JP2018151821A - Abnormality diagnosis system of facility apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enable the achievement of abnormality diagnosis about data of a facility apparatus even in the case that normal data cannot be defined because the normal data to be used for leaning processing is not clear when data clustering processing is used to perform abnormality diagnosis of the facility apparatus.SOLUTION: An abnormality diagnosis system 1 of a facility apparatus for performing abnormality diagnosis by using a diagnosis method composed of learning processing with normal data and diagnosis processing with a model created by the leaning processing on the basis of a plurality of pieces of data from a facility apparatus 2 and used in the case that data during normal operation is unknown about the plurality of pieces of data from the facility apparatus includes: a first processing part 13 for performing clustering processing of the plurality of pieces of data from the facility apparatus; a second processing part 14 for defining a normal cluster on the basis of the number of pieces of data corresponding to each cluster; a third processing part 16 for extracting data corresponding to the normal cluster as normal cluster; and a fourth processing part 17 for performing learning processing by using the extracted normal data.SELECTED DRAWING: Figure 10

Description

  The present invention relates to an abnormality diagnosis system for equipment using data clustering processing.

  In various plants, detecting an abnormal state of the plant at an early stage before an alarm is output is effective in improving operation reliability. As a method for detecting an abnormality of a plant at an early stage, it is effective to learn a data pattern when the plant is normal in advance and detect a state change by comparing this with a pattern of measurement data. One such data analysis is a data clustering technique.

  Clustering analysis is a method of classifying input data into a plurality of clusters (= groups) based on a data pattern. A cluster represents a collection of data having similarity to data patterns. In the clustering analysis, a plurality of input data is mapped on a multidimensional space, and a cluster is defined from the positional relationship in the space. Abnormality determination is performed on diagnostic data with reference to a cluster defined by normal data. An outline of this processing is shown in FIG.

  FIG. 2 shows a two-dimensional surface that employs the input signal from the sensor 1 on the horizontal axis and the input signal from the sensor 1 on the vertical axis, and a case where two input signals composed of the sensor 1 and the sensor 2 are clustered. Show. In this example, clustering is performed on a two-dimensional surface, and the measured values of the sensors 1 and 2 are plotted as the values of the respective axes, and cluster 1, cluster 2, and cluster 3 are obtained as a result of clustering. In each cluster, a cross indicates the position of the center of gravity of the category. In FIG. 2, clustering on a two-dimensional surface is shown as a simple example. However, since a large number of sensor signals measured by equipment are input, it is generally processed in a multidimensional space. It is.

  In FIG. 2, both the white circle learning data and the black circle diagnosis data are plotted, and three clusters created by the learning data are shown. Here, the learning data is data defined as normal by the user. When performing abnormality diagnosis of equipment by clustering, a cluster is created in advance using learning data, and this is used as a normal state standard. Next, the diagnostic data is plotted in the same space, and abnormality determination is performed from the positional relationship with the cluster created by the learning data. As exemplified in cluster 1 and cluster 3, if the diagnostic data is similar to the normal data pattern, the plotted points are included in any of the three clusters. On the other hand, as exemplified in cluster 2, if the diagnostic data is significantly different from the normal data pattern, it is plotted at a position outside the cluster.

  As one method of abnormality determination, a concept of abnormality level indicating the degree of deviation from normal can be considered. Although there are several possible definitions of the degree of abnormality, as shown in the figure, one method is to use the distance from the center of the nearest cluster for each diagnostic data. For example, in cluster 2, the distance between the barycentric position of the category indicated by x and the diagnostic data is determined, and the degree is determined based on the distance from the cluster area. The distance increases as the diagnostic data greatly differs from the data pattern included in the learning data. On the other hand, even when the diagnostic data is included in a normal cluster, the degree of abnormality does not become zero. This is because the data includes fluctuation, which corresponds to the distance from the cluster centroid. Normal abnormality determination is performed by performing threshold determination for the degree of abnormality.

  As an example in which clustering is utilized for abnormality determination of equipment, Patent Document 1 is cited. Patent Document 1 describes a diagnosis method for time-series waveform data, but the basic concept of abnormality determination is the same as described above. As shown in FIG. 6 of Patent Document 1, abnormality is determined based on the distance between the diagnostic data and the cluster center. In Patent Document 1, as shown in Formula 1, a value obtained by dividing a distance in space by the radius of a cluster is used as the degree of abnormality.

JP 2017-33471 A

  As described above, methods and systems for diagnosing abnormality of equipment using various clustering methods have been proposed. However, these methods require learning processing using data defined as normal in advance. In other words, the cluster generated from the learning data is used as a standard for the data pattern at the normal time, and the abnormality determination is performed in comparison with this.

  As an example in which this method cannot be applied, there is sensor data measured by equipment, but there is a case where it is unknown which data is in normal operation and which data is abnormal. That is, there is a possibility that data at the time of abnormality may be included in the data. If the clustering process is performed using learning data including abnormal data, the abnormal data pattern is also treated as normal. For this reason, even if a similar abnormality occurs, the system cannot detect the abnormality.

  As described above, there has been a demand for a processing method capable of diagnosing abnormality even for data whose normality / abnormality is unknown.

From the above, in the present invention, “based on a plurality of data from equipment, an abnormality diagnosis is performed using an analysis method comprising a learning process with normal data and a diagnosis process with a model created by the learning process. In addition to performing a plurality of data from the equipment, the equipment equipment abnormality diagnosis system when the data at the time of normal operation of the equipment is not known,
A first processing unit that performs a clustering process on a plurality of data from the equipment;
A second processing unit that defines a normal cluster based on the number of data corresponding to each cluster;
A third processing unit for extracting data corresponding to the normal cluster as normal data;
An abnormality diagnosis system for facility equipment, comprising a fourth processing unit that performs learning processing using extracted normal data. ".

  According to the abnormality diagnosis system for equipment according to the present invention, abnormality diagnosis can be realized even when the period during which the equipment is operating normally is unknown and normal data used for learning cannot be defined.

The figure which shows the structural example of the abnormality diagnosis system of the equipment apparatus which concerns on Example 1. FIG. The figure which shows the concept of the abnormality diagnosis by clustering. The figure which shows the structural example of sensor database DB1. The figure which shows the structural example of normalization database DB2. The figure which shows the structural example of clustering result database DB3A. The figure which shows the structural example of clustering result database DB3B. The figure which shows the structural example of cluster information database DB4. The figure which shows the structural example of abnormality degree calculation result database DB5. The figure which shows the example of a display screen of the abnormality diagnosis system of an installation apparatus. The figure which shows the display screen 90c displayed on the input / output device 3 of the abnormality diagnosis system of the equipment apparatus which concerns on Example 2. FIG. The figure which shows the structural example of the abnormality diagnosis system of the equipment equipment which concerns on Example 3. FIG. The figure which shows the structural example of normal data database DB6. The figure which shows the structural example of model analysis result database DB7. The figure which shows the display screen 90d displayed on the input / output device 3 of the abnormality diagnosis system of the equipment apparatus which concerns on Example 3. FIG.

  A configuration of an abnormality diagnosis system for equipment according to the present invention will be described below with reference to the drawings.

  The structural example of the abnormality diagnosis system of the equipment apparatus which concerns on Example 1 of this invention is shown in FIG. In FIG. 1, 1 is an equipment equipment abnormality diagnosis system, 2 is equipment equipment to be subjected to abnormality diagnosis, 3 is a diagnostic information output by the equipment equipment abnormality diagnosis system 1, and is input by the user. An input / output device for registering data.

  The equipment apparatus abnormality diagnosis system 1 includes a plurality of processing units (11 to 15), a storage unit 20 including a plurality of databases (DB1 to DB5), and an input / output device control unit 31. Among these, the memory | storage part 20 is comprised from several database DB for every data type, The writing and reading of the data by each process part (11 to 15) which comprise the abnormality diagnosis system 1 with respect to the memory | storage part 20 are possible. It has a possible configuration.

  Among the plurality of processing units (11 to 15), the data capturing unit 11 stores and stores sensor data captured from the equipment 2 in the sensor database DB 1 in the storage unit 20.

  FIG. 3 is a diagram illustrating a configuration of the sensor database DB1. As sensor data, data from a plurality of sensors (here, the data of sensor 1, sensor 2, etc.) is written and stored together with date and time information in time series.

  Next, the normalization processing unit 12 constituting the abnormality diagnosis system 1 reads data from the sensor database DB1 and performs normalization processing. The standardized data is stored in the standardized result database DB2. FIG. 4 shows the configuration of the standardization result database DB2. All data is standardized as a numerical value between 0 and 1, for example, by the standardization process. Thus, the standardization result database DB2 stores the standardized value for each sensor together with the date / time information in the same format as the sensor database DB1 described above.

  As a standardization method, an upper limit / lower limit is defined based on the maximum / minimum of the design value or sensor data, and the values are converted to 1.0 / 0.0, respectively, or There are multiple methods such as dividing by the standard deviation of sensor data. The purpose of the normalization process is to treat data having different scales depending on the sensor equivalently in the clustering process. For example, even if sensor data that changes in the order of several hundreds and sensor data that changes in digits after the decimal point are subjected to clustering processing with raw values, a meaningful result cannot be obtained. In this case, the degree of abnormality defined by the distance in space greatly depends on the change width of sensor data having an order of several hundreds. In order to avoid such a situation, it is common to perform clustering processing using a standardized value for sensor data as an input.

  Next, the clustering processing unit 13 takes in the normalized sensor data from the standardization result database DB2 as time series data, and performs clustering processing using this as input. A plurality of methods of clustering are known, but any method can be applied to the system according to the embodiment of the present invention regardless of the clustering method. The result of the clustering process in the clustering processing unit 13 is stored in the clustering result database DB3.

  The clustering result database DB3 includes a database DB3A for the cluster identification number shown in FIG. 5A and a database DB3B for the cluster information shown in FIG. 5B.

  Among these, the cluster identification numbers in the database DB3A are stored in chronological order with the cluster identification numbers in a format corresponding to the date and time information stored in the standardization result database DB2. If this number is the same, it represents that the data pattern which consists of several sensor data is similar. The cluster information in the database DB3B stores the barycentric coordinates of each cluster. For each cluster number, a coordinate value represented by each input signal (each sensor in the first embodiment) is stored. This data is used in the calculation of the degree of abnormality described later.

  Next, the cluster information processing unit 14 takes in the clustering number determination result shown in FIG. 5A from the clustering result database DB3 and counts the number of data for each cluster number. This result is stored in the cluster information database DB4.

  FIG. 6 shows the configuration of the cluster information database DB4. The number of data for each cluster number is described. Further, the cluster information processing unit 14 selects a normal cluster based on the number of data in addition to counting the number of data for each cluster. Here, a threshold for the number of data defined as normal is provided in advance, and a cluster exceeding this threshold is defined as a normal cluster. Here, the threshold value may be the number of data or a ratio to the total number. If the threshold value is set to 10, as shown in the cluster information database DB4 in FIG. 6, the cluster numbers 1 and 3 with the number of data exceeding 10 are defined as “normal”. At this time, “True” is set in the column indicating the normal cluster in the cluster information database DB4. Otherwise, “False” is set.

  In the abnormality diagnosis system for facility equipment according to the first embodiment, a cluster having a large number of corresponding data is defined as “normal”. This is based on the premise that the equipment to be diagnosed is normally operating normally, and the abnormality is exceptionally occurring for a short period of time. That is, most of the sensor data stored in the sensor database DB1 is normal data, and a very small part of the sensor data contains abnormal data. In such a case, when the data pattern is classified by the clustering process, it is estimated that a cluster including a large number of data is “normal”. Using the normal cluster defined here as a reference, an abnormality determination process is performed based on the positional relationship between the normal cluster and each data in the multidimensional space.

  Next, the abnormality degree calculation unit 15 calculates the abnormality degree of each data. In the abnormality diagnosis system for facility equipment according to the first embodiment, the cluster information processing unit 14 defines a normal cluster. Therefore, a known method as shown in FIG. 2 can be used as a method for calculating the degree of abnormality. First, the degree-of-abnormality calculation unit 15 acquires information on the cluster number defined as normal from the cluster information database DB4 illustrated in FIG. Next, the cluster centroid coordinate data corresponding to the normal cluster is acquired from the clustering result database DB3B shown in FIG. 5B. Next, the normalized value of the sensor data for each date and time is fetched from the normalized result database DB2 shown in FIG. 4, and the degree of abnormality is calculated. The degree of abnormality is calculated as the distance from the center of gravity of the nearest normal cluster selected in the multidimensional space. The abnormality degree calculation result is stored in the abnormality degree calculation result database DB5.

  FIG. 7 shows the configuration of the abnormality calculation result database DB5. In the abnormality calculation result database DB5, the calculated value of the degree of abnormality for each date and time is stored as time series data.

  FIG. 8 shows a display example of the display unit 90 in the input / output device 3. Two types of display screens are set on the display unit 90, and different types of contents are displayed. On the display screen 90a, a time-series trend of data from the sensors (sensor 1, sensor 2, sensor 3) of the equipment 1 that is an input of the clustering process is displayed. The display screen 90b shows a result of abnormality diagnosis corresponding to the sensor data shown on the display screen 90a. Data used for displaying these graphs is stored in the storage unit 20. The user requests graph display of specific data through the input / output device 3, and the input / output device control unit 31 extracts the requested data and outputs it to the input / output device 3.

  In the display screen 90b, the cluster number identified in the upper part and the abnormality degree in the lower part are shown as a trend. In both graphs, the data corresponding to the cluster number defined as normal by the cluster information processing unit 14 is defined as a normal period, and is displayed separately by color coding. As shown in the graph of the degree of abnormality in the figure, the data corresponding to the normal cluster tends to have a lower value of the degree of abnormality, but does not necessarily become zero. This is because, as described above, fluctuations are included in the sensor data, and a distance from the center of gravity of each cluster occurs in a multidimensional space in clustering. On the other hand, data with a high degree of abnormality means that the data pattern is greatly different from that defined as a normal cluster, that is, the possibility of abnormality is high. In this way, it is possible to detect an abnormality included in the sensor data from the abnormality degree graph.

  According to the first embodiment, even when the learning data for learning the normal characteristics cannot be defined in the diagnosis target data, the normal period is automatically defined by the clustering process, and the abnormality diagnosis is based on the normal period. Can be performed.

  The configuration of the abnormality diagnosis system for equipment according to the second embodiment of the present invention is basically the same as the system shown in the first embodiment. The differences from the first embodiment are the processing content of the cluster information processing unit 14, the writing processing to the cluster information database DB 4, and the processing content of the input / output device control unit 31 through the input / output device 3. Only the differences will be described here.

  In the first embodiment, the cluster information processing unit 14 automatically performs processing related to the definition of a normal cluster. That is, a cluster in which the number of corresponding data exceeds a preset threshold is selected from each cluster, defined as a “normal” cluster, and the data is written in the cluster information database DB4 shown in FIG.

  On the other hand, in the equipment apparatus abnormality diagnosis system according to the second embodiment, the user visually confirms the clustering result and defines a “normal” cluster based on the information.

  FIG. 9 illustrates a display screen 90c displayed on the input / output device 3 of the abnormality diagnosis system for facility equipment according to the second embodiment. The graph on the display screen 90c shows the number of corresponding data for each cluster stored in the cluster information database DB4 shown in FIG. Based on this graph, the user defines the number of data that is a criterion for determining a normal cluster. Reference numeral 300 in FIG. 9 denotes a bar for adjusting the reference value. In the example of the figure, the number of data of 10 or more is a “normal” cluster. The input / output device control unit 31 takes in the reference value set by the user and writes the data of the column “normal cluster” in the cluster information database DB4 shown in FIG. In other words, a cluster having 10 or more data is used as a reference for “normal” in abnormality diagnosis, so this data is “True” and the others are “False”.

  The subsequent processing is the same as in the first embodiment. The degree of abnormality calculation unit 15 calculates the degree of abnormality of each data with reference to the “normal cluster” set by the user.

  According to the second embodiment, even when the learning data for learning the characteristics at normal time cannot be defined in the data to be diagnosed, the user visually confirms the clustering result, and based on the information, the “normal cluster” By defining “”, it becomes possible to perform abnormality diagnosis.

  FIG. 10 shows a configuration example of an abnormality diagnosis system for facility equipment according to Embodiment 3 of the present invention. The difference from the abnormality diagnosis system for facility equipment according to the first embodiment shown in FIG. 1 is that a normal data extraction unit 16, a normal data database DB6, a model analysis unit 17, and a model analysis result database DB7 are added. . Here, only differences from the first embodiment will be described.

  The normal data extraction unit 16 refers to the time series data of the degree of abnormality stored in the degree of abnormality calculation result database DB5 shown in FIG. Here, the normal data extraction unit 16 compares the value of the degree of abnormality with a preset threshold value, acquires date and time data that is equal to or less than the threshold value, and outputs it to the normal data database DB6. FIG. 11 shows a configuration example of the normal data database DB6. Date and time data that satisfies the condition that the degree of abnormality is equal to or less than the threshold is stored.

  Next, the model analysis unit 17 refers to the date / time data stored in the normal data database DB6, and further reads the standardized data corresponding to this date / time from the standardization result database DB2 shown in FIG. At this time, the read standardized data is only data whose degree of abnormality is equal to or less than a threshold, that is, data estimated to be normal. The model analysis unit 17 performs learning processing using the read data as learning data. Next, the model analysis unit 17 reads data corresponding to all the dates and times stored in the standardization result database DB2, and performs a diagnosis process.

  The analysis method in the model analysis unit 17 is not particularly limited as long as it is an analysis method including two steps of learning and diagnosis. It is also possible to use different clustering methods for the clustering processing unit 13 and the model analysis unit 17. Furthermore, the analysis method that can be used in the model analysis unit 17 is applicable to any analysis method that constructs a normal model using learning data, such as a neural network model or statistical model other than clustering analysis.

  The analysis result in the model analysis unit 17 is stored in the model analysis result database DB7. FIG. 12 shows a configuration example of the model analysis result database DB7. Data having a high degree is stored together with time information indicating the state.

  In the above-described processing, the normal data extraction processing in the normal data extraction unit 16 is performed based on a preset threshold value. It is also possible to set a threshold after the user visually confirms this. FIG. 13 shows an example of the display screen 90d of the input / output device 3 when the user sets a threshold value used in normal data extraction processing. On the display screen 90d, number 1 is selected from the clusters defined as normal in the cluster information database DB4 shown in FIG. The graph on the display screen 90d represents the distribution of the number of data with respect to the degree of abnormality for the data corresponding to cluster 1. As described above, the degree of abnormality does not become 0 even for data defined as normal. This is because the distance from the cluster centroid is generated due to fluctuations included in the data. The user adjusts the threshold value using the bar 310 while confirming the distribution of the degree of abnormality using the display screen 90d. In this example, the threshold value is 0.05. That is, the normal data extraction unit 16 extracts, as normal data, data whose abnormality level is less than 0.05 for data corresponding to the cluster 1.

  According to the third embodiment, the normal period can be automatically extracted by the clustering process even when the learning data for learning the normal characteristics cannot be defined in the diagnosis target data. It is possible to perform a model learning process using the extracted normal data, and then a diagnostic process using all the data. The model analysis method according to the third embodiment is not limited to the clustering method, and any method including two steps of learning and diagnosis processing can be applied.

  Further, according to the third embodiment, the user can set the threshold for the degree of abnormality that is a reference for normal data extraction while visually confirming the distribution of the degree of abnormality for each cluster. Thereby, normal data extraction conditions can be set in more detail, and the accuracy of abnormality diagnosis by model analysis can be improved.

  The system according to the present invention can be used for abnormality diagnosis of general mechanical equipment.

1: equipment equipment abnormality diagnosis system, 2: equipment equipment to be diagnosed, 3: input / output device, 11: data fetching section, 12: normalization processing section, 13: clustering processing section, 14 cluster information processing section 15: Abnormality calculation unit, 16: Normal data extraction unit, 17: Model analysis unit, 20: Storage unit, DB1: Sensor database, DB2: Normalization result database, DB3: Clustering result database, DB4: Cluster information database, DB5: Abnormality calculation result database, DB6: normal data database, DB7: model analysis result database, 31: input / output device controller

Claims (10)

Based on a plurality of data from the equipment, an abnormality diagnosis is performed using an analysis method including a learning process with normal data and a diagnosis process with a model created by the learning process. About the data of the equipment equipment abnormality diagnosis system when the data at the time of normal operation of the equipment equipment is not known,
A first processing unit that performs a clustering process on a plurality of data from the equipment;
A second processing unit that defines a normal cluster based on the number of data corresponding to each cluster;
A third processing unit for extracting data corresponding to the normal cluster as normal data;
An abnormality diagnosis system for facility equipment, comprising: a fourth processing unit that performs learning processing using the extracted normal data and evaluates a plurality of data from the facility equipment. An abnormality diagnosis system for equipment according to claim 1,
The second processing unit sets a threshold value for the number of data corresponding to each cluster, and defines a cluster exceeding the threshold value as a normal cluster. The abnormality diagnosis system for facility equipment according to claim 1 or 2, comprising a display unit,
An abnormality diagnosis system for facility equipment, wherein the number of data corresponding to each cluster is displayed on the screen of the display unit. The abnormality diagnosis system for equipment according to claim 3, having a first threshold setting function,
On the screen of the display unit, the number of data corresponding to each cluster and a threshold for the number of data set by the first threshold setting function are displayed, and the user uses the first threshold setting function to display the data An abnormality diagnosis system for facility equipment, characterized by setting a threshold value for the number. It is an abnormality diagnosis system for facility equipment according to any one of claims 1 to 4,
A facility device in which a threshold is set for the degree of abnormality of each data obtained from clustering processing, and data corresponding to the normal cluster and having the degree of abnormality equal to or less than the threshold is extracted as normal data Abnormality diagnosis system. The abnormality diagnosis system for equipment according to claim 5, comprising a display unit,
An abnormality diagnosis system for facility equipment, wherein a distribution of the degree of abnormality of data corresponding to a normal cluster is displayed on the screen of the display unit. The abnormality diagnosis system for equipment according to claim 6, having a second threshold setting function,
On the screen of the display unit, the distribution of the degree of abnormality of the data corresponding to the normal cluster and the threshold for the degree of abnormality set by the second threshold setting function are displayed, and the user uses the second threshold setting function. And setting a threshold for the degree of abnormality. The abnormality diagnosis system for facility equipment according to any one of claims 1 to 7, comprising a display unit,
An abnormality diagnosis system for facility equipment, characterized in that a cluster number or degree of abnormality as a result of clustering processing in the first processing unit is displayed on the display unit. An abnormality diagnosis system for facility equipment according to claim 8,
An abnormality diagnosis system for facility equipment, characterized in that, when a cluster number or abnormality level is displayed on the display unit, the cluster number or abnormality level corresponding to normal data is displayed separately. An abnormality diagnosis system for facility equipment according to any one of claims 1 to 9,
The third processing unit calculates an abnormality level based on the extracted normal cluster, extracts only normal data based on the calculated abnormality level, and the fourth processing unit extracts the extracted normal data. An abnormality diagnosis system for equipment that performs analysis for abnormality diagnosis using as learning data for models.

Images