned self-organization map for each operation mode formed by the self-organization map formation means, It functions to calculate an abnormal operation mode ratio vector representing a ratio of each operation mode to the above-mentioned all operation modes for the plurality of abnormal data sets.
- the diagnosis device of the present invention is the data processing device according to claim 4, and a plurality of sets of actual operation data sets detected by the detection means during actual operation of the object.
- the self-organization map that has the highest similarity to the actual operation data set Obtained for each actual operation data set obtains the actual operation mode ratio vector representing the ratio of each operation mode to the above all operation modes, and calculated by the operation mode ratio calculation means described above.
- each abnormal operation mode ratio vector a variation vector having the highest similarity to the above-described actual operation mode ratio vector is selected, and each variation value of the variation vector corresponding to the selected abnormal operation mode ratio vector is calculated. It is characterized by having a determination means for determining abnormality of the object by obtaining.
- the abnormality of the object can be determined using the correspondence between the abnormal operation mode ratio vector and the variation vector that increases or decreases the value of each parameter. And can be diagnosed more accurately.
- FIG. 1 is a block diagram showing a diagnostic apparatus according to an embodiment of the present invention.
- FIG. 2 is a flowchart showing a diagnostic method according to an embodiment of the present invention.
- FIG. 3 is a flowchart showing specific processing of a self-organization map forming step.
- FIG. 4 is a graph showing output values of sensors for operation modes 1 to 4 of a hydraulic excavator according to an embodiment of the present invention.
- FIG. 5 is a flowchart showing a specific process of a self-organization map forming step.
- FIG. 6 is a diagram visually showing a minimum distance between a training data point (detected data point) and -Euron in the self-woven and weaving map according to one embodiment of the present invention.
- FIG. 7 (a) A diagram for explaining a self-organizing map according to an embodiment of the present invention.
- training data for engine speed P and left hydraulic pump pressure P in operation mode 1 are used.
- FIG. 7 (b) is a diagram for explaining a self-organizing map according to an embodiment of the present invention.
- training data for engine speed P and right hydraulic pump pressure P in operation mode 1 are used.
- FIG. 7 (c) A diagram for explaining the self-organization map according to the embodiment of the present invention, and is a training data for the left hydraulic pump pressure P and the right hydraulic pump pressure P in the operation mode 1.
- [7 (d)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, and is based on training data for engine speed P and fuel consumption P in operation mode 1.
- ⁇ 8 (a)] is a diagram for explaining the self-organization map according to the embodiment of the present invention, and is based on the training data of the engine speed P and the left hydraulic pump pressure P in the operation mode 2.
- ⁇ 8 (b)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention.
- FIG. 8 (c)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, and is a training data for the left hydraulic pump pressure P and the right hydraulic pump pressure P in operation mode 2.
- [8] (d)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, which is based on training data for engine speed P and fuel consumption P in operation mode 2.
- ⁇ 9 (a)] is a diagram for explaining a self-organizing map according to one embodiment of the present invention, in which training data for engine speed P and left hydraulic pump pressure P in operation mode 1 (
- ⁇ 9 (b)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, in which training data for engine speed P and right hydraulic pump pressure P in operation mode 1 (
- FIG. 9 (c)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, and is a training data for the left hydraulic pump pressure P and the right hydraulic pump pressure P in the operation mode 1. (Small dot in the figure), complete learning and elimination of idling neurons FIG.
- FIG. 10 (a)] is a diagram for explaining a self-organizing map according to an embodiment of the present invention, and is a training data for the engine speed P and the left hydraulic pump pressure P in operation mode 2.
- FIG. 10 (b)] is a diagram for explaining a self-organizing map according to one embodiment of the present invention, and is a training data for engine speed P and right hydraulic pump pressure P in operation mode 2.
- FIG. 10 (c)] is a diagram for explaining the self-organization map according to the embodiment of the present invention, and is a training data for the left hydraulic pump pressure P and the right hydraulic pump pressure P in operation mode 2.
- ⁇ 10 (d)] is a diagram for explaining the self-organization map according to one embodiment of the present invention, and is training data for engine speed P and fuel consumption P in operation mode 2 (in the figure
- FIG. 11 is a diagram showing a variable parameter vector model map according to an embodiment of the present invention.
- FIG. 12 is a flowchart showing a specific process for calculating an operation mode ratio shown in FIG.
- FIG. 13 is a diagram showing an operation mode ratio vector model map according to one embodiment of the present invention.
- FIG. 14 shows an operation mode for parameter P of abnormal data according to an embodiment of the present invention.
- FIG. 15 shows an operation mode for parameter P of abnormal data according to an embodiment of the present invention.
- FIG. 1 A first figure.
- FIG. 16 is a diagram plotting ratios of operation modes 1 and 2 according to an embodiment of the present invention.
- FIG. 17 is a flowchart showing specific processing of a determination step.
- FIG. 18 is a schematic diagram for explaining a diagnostic method according to an embodiment of the present invention.
- FIG. 19 is a view showing a diagnostic apparatus as a modification of the present invention.
- FIG. 20 is a diagram showing a conventional self-organizing map (visualized two-dimensional map).
- FIG. 1 and FIG. 2 are diagrams for explaining a diagnostic apparatus according to an embodiment of the present invention.
- FIG. 1 is a block diagram thereof
- FIG. 2 is a flowchart showing the processing.
- This diagnosis device is provided in machinery such as a work machine, for example, and diagnoses how much deterioration or abnormality has occurred in which part of the machinery.
- a diagnostic device when applied to a hydraulic excavator of a work machine will be described. Note that the scope of application of this diagnostic device is not limited to this, but operates in multiple operating modes (variation modes) ( It can be applied to all the various objects that can be changed.
- the diagnostic apparatus includes a plurality of sensors (detection means) la to Ld, a self-organizing map forming unit 2, a storage unit 3, a determination unit 4, It is composed mainly of an ECU (Electronic Control Unit) 5 having various functions corresponding to the abnormal data creation means 7 and the operation mode ratio calculation means 8 and a display device 6.
- the ECU 5 includes an input / output device, a storage device (RAM, ROM, etc.) with a built-in processing program, a central processing unit (CPU), and the like.
- the sensor la ⁇ : Ld, self-organization map forming means 2, abnormal data creating means 7 and operation mode ratio calculating means 8 constitute a data processing device.
- Sensors la to: Ld is provided corresponding to each parameter (variation factor) related to the hydraulic excavator, and varies according to the operation of the hydraulic excavator when the hydraulic excavator can operate in a plurality of operation modes.
- the parameter value is detected for each operation mode.
- the sensor la ⁇ : Ld includes not only the value of the corresponding parameter that is directly detected, but also the value of the corresponding parameter that is obtained as an estimated value by processing certain detection data by calculation or the like.
- the parameters relating to the hydraulic excavator here include, for example, engine speed, fuel consumption, hydraulic pump pressure (one or more hydraulic pump pressures), oil temperature in the hydraulic circuit, vehicle body advance, Fluctuates according to hydraulic excavator operation, such as the operating pressure that controls reverse or turning, the operating pressure of the bucket cylinder that controls the packet, the operating pressure of the stick cylinder that controls the stick, and the operating pressure of the boom cylinder that controls the boom About each element
- sensors la to Ld that detect values of engine speed, fuel consumption, and hydraulic pump pressure are provided. That is, the engine speed sensor la for detecting the engine speed, the fuel consumption sensor lb for detecting the fuel consumption, and the left hydraulic pump pressure sensor lc for detecting the left and right hydraulic pump pressures of the excavator, respectively. , The right hydraulic pump pressure sensor Id and four sensors la ⁇ Id. Of course, a sensor for detecting the operating pressure of the bucket cylinder, stick cylinder, boom cylinder or the like as described above may be provided.
- the self-organizing map forming means 2 detection based on a plurality of parameter values detected by the engine speed sensor la, the fuel consumption sensor lb, the left hydraulic pump pressure sensor lc, and the right hydraulic pump pressure sensor Id
- the training data (Training Data) is used to form a self-organizing map (hereinafter also referred to as SOM) as an individual separation model corresponding to each operation mode of hydraulic excavators. Yes.
- each operation mode of the hydraulic excavator means a certain fixed operation (specific operation), for example, a series of operations of loading accumulated sediment on a truck vessel (loading platform).
- operation modes ⁇ Starting and scooping up the sand and sand with the packet and moving it to the end (operation mode 1) '' ⁇ Transporting operation (operation mode 2) '', ⁇ Opening the packet, starting the transfer of earth and sand onto the vessel and completing the transfer of the earth and sand with the force (operation mode 3) '', ⁇ Accumulating packets
- operation mode 0 a description will be given of a case where the excavator operates in five operation modes including the “idling (standby state) (operation mode 0)” mode.
- the self-organization map generally refers to a recognition model in which multidimensional data is represented in two-dimensional space and visualized. It can be used as a means of classifying (classifying) itself into classes prepared in advance.
- D data group (or data group) obtained by measurement ⁇ d, d, ..., d, ..., d ⁇
- the training data is used to learn the self-organizing map (recognition model) (ie, gradually update the self-organizing map), and the method of repeating this learning is called “supervised learning” .
- the self-organization map obtained in this way is used as a means to solve the classification problem.
- the more accurate training data is used, the more accurate self-organization map can be constructed.
- the amount of training data is limited to some extent. If the data amount is reached, the accuracy of the self-organizing map will be slightly improved even if the amount of data is further increased. Therefore, it is preferable to set the number of training data inputs to a predetermined number. Further, the “class” here corresponds to the “operation mode” in the present embodiment.
- OM (SOM, SOM, ..., SOM) is formed. Therefore, in the present embodiment, five blocks j 1 2 z
- One self-organizing map is created for each of the laths (operation modes).
- Each self-organizing map as a separation model is learned using a large amount of training data that clearly represents only one operation mode.
- Detection data from Ld is four (n) parameter values indicating the momentary state of the hydraulic excavator d (k) And the time differential value (parameter) of 4 (n) parameter values indicating the instantaneous change state of the hydraulic excavator (Including the time derivative equivalent value such as the rate of change of meter value) A d (k), and these four parameter values d (k) themselves and the time derivative values of the four parameter values ⁇ d (k) It is composed as 8 dimensional (2n dimensional) data.
- the self-organizing map forming means 2 the parameter value d (k) at the current time obtained by using only the parameter value d (k) at the current time and the parameter value d at the time prior to the current time.
- the parameter value d (k) at the current time alone cannot provide sufficient representative information for the overall dynamic operation of the excavator.
- ⁇ d (k) must be taken into account.
- d (k) and ⁇ d (k) t need to form a double-sized self-organizing map, which requires a long learning time. Since it only needs to be performed once when learning the organization map, there is no burden on the equipment when determining the operation mode while the excavator is operating.
- the self-organization map forming means 2 randomly arranges a predetermined number of eulons in an 8-dimensional (2n-dimensional) space, learns them using the training data described above, and sets the training data points.
- -Minimum distance from Euron -Euron is the winner -Euron is the winner for each training data point -Eron is determined to form a self-organizing map candidate and the above self-organizing map candidate From the multiple self-organization key map candidates obtained by performing multiple formations (predetermined number), the one closest to the training data characteristic is selected as the self-organization key map! /,
- the self-organization map forming means 2 randomly arranges a predetermined number of eulons in an 8-dimensional (2n-dimensional) space, learns them using the training data described above, and sets the training data points.
- -Minimum distance from Euron -Euron is the winner -Euron is the winner for each training data point -Eron is determined to form a self-organizing map candidate and the
- the self-organization key map forming means 2 determines the average value of the distance between the training data point and each winner-Euron and the training data for each of the above self-organization key map candidates! Calculate the standard deviation of the distance between the points and each of the above-mentioned winner neurons, and select the smallest self-organizing map candidate as the self-organizing map for either the average value or the standard deviation. It has become.
- the winner-Euron here means all of the winners that have a history of the winner neuron (that is, who have ever become a winner-Euron).
- the self-organization map forming means 2 determines that the average value is the smallest when there is no smallest self-organization map map candidate for any of the above average value and standard deviation. Can be selected as a self-organizing map candidate! /
- the self-organization map forming means 2 erases the euron that has never been a winner-Euron among the neurons in the selected self-organization map.
- the abnormal time data creation means 7 sets the value of each parameter of the data point detected by the sensor la to Ld to the same number of fluctuation values as the number of these parameters ( Both the fluctuation rate and the fluctuation vector are increased or decreased by a plurality of (E) different fluctuation vectors, so that the virtual abnormality data point at the time of the hydraulic excavator abnormality can be changed for each detection data point. Create only number (E)
- the operation mode ratio calculation means 8 has the highest similarity to the input data point among the self-organization map for each operation mode formed by the self-organization map formation means 2. A self-organization map is selected, and an operation mode ratio vector representing the ratio of each operation mode to all operation modes is calculated for a plurality of sets of data points. Further, the operation mode ratio calculation means 8 receives the abnormal data point from the abnormal data creation means 7 and the self-organization map of each operation mode formed by the self-organization map formation means 2 is obtained. The self-organization map with the highest similarity to the abnormal data point is selected from among the above, and each of the above-mentioned multiple sets of abnormal data points for each operation mode is selected. It functions to calculate an abnormal operation mode ratio vector representing the operation mode ratio.
- the determination means 4 acquires a plurality of sets of data points during actual operation detected by the sensors la ⁇ : Ld during actual operation of the hydraulic excavator.
- the determination means 4 has the self-organization map having the highest similarity to the data points during actual operation among the self-organization map for each operation mode calculated by the operation mode ratio calculation means 8. Obtain for each operating data point and obtain the actual operating mode ratio vector that represents the ratio of each operating mode to all operating modes. Then, the determination unit 4 selects the variation vector having the highest similarity to the actual operation mode ratio vector from among the abnormal operation mode ratio vectors calculated by the operation mode ratio calculation unit 8, and selects the selected vector. Abnormalities in the hydraulic excavator are determined by obtaining each fluctuation value of the fluctuation vector corresponding to the abnormal operation mode ratio beta.
- the display device 6 can display the determination result in the determination means 4.
- the diagnostic device is configured as described above, and the processing is executed along the flow shown in FIG.
- the flow shown in Fig. 2 is as follows: self-organization map creation step (step W1), abnormal data creation step (step W2), operation mode ratio calculation step (step W3), judgment step (step W4) This will be explained.
- Step W1 the self-organization map formation step shown in Fig. 2 will be explained.
- each operation mode of the hydraulic excavator is clarified by the self-organization map formation means 2.
- a self-organizing map is created as a separation model for each operation mode.
- the self-organization map forming step is performed in the off-line state of the hydraulic excavator as described above. More specifically, as shown in FIG. 3, the data creation detection step (step S100), the calculation step (step S110), forming step (step S120).
- step S 100 a large amount of highly reliable detection data is acquired for each operation mode of the excavator. That is, this embodiment In each operation mode, multiple sets of parameter values are detected from the four sensors la to Id. Here, the parameter value at the current time k is d (k).
- each parameter value detected by! / In the data creation detection step is processed, and the time differential value of each parameter value [change rate of parameter value (for example, detection cycle time, etc. Including time derivative equivalent values such as change in time)] Calculate A d (k).
- step S120 In the formation step (step S120), a plurality of sets of parameter values d (k) acquired in the data creation detection step and a time differential value A d (k) of the plurality of sets of parameter values calculated in the calculation step Using the detection data ⁇ d (k); A d (k) ⁇ based on the training data, a self-organizing map is formed as a separation model for each operation mode.
- Fig. 4 shows the parameter values of sensor la ⁇ : Ld when the excavator is repeatedly operated in a series of operation modes of 1 to 4 force, and the horizontal axis is a common time scale.
- the same parameter value waveform
- the parameter value may be different even in the same operation mode. is there. Therefore, a self-organization key map that expresses the characteristics of each operation mode more clearly is formed by repeatedly learning the self-organization key map using a large amount of reliable training data in this offline processing. It is possible.
- each self-organizing map learns for only one mode of operation, so on a two-dimensional map graph that can be expressed using commonly known self-weaving and weaving map software- It is necessary to show the topological distance (neighborhood) of Euron.
- step S200 a predetermined number of neurons are randomly arranged in an 8-dimensional space (step S200, first step). Then, each detected data point (this offline In-process is used as training data for forming a self-weaving and weaving map), and the distance to each -Euron is obtained (step S210). Then, the above distance is the minimum-Euron is determined to be the winner-Eurone. Also, at this time, this winner-the winner only by Euron-the Eurone in the vicinity of Euron-is also learned at the same time.
- the minimum distance MD is defined as the minimum value of the distance between the i-th detected data point and each -Euron in the 2n-dimensional space. For example, if the distance to the jth Euron is the minimum, the minimum jth Euron is called the winner Euron.
- This minimum distance MD is expressed by the following equation (1).
- r (i, j) represents the distance between the i th detected data point and the j th -Euron.
- the distance r (i, j) is calculated as a Eugrid distance.
- TD represents the number (set) of training data.
- step S230 it is determined whether or not the ability has been learned for all the sets. If not yet conducted (NO), the process proceeds to step S210. On the other hand, if all the operations have been completed (YES), the process moves to step S240, and one self-organizing map candidate is formed.
- the self-organization map obtained at this point is not necessarily the best self-organization map that clearly expresses one operation mode, so it is treated as one candidate.
- Steps S210 to S240 are the second step, and the self-organizing map candidate formation step includes the first step and the second step.
- one self-yarn and weave map candidate is formed for one operation mode.
- a characteristic of one operation mode is formed.
- these multiple self-organization maps The best one is selected from the candidates. Therefore, in step S250, it is determined whether or not a predetermined number of self-organizing map candidates previously formed before forming the self-organizing map is determined. If NO, the process proceeds to step S200. Then, another self-organization map candidate is formed. If YES, the process moves to step S260.
- step S260 selection step
- a candidate having a characteristic closest to the training data characteristic is selected from the candidates as a self-organizing map.
- the best self-organization map in step S260 will be described in more detail.
- FIG. 6 shows, as an example, 10 detected data points in the 2n-dimensional space (they are treated as training data in offline processing! /, So are shown as training data points in FIG. 6) d to d Visually shows the minimum distance MD between and 7 -Eurons n to n
- the average minimum distance AV is the average value of these minimum distances MD. This average
- the minimum distance AV is expressed by the following equation (2).
- the standard deviation ST is also expressed by a well-known equation [the following equation (3)], similar to the equation of the average minimum distance AV.
- step S260 shown in Fig. 5 the average minimum distance AV and the standard deviation thus obtained are calculated.
- Candidates are selected as the best self-organizing map that best matches the training data characteristics.
- step S270 the neuron that once became the winner neuron among the neurons in the selected self-thread and weave map (here, "Idling-Euron (in this case) Idling Neuron) ").
- Figure 6 shows two idlings – the forces shown by Euron ⁇ , n.
- Ring-Euron ⁇ , n is deleted after learning the self-organizing map. like this
- training data characteristics can be represented by a self-organizing map with a significantly reduced number of neurons, so the capacity to store the self-organizing map After that, the calculation processing time using this self-organizing map can be shortened.
- the merit of using one self-organizing map (separation model) for one operation mode is the huge amount of detection data that characterizes this one operation mode.
- the point is that the number can be greatly reduced-the memory capacity can be reduced by approximating the euron.
- Fig. 7 (a) to Fig. 7 (d) are graphs of detected data points for operation mode 1.
- Fig. 7 (a) shows the relationship between engine speed p and left hydraulic pump pressure P, and
- Fig. 7 (d) shows the relationship between engine speed P and fuel consumption P.
- Each self-weaving and weaving map (separation model) in Fig. 7 (a) to Fig. 7 (d) is 8D, so it is actually formed as a map where Winner-Euron is located in 8D space.
- FIGS. 8A to 8D are graphs of detected data points for the operation mode 2.
- each self-organizing map (separation model) has 8 dimensions, It is formed as a map where Winner-Ron is located in an 8-dimensional space.
- Figures 9 (a) to 9 (d) show the best self-organization map of operation mode 1.
- the small dots in Fig. 9 (a) to Fig. 9 (d) indicate the detection data points in operation mode 1, and the large dots indicate the long after complete learning and idling-ron elimination. Is shown.
- Fig. 10 (a) to Fig. 10 (d) show the best self-organizing maps in operation mode 2. Note that the small dots in Figs. 10 (a) to 10 (d) are the detected data points in operation mode 2, and the large dots are after complete learning and idling-lon elimination. Show Ron. From these Fig. 9 (a) to Fig. 9 (d) and Fig. 10 (a) to Fig. 10 (d), it is easy to see that the neurons are mainly arranged in the region with the highest data density.
- the small dots in Figs. 10 (a) to 10 (d) are the detected data points in operation mode 2, and the large dots are after complete learning and idling-lon elimination. Show Ron. From these Fig. 9 (a) to Fig. 9 (d) and Fig. 10 (a) to Fig. 10 (d), it is easy to see that the neurons are mainly arranged in the region with the highest data density.
- the abnormal data creation step (step W2) shown in FIG. 2 This process is performed in the off-line state of the hydraulic excavator as in the above-described self-organization map forming step.
- the abnormal data creation means 7 obtains data (normal data) during normal operation of the hydraulic excavator from the sensors la to Id, and sets each detected data point of the obtained normal data. Then, fluctuate at a preset fluctuation rate to create virtual data (abnormal data) when an excavator malfunction occurs.
- the detection data in the normal operation state detected in the self-organization map forming step described above may be used as it is.
- the abnormal data for can be obtained by the following equation (4).
- M Total number of data in the data group That is, as in the present embodiment, the parameter number n is the engine speed P, the fuel consumption P,
- Fig. 11 shows a conversion table (hereinafter referred to as "variable parameter vector model map") used when creating abnormal data points when the number of parameters ⁇ is 4 and the rate of change ⁇ is 5 levels. Show me! /! In Fig. 11, ⁇ is engine speed, ⁇ is fuel consumption, ⁇ is left hydraulic pump pressure, ⁇ is right hydraulic pressure
- ID 2 “625” fluctuates only the right hydraulic pump pressure P with a fluctuation rate of “+0.2”.
- the abnormal time data creation means 7 converts each detected data point of normal data into 625 sets of virtual abnormal data points using the variation parameter vector model map. .
- 1000 x 625 sets of abnormal data points are created. Will be.
- step W3 the operation mode ratio calculation step (step W3) shown in FIG. 2 will be described.
- This process is performed in the off-line state of the hydraulic excavator as in the above-described self-organization map forming step.
- the operation mode ratio calculating means 8 is first formed by the self-organization map forming means 2 for each abnormal data point of the abnormal data created by the abnormal data creating means 7. Use the self-organization map of each operation mode to recognize the operation mode. For this reason, the operation mode ratio calculation means 8 The force that most closely resembles the self-organization map of each operation mode is calculated, and the operation mode corresponding to the self-organization map with the highest similarity is obtained.
- a similarity degree SD (Similarity Degree) between one abnormal data point and the self-organizing map of each operation mode is obtained (step S 500).
- this similarity degree SD is obtained using the Eugrid distance, that is, the distance between the abnormal data point and the winner-Euron in the self-organization map.
- the similarity SD obtained in this way is divided by the average minimum distance AV to obtain an abnormality.
- This relative distance RD is calculated for each of the five self-organizing maps (step S 510).
- the relative distance RD calculated as described above is equal to or less than a predetermined value (1 + j8), that is, RD ⁇ 1 + ⁇ (j8 is a predetermined threshold) with respect to the self-organization map. It is determined whether or not there is (step S520), and if it is equal to or less than a predetermined value, it is determined that the self-organization map is satisfied (step S530). In this way, each abnormal data point is classified into an operation mode corresponding to a suitable self-organizing map.
- step S540 if the relative distance RD is equal to or greater than a predetermined value, it is determined that the self-organization map is not compatible (step S540). That is, in this case, the abnormal data point cannot be classified into any operation mode. It should be noted that by appropriately setting the predetermined value (1 + j8), it is possible to determine a criterion for determining whether or not the detected data point matches the self-organizing map according to the situation.
- the operation mode ratio calculation means 8 performs such determination for each of the five self-organization key maps for one abnormal data point, and when there are a plurality of matching self-organization key maps ( That is, when there are a plurality of suitable operation modes), the self-organization map having the smallest relative distance RD is selected and recognized as the operation mode corresponding to the self-organization map. If there is no suitable self-organization map (that is, there is no suitable operation mode), it is determined that the mode cannot be determined. Recognize it as an “intellectual driving mode”.
- the operation mode ratio calculation means 8 calculates the operation mode ratio vector (the operation mode ratio vector at the time of abnormality) representing the ratio of each operation mode to all the operation modes after performing the operation mode recognition as described above. To do.
- This operation mode ratio vector V can be obtained by the following equation (5).
- FIG. 13 shows an example of an operation mode ratio vector model map corresponding to the variation parameter vector model map shown in FIG.
- MO indicates an idling (standby) operation mode
- M1 to M4 indicate the above-described operation modes 1 to 4, respectively.
- “Fail” indicates a ratio not applicable to any of the above modes.
- the ratio of “0.185”, operation mode 1 is “0.148”, the ratio of operation mode 4 is “0.188”, and the ratio of Fail is “0.068”. If all these ratios are added, it will be “1. 000”.
- ID “1”
- 185 abnormal data points were recognized as operation mode 0, and operation mode 1 was recognized. This indicates that there are 148 abnormal data points, 188 abnormal data points recognized as operation mode 4, and 68 abnormal data points recognized as Fail.
- FIG. 14 and FIG. 15 show, as an example, the operation mode ratio of the abnormal data.
- Fig. 14 shows that only engine speed P is varied from "-0.25" to "+0.25" in increments of "0.05".
- the operation mode ratio increases in the order of the mode 1 and operation mode 0. At this time, the ratio of Fail is 0. When the engine speed P is changed, the operation mode ratio changes.
- FIG. 15 shows that only the fuel consumption P is changed in units of “0.05” from “1.0.25” to “+0.25”.
- the operation mode ratio when moving is shown. As can be seen from FIG. 15, when the fluctuation rate of the fuel consumption P is “0”, the operation mode 2, the operation mode 3, the operation mode 4, and the operation mode
- the operation mode ratio increases in the order of the mode 1 and operation mode 0. At this time, the ratio of Fail is 0, but when the fuel consumption P is changed, each operation mode ratio changes,
- Fig. 16 is a plot of the ratio of operation mode 1 and operation mode 2 when the fluctuation rate ⁇ is fluctuated in units of "0.1" from “one 0.2" to “+0.2" ( Figure Fig. 16 is a small point).
- the big points in Fig. 16 are plots of the ratios of operation mode 1 and operation mode at normal data points (that is, when the variation rate ⁇ of each parameter value is all "0"). From Fig. 16, it can be seen that the operation mode ratio at each abnormal data point is spread and distributed around the point indicating the operation mode ratio at the normal data point (large point in Fig. 16).
- step W4 the determination step shown in FIG. 2 will be described.
- the above-described processes (1) to (3) are pre-processes for diagnosing the hydraulic excavator.
- This determination step is a process for actually diagnosing the hydraulic excavator. This is performed after shipment (in this embodiment, this is referred to as the “online state” of the excavator).
- the hydraulic excavator is actually operated, and each sensor la ⁇ : Ld detects four parameter values, that is, detection data (actual operation data). Step S600).
- detection data is acquired for one day, for example, and the storage unit 3
- the operation mode ratio calculation means 8 obtains a similarity degree SD (Similarity Degree) between the detected data point and the self-yarn / weave map of each operation mode (step S610).
- a similarity degree SD Similarity Degree
- the similarity is obtained using the Eugrid distance, that is, the distance between the detected data point and the winner-Euron in the self-organization map.
- the similarity SD obtained in this way is divided by the average minimum distance AV to detect it.
- Relative distance RD SDZAV between data points and self-organizing map winner-Euron
- the winner-Euron here means the shortest distance to the detected data point (one)-Euron. This relative distance RD is calculated for each of the five self-organizing maps (step S620).
- the relative distance RD calculated as described above is not more than a predetermined value (1 + ⁇ ) with respect to the self-organization map, that is, RD ⁇ 1 + ⁇ (where ⁇ is a predetermined threshold). It is determined whether or not there is (Step S630), and if it is equal to or less than a predetermined value, it is determined that the self-organization map is suitable (Step S640). Judge that the map does not fit (step S650). In other words, the above detection data points cannot be classified into any operation mode. It should be noted that by setting the above predetermined value (1 + ⁇ ) as appropriate, it is possible to determine a criterion for determining whether or not the detected data point is compatible with the self-organizing map according to the situation.
- the operation mode ratio calculation means 8 makes such a determination for each of the five self-yarn and weaving maps, and when there are a plurality of suitable self-organizing maps (that is, there are a plurality of suitable operation modes). In the case), the self-organization map having the smallest relative distance RD is selected and recognized as the operation mode corresponding to the self-organization map. If there is no suitable self-organizing map (ie, there is no suitable operating mode), it is determined that the mode cannot be determined. Recognize as mode.
- the operation mode ratio calculation means 8 recognizes the operation mode for each detected data point as described above, and then represents the operation mode that represents the ratio of each operation mode to the entire operation mode for the day. Find the ratio vector (actual operation mode ratio vector).
- the judging means 4 is the actual operation mode ratio vector force obtained by the operation mode ratio calculating means 8 and the abnormal operation mode ratio beta value obtained in step W3 shown in FIG.
- the operation mode ratio vector that is closest is selected according to the Eugrid distance.
- the determination means 4 obtains the fluctuation parameter vector corresponding to the operation mode ratio vector selected as described above from the fluctuation parameter vector model map (see FIG. 11) obtained in step W2 shown in FIG. .
- the fluctuation rate ⁇ of each parameter of the operation mode ratio vector thus obtained which meter (engine speed, fuel consumption, left hydraulic pump pressure, Predict signs of machine deterioration and abnormalities by determining how much fluctuation is occurring in the right hydraulic pump pressure).
- the display device 6 displays the prediction result, and notifies the signs of machine deterioration or abnormality.
- the normal state of each parameter during normal operation (offline) before shipment of the hydraulic excavator is normal.
- Data is acquired, and an abnormal operation mode ratio vector model map is created using the normal data and the fluctuation parameter vector model map.
- the actual operation data of each parameter is obtained, and the actual operation mode ratio is obtained using this actual operation data.
- the vector is obtained, and the vector closest to the actual operation mode ratio vector during actual operation is selected as the abnormal operation mode ratio vector model map force created in the offline state, and the variation parameter vector corresponding to the selected vector is selected as the variation parameter vector. Obtained from the model map.
- the force object described by taking the hydraulic excavator as an example of the object that can operate in a plurality of operation modes is not limited to this.
- trucks, buses, ships, etc. It can be widely applied to the determination of the quality of operations of vehicles and various machines including industrial machines, etc., and also to the determination of the quality of life forms such as animals and plants, and estimation of changes in weather or the celestial bodies such as the earth. Applicable.
- a configuration has been described in which a diagnosis device is provided in a hydraulic excavator and diagnosis is performed collectively on the hydraulic excavator side.
- the self-organizing map forming means 2, the storage unit 3, the judging means 4, the abnormal data creating means 7, the operation mode described in the present embodiment are provided in the business office that owns this mobile machine.
- the mobile machine is in a remote location by installing a computer equipped with the ratio calculation means 8 and display device 6 and transmitting the sensor force detection data to the above computer by wireless communication etc. and displaying it
- Fig. 19 shows an example in which a management system is interposed between the mobile machine and the operator.
- diagnostic apparatus of the present invention is configured as described above, such needs can be met.
- Each parameter of an object such as a machine that can operate in a plurality of operation modes can be diagnosed more accurately, and its usefulness is considered to be extremely high.
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Automation & Control Theory (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Civil Engineering (AREA)
- Structural Engineering (AREA)
- Mining & Mineral Resources (AREA)
- Testing And Monitoring For Control Systems (AREA)
- Component Parts Of Construction Machinery (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
Description
Claims
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| EP05737347A EP1777601A4 (en) | 2004-08-13 | 2005-04-28 | DATA PROCESSING METHOD, DATA PROCESSING DEVICE, DIAGNOSTIC METHOD AND DIAGNOSTIC DEVICE |
| US11/630,571 US7743005B2 (en) | 2004-08-13 | 2005-04-28 | Diagnosis of abnormal operation modes of a machine utilizing self organizing map |
| CN200580025799A CN100595710C (zh) | 2004-08-13 | 2005-04-28 | 数据处理方法,数据处理设备,诊断方法和诊断设备 |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2004235898A JP4032045B2 (ja) | 2004-08-13 | 2004-08-13 | データ処理方法及びデータ処理装置、並びに診断方法及び診断装置 |
| JP2004-235898 | 2004-08-13 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2006016440A1 true WO2006016440A1 (ja) | 2006-02-16 |
Family
ID=35839216
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/JP2005/008172 Ceased WO2006016440A1 (ja) | 2004-08-13 | 2005-04-28 | データ処理方法及びデータ処理装置、並びに診断方法及び診断装置 |
Country Status (5)
| Country | Link |
|---|---|
| US (1) | US7743005B2 (ja) |
| EP (1) | EP1777601A4 (ja) |
| JP (1) | JP4032045B2 (ja) |
| CN (1) | CN100595710C (ja) |
| WO (1) | WO2006016440A1 (ja) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103994805A (zh) * | 2014-04-29 | 2014-08-20 | 江苏农林职业技术学院 | 一种汽车燃油油量测量的传感器测试装置 |
| CN105203346A (zh) * | 2015-10-23 | 2015-12-30 | 珠海格力电器股份有限公司 | 基于emd降噪的吸油烟机故障诊断的方法及系统 |
| WO2022024991A1 (ja) * | 2020-07-31 | 2022-02-03 | 株式会社デンソー | 異常検出装置、異常検出方法および異常検出プログラム |
Families Citing this family (46)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4886460B2 (ja) * | 2006-10-12 | 2012-02-29 | パナソニック電工Sunx株式会社 | 異常監視装置 |
| JP4886461B2 (ja) * | 2006-10-12 | 2012-02-29 | パナソニック電工Sunx株式会社 | 異常監視装置 |
| EP2085594B1 (en) * | 2008-01-29 | 2010-06-30 | Honda Motor Co., Ltd. | Control system for internal combustion engine |
| AU2010228582B2 (en) * | 2009-03-24 | 2012-09-20 | Hitachi Construction Machinery Co., Ltd. | Device for detecting abnormality in construction machine |
| JP5363927B2 (ja) * | 2009-09-07 | 2013-12-11 | 株式会社日立製作所 | 異常検知・診断方法、異常検知・診断システム、及び異常検知・診断プログラム |
| US8676721B2 (en) * | 2009-09-18 | 2014-03-18 | Apo Offshore, Inc. | Method, system and apparatus for intelligent management of oil and gas platform surface equipment |
| JP5271300B2 (ja) * | 2010-03-19 | 2013-08-21 | 株式会社小松製作所 | 建設機械の表示装置 |
| TW201216180A (en) * | 2010-10-11 | 2012-04-16 | yi-long Chen | Intelligent car maintenance and management method |
| EP2472440B1 (en) * | 2011-01-04 | 2014-06-25 | Siemens Aktiengesellschaft | Method and system for diagnosis of plant status |
| US20120215491A1 (en) * | 2011-02-21 | 2012-08-23 | Snap-On Incorporated | Diagnostic Baselining |
| BR112013022489A2 (pt) | 2011-03-03 | 2016-12-06 | Eaton Corp | sistema de controle para controlar um sistema de atuação hidráulico, método para operar um sistema de controle para um circuito hidráulico em um veículo, método para controlar um sistema de atuação hidráulico, método para configurar um controlador para um sistema hidráulico, método para reconfigurar um controlador para um sistema hidráulico e sistema de controle de atuador hidráulico |
| US9650762B2 (en) | 2012-01-24 | 2017-05-16 | Harnischfeger Technologies, Inc. | System and method for monitoring mining machine efficiency |
| GB2500388B (en) * | 2012-03-19 | 2019-07-31 | Ge Aviat Systems Ltd | System monitoring |
| US20130268527A1 (en) * | 2012-04-10 | 2013-10-10 | Baker Hughes Incorporated | Processing acquisition data into deliverable format |
| DE102012015485A1 (de) | 2012-08-07 | 2014-05-15 | Prüftechnik Dieter Busch AG | Verfahren zum Überwachen von rotierenden Maschinen |
| KR101326850B1 (ko) * | 2012-10-04 | 2013-11-11 | 기아자동차주식회사 | 오일펌프 제어 시스템 및 방법 |
| US9340950B2 (en) * | 2013-01-08 | 2016-05-17 | Deere & Company | Hydraulic fluid warm-up |
| CN103912529B (zh) * | 2013-01-08 | 2017-10-20 | 迪尔公司 | 改进的液压流体预热装置 |
| KR101554216B1 (ko) * | 2013-06-18 | 2015-09-18 | 삼성에스디에스 주식회사 | 시계열 형태의 센싱 데이터 배드 패턴 검증 방법 및 그 장치 |
| JP6116466B2 (ja) * | 2013-11-28 | 2017-04-19 | 株式会社日立製作所 | プラントの診断装置及び診断方法 |
| CN105099804B (zh) * | 2014-05-15 | 2019-01-18 | 中国石油天然气股份有限公司 | 机组故障的检测方法、服务器及终端 |
| US20150350303A1 (en) * | 2014-05-29 | 2015-12-03 | Chia-I Lin | Manufacturing optimization platform and method |
| JP2016023489A (ja) * | 2014-07-23 | 2016-02-08 | 住友重機械工業株式会社 | ショベル支援装置及びショベル管理方法 |
| US10295995B2 (en) * | 2015-01-20 | 2019-05-21 | Hitachi, Ltd. | Machine diagnostic apparatus and machine diagnostic method |
| WO2016143118A1 (ja) * | 2015-03-12 | 2016-09-15 | 株式会社日立製作所 | 機械診断装置および機械診断方法 |
| JP6796373B2 (ja) * | 2015-09-25 | 2020-12-09 | 三菱重工業株式会社 | プラント運転システム及びプラント運転方法 |
| WO2017090098A1 (ja) * | 2015-11-25 | 2017-06-01 | 株式会社日立製作所 | 設備管理装置および方法 |
| EP3232282B1 (de) | 2016-04-12 | 2018-09-12 | Siemens Aktiengesellschaft | Diagnoseeinrichtung und verfahren zur überwachung des be-triebs einer technischen anlage |
| JP6432890B2 (ja) * | 2016-06-01 | 2018-12-05 | 三菱日立パワーシステムズ株式会社 | 監視装置、対象装置の監視方法、およびプログラム |
| EP3279756B1 (de) | 2016-08-01 | 2019-07-10 | Siemens Aktiengesellschaft | Diagnoseeinrichtung und verfahren zur überwachung des betriebs einer technischen anlage |
| EP3407273A1 (de) * | 2017-05-22 | 2018-11-28 | Siemens Aktiengesellschaft | Verfahren und anordnung zur ermittlung eines anomalen zustands eines systems |
| EP3435184B1 (en) * | 2017-07-28 | 2024-04-17 | Siemens Aktiengesellschaft | System, method and control unit for controlling a technical system |
| CN109521725A (zh) | 2017-09-20 | 2019-03-26 | 西门子公司 | 检测异常数据的方法、装置和设备以及机器可读介质 |
| JP7047386B2 (ja) * | 2018-01-10 | 2022-04-05 | セイコーエプソン株式会社 | 異常を警告する方法および異常警告システム |
| JP7063022B2 (ja) * | 2018-03-14 | 2022-05-09 | オムロン株式会社 | 異常検知システム、サポート装置およびモデル生成方法 |
| US11125653B2 (en) * | 2018-10-11 | 2021-09-21 | Palo Alto Research Center Incorporated | Motion-insensitive features for condition-based maintenance of factory robots |
| JP7206985B2 (ja) * | 2019-02-08 | 2023-01-18 | コベルコ建機株式会社 | 損害推定装置及び機械学習装置 |
| US12176227B2 (en) * | 2019-03-26 | 2024-12-24 | Tokyo Electron Limited | State determination device, state determination method, and computer-readable recording medium |
| US11560908B2 (en) * | 2019-05-13 | 2023-01-24 | Caterpillar Inc. | Control mapping for hydraulic machines |
| US11623767B2 (en) * | 2019-07-24 | 2023-04-11 | The Boeing Company | Relational time-series classification method and system for corrosion maintenance dispatch |
| CN110374164A (zh) * | 2019-07-25 | 2019-10-25 | 徐州徐工矿业机械有限公司 | 一种液压挖掘机动力系统故障监测与诊断系统及方法 |
| US11587448B2 (en) * | 2019-07-26 | 2023-02-21 | General Electric Company | Systems and methods for manifolds learning of airline network data |
| EP3901595B1 (en) * | 2020-04-22 | 2025-06-04 | ABB Schweiz AG | A fault state detection apparatus |
| CN112580262B (zh) * | 2020-12-23 | 2023-05-02 | 武汉理工大学 | 一种液压机能耗智能监测方法、系统、装置及存储介质 |
| DE102021124253A1 (de) | 2021-09-20 | 2023-03-23 | Festo Se & Co. Kg | Maschinelles Lernverfahren zur Anomaliedetektion in einem elektrischen System |
| DE102021124254A1 (de) | 2021-09-20 | 2023-03-23 | Festo Se & Co. Kg | Maschinelles Lernverfahren zur Leckageerkennung in einem pneumatischen System |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH11338848A (ja) * | 1998-05-26 | 1999-12-10 | Ffc:Kk | データ異常検出装置 |
| JP2000266570A (ja) * | 1999-03-15 | 2000-09-29 | Omron Corp | 定常状態と非定常状態を弁別する信号処理装置 |
| JP2005025351A (ja) * | 2003-06-30 | 2005-01-27 | Kagawa Univ | 情報処理装置,状態判定装置及び診断装置,並びに情報処理方法,状態判定方法及び診断方法 |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5239594A (en) * | 1991-02-12 | 1993-08-24 | Mitsubishi Denki Kabushiki Kaisha | Self-organizing pattern classification neural network system |
| DE19734947B4 (de) * | 1997-08-13 | 2010-04-29 | Otte, Ralf, Dr. | Verfahren zur Steuerung von Prozeßvorgängen |
| US6647341B1 (en) * | 1999-04-09 | 2003-11-11 | Whitehead Institute For Biomedical Research | Methods for classifying samples and ascertaining previously unknown classes |
| DE19957163C1 (de) * | 1999-11-27 | 2001-08-09 | Thyssenkrupp Stahl Ag | Verfahren und Vorrichtung zur Qualitätskontrolle der Naht an mit einem Laser stumpf geschweißten Blechen oder Bändern |
| AU2001247627A1 (en) * | 2000-03-22 | 2001-10-03 | 3-Dimensional Pharmaceuticals, Inc. | System, method, and computer program product for representing object relationships in a multidimensional space |
| US6571228B1 (en) * | 2000-08-09 | 2003-05-27 | Po-Tong Wang | Hybrid neural networks for color identification |
| US6477469B2 (en) * | 2001-01-08 | 2002-11-05 | Halliburton Energy Services, Inc. | Coarse-to-fine self-organizing map for automatic electrofacies ordering |
| JP2002323013A (ja) | 2001-04-24 | 2002-11-08 | Komatsu Ltd | 作業機械の異常診断装置 |
| AT412678B (de) * | 2002-09-30 | 2005-05-25 | Gerhard Dr Kranner | Verfahren zur rechnergestützten erstellung von prognosen für operative systeme sowie system zur erstellung von prognosen für operative systeme |
| US7552005B2 (en) * | 2004-03-16 | 2009-06-23 | Honeywell International Inc. | Method for fault diagnosis of a turbine engine |
-
2004
- 2004-08-13 JP JP2004235898A patent/JP4032045B2/ja not_active Expired - Fee Related
-
2005
- 2005-04-28 CN CN200580025799A patent/CN100595710C/zh not_active Expired - Fee Related
- 2005-04-28 US US11/630,571 patent/US7743005B2/en not_active Expired - Fee Related
- 2005-04-28 EP EP05737347A patent/EP1777601A4/en not_active Withdrawn
- 2005-04-28 WO PCT/JP2005/008172 patent/WO2006016440A1/ja not_active Ceased
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH11338848A (ja) * | 1998-05-26 | 1999-12-10 | Ffc:Kk | データ異常検出装置 |
| JP2000266570A (ja) * | 1999-03-15 | 2000-09-29 | Omron Corp | 定常状態と非定常状態を弁別する信号処理装置 |
| JP2005025351A (ja) * | 2003-06-30 | 2005-01-27 | Kagawa Univ | 情報処理装置,状態判定装置及び診断装置,並びに情報処理方法,状態判定方法及び診断方法 |
Non-Patent Citations (1)
| Title |
|---|
| See also references of EP1777601A4 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103994805A (zh) * | 2014-04-29 | 2014-08-20 | 江苏农林职业技术学院 | 一种汽车燃油油量测量的传感器测试装置 |
| CN105203346A (zh) * | 2015-10-23 | 2015-12-30 | 珠海格力电器股份有限公司 | 基于emd降噪的吸油烟机故障诊断的方法及系统 |
| WO2022024991A1 (ja) * | 2020-07-31 | 2022-02-03 | 株式会社デンソー | 異常検出装置、異常検出方法および異常検出プログラム |
Also Published As
| Publication number | Publication date |
|---|---|
| US20070244841A1 (en) | 2007-10-18 |
| CN101027617A (zh) | 2007-08-29 |
| JP2006053818A (ja) | 2006-02-23 |
| CN100595710C (zh) | 2010-03-24 |
| US7743005B2 (en) | 2010-06-22 |
| JP4032045B2 (ja) | 2008-01-16 |
| EP1777601A1 (en) | 2007-04-25 |
| EP1777601A4 (en) | 2010-11-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2006016440A1 (ja) | データ処理方法及びデータ処理装置、並びに診断方法及び診断装置 | |
| JP4133627B2 (ja) | 建設機械の状態判定装置及び建設機械の診断装置,並びに建設機械の状態判定方法及び建設機械の診断方法 | |
| CN102707708B (zh) | 多模型飞行控制系统故障诊断方法及装置 | |
| CN110738360B (zh) | 一种设备剩余寿命预测方法及系统 | |
| US10664754B2 (en) | Information processing apparatus | |
| CN102789545B (zh) | 基于退化模型匹配的涡轮发动机剩余寿命的预测方法 | |
| CN104507792A (zh) | 用于船只性能的评估的方法和系统 | |
| JP4175296B2 (ja) | 建設機械のデータ処理装置及び建設機械のデータ処理方法 | |
| US8855954B1 (en) | System and method for prognosticating capacity life and cycle life of a battery asset | |
| CN114462305A (zh) | 一种Argo浮标长航时能耗优化与寿命预估方法 | |
| CN113139514A (zh) | 一种基于模糊神经网络的液压系统故障诊断方法 | |
| JP2007257366A (ja) | 診断装置及び診断方法 | |
| Chen et al. | Data-driven propulsion load profile prediction for all-electric ships | |
| Mosallam et al. | Integrated bayesian framework for remaining useful life prediction | |
| Peysson et al. | A generic prognostic methodology using damage trajectory models | |
| Salah et al. | Echo state network and particle swarm optimization for prognostics of a complex system | |
| CN117444978B (zh) | 一种气动软体机器人的位置控制方法及其系统、设备 | |
| CN117408677A (zh) | 一种事故车辆的维修优化方法及系统 | |
| KR20240074365A (ko) | 시운전 선박의 연료 소모량 예측 방법, 시스템 및 비일시성의 컴퓨터 판독 가능 기록 매체 | |
| Shen et al. | Hybrid Machine Learning-Based Methodologies for Fault Diagnosis, Prognosis and Health Management | |
| EP4280122A1 (en) | Actionable advice for the optimization of process systems | |
| Assafo et al. | Self-Organizing Map Applications for Predictive Maintenance: A Review | |
| Sun et al. | A hybrid data and mechanism-driven method for operational state estimation in multi-domain coupled industrial equipment | |
| Xu et al. | Research on Real-Time Vehicle Load Prediction Method Based on Neural Networks | |
| Garvey et al. | An integrated fuzzy inference-based monitoring, diagnostic, and prognostic system for intelligent control and maintenance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| AK | Designated states |
Kind code of ref document: A1 Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS KE KG KM KP KR KZ LC LK LR LS LT LU LV MA MD MG MK MN MW MX MZ NA NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW |
|
| AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG |
|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
| WWE | Wipo information: entry into national phase |
Ref document number: 2005737347 Country of ref document: EP |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 11630571 Country of ref document: US |
|
| WWE | Wipo information: entry into national phase |
Ref document number: 200580025799.9 Country of ref document: CN |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| WWP | Wipo information: published in national office |
Ref document number: 2005737347 Country of ref document: EP |
|
| WWP | Wipo information: published in national office |
Ref document number: 11630571 Country of ref document: US |