JP2018081523A - Equipment diagnostic device and equipment diagnostic method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve accuracy of abnormality detection and cause diagnosis of equipment.SOLUTION: The equipment diagnostic device diagnosing abnormality of the equipment from values measured by sensors comprises elements constituting the equipment, equipment element information setting relations among the sensors installed in the elements, time sequence sensor data including identifiers and values of the sensors, input sensor information setting an input sensor of the sensors within the elements, an abnormality detection part identifying the sensor detecting the abnormality from the value of the sensor data as an abnormality cause sensor, and identifying the element including the abnormality cause sensor by referring to the equipment element information, and an abnormality filter part determining whether the abnormality cause sensor is the input sensor of the identified element by referring to the input sensor information, and if the abnormality cause sensor is not the input sensor, classifying the element as a first abnormal element, and if the abnormality cause sensor is the input sensor, classifying the element as a second abnormal element.SELECTED DRAWING: Figure 12

Description

  The present invention relates to a diagnostic apparatus for detecting an abnormality of equipment.

  In factories, power plants, plants, etc., the operating time of equipment is directly linked to changes in earnings. For this reason, when an abnormality occurs in the facility, detecting the abnormality and narrowing down the cause of the abnormality with high accuracy at an early stage shortens the time until the problem is solved and leads to an increase in profit.

  In order to improve the accuracy of equipment diagnosis, as described in Patent Document 1, there is known a technique of dividing an apparatus and piping constituting a plant and individually detecting an abnormality. Furthermore, it has design information for storing the association between devices and pipes, and narrows down the cause of the abnormality by tracing the device upstream of the device where the abnormality has occurred. By detecting an abnormality for each device, it is possible to detect an abnormality with high accuracy. Furthermore, the cause information can be narrowed down by using the design information to diagnose the cause of the abnormality.

  Further, as described in Patent Document 2, a technique is known in which device data is classified into input data and output data, and device abnormality diagnosis is performed based on the responsiveness of the output data to the input data. This abnormality diagnosis method that introduces a causal relationship between input data and output data can realize abnormality diagnosis with higher accuracy than the equipment diagnosis method that evaluates the correlation between sensors.

International Publication No. 2011/061793 JP 2012-128583 A

  The facility diagnosis method described in Patent Document 1 requires design information for recording the upstream and downstream associations between devices and pipes. However, depending on the facility, it may be difficult to identify the upstream and downstream relationships between devices. For example, in the vibration analysis of the generator shaft, it is difficult to specify from the design drawing which of the plurality of vibration sensors is upstream. For this reason, it is difficult to apply the abnormality cause diagnosis method of Patent Document 1 to various facilities.

  The facility diagnosis method described in Patent Literature 2 analyzes time series data and classifies the data into input data and output data. However, there is a case where the input data and the output data are difficult to apply because the output is a time delay in a similar part between time series. For example, it is difficult to detect a time delay of several seconds from time-series data with 10-minute intervals. Therefore, it is difficult to apply the abnormality cause diagnosis method of Patent Document 1 using various time series data.

  Therefore, an object of the present invention is to extract relationship information between sensors from the equipment operation log in order to improve the accuracy of equipment abnormality detection and cause diagnosis. By using this relationship information, the equipment diagnosis accuracy can be improved.

  A facility diagnostic apparatus having a processor and a memory and diagnosing facility abnormality from a value measured by a sensor, wherein a relationship between one or more elements constituting the facility and a sensor installed in the element is set Facility element information, time-series sensor data including the sensor identifier and the value, input sensor information in which an input sensor is set among the sensors in the element, and the value of the sensor data using the sensor data as an input An abnormality detecting unit that identifies an abnormality detecting sensor as an abnormality cause sensor, identifies the element including the abnormality cause sensor with reference to the facility element information, and the abnormality cause sensor with reference to the input sensor information. Is the input sensor of the specified element, and if the abnormality cause sensor is not the input sensor, the element is classified as a first abnormal element Have, and abnormal filter unit which classifies the elements in the second abnormality elements when the abnormality cause sensor is the input sensor.

  Therefore, according to the present invention, even when an abnormality is detected with a plurality of elements at the time of facility diagnosis, the elements causing the abnormality can be narrowed down, and the diagnosis accuracy can be improved. The present invention can also be applied to equipment in which it is difficult to specify the input or output of an element from a design drawing. Further, the present invention can be applied to facilities in which the data interval of time series data is long and a time delay between inputs or outputs cannot be detected. Thereby, a highly accurate equipment diagnosis can be provided for a wide range of equipment.

1 is a block diagram illustrating an example of a configuration of a diagnostic system according to a first embodiment of this invention. FIG. It is Example 1 of this invention, and is a block diagram which shows an example of a structure of a diagnostic apparatus. It is Example 1 of the present invention, and is a lineblock diagram of rolling equipment as a diagnostic object. It is a figure which shows Example 1 of this invention and shows an example of equipment operation log DB. It is a figure which shows Example 1 of this invention and shows an example of the element division | segmentation screen of an element division part. It is a figure which shows Example 1 of this invention and shows an example of equipment element DB. It is a block diagram. It is a flowchart which shows Example 1 of this invention and shows an example of a process of the abnormality detection part. It is Example 1 of this invention, and is a graph which shows an example of the data of equipment operation log DB regarding the sensor A and the sensor B. FIG. It is a graph which shows Example 1 of this invention and shows an example of an abnormality detection log. It is a flowchart which shows Example 1 of this invention and shows an example of a process of the input sensor estimation part. It is a figure which shows Example 1 of this invention and shows an example of input sensor DB. It is a figure which shows Example 1 of this invention and shows the example of a display of the diagnostic result display part at the time of abnormality non-occurrence | production. It is a figure which shows Example 1 of this invention and shows the example of a display of the diagnostic result display part at the time of abnormality occurrence. It is a block diagram which shows Example 2 of this invention and shows an example of a structure of the power generation equipment as a diagnostic object. It is a figure which shows Example 2 of this invention and shows an example of equipment element DB. It is a figure which shows Example 2 of this invention and shows an example of input sensor DB. It is a figure which shows Example 2 of this invention and shows the example of a display of a diagnostic result display part. It is a flowchart which shows Example 1 of this invention and shows an example of the process which calculates an abnormality degree and a contribution degree. It is a graph which shows Example 1 of this invention and shows an example of the process which calculates an abnormality degree and a contribution degree. It is Example 1 of this invention, and is a flowchart which shows an example of the process performed in the abnormality filter part.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

<Overview>
The first embodiment is a diagnostic system for diagnosing equipment abnormality and its cause. This diagnostic system improves the diagnostic accuracy by dividing the equipment into a plurality of elements and individually diagnosing each element. Further, sensors related to each element are classified into input sensors and other sensors. The input sensor is a sensor that is easily influenced by external factors or represents the influence of external factors. By using the information of the input sensor, it is possible to determine whether the abnormality that has occurred in each element is an external influence or due to a factor in the element, thereby improving the diagnostic accuracy.

<Concept>
In classifying an input sensor and other sensors, in a facility in which an output is generated based on the input, for example, a feature that vibration of the input is attenuated by the output is used. Specifically, using the feature that abnormalities (vibrations) that occur outside the facility decrease as they propagate from input to output, the degree of abnormality (or the average value of the degree of abnormality) in the normal state of the elements that make up the equipment Select the larger sensor as the input sensor.

<Diagnostic system>
FIG. 1A is a block diagram illustrating an example of a configuration of a diagnostic system.

  The diagnosis system includes a facility to be diagnosed 130 that is a facility to be diagnosed, a facility diagnosis device 100 that diagnoses elements of the facility to be diagnosed 130, and a network 16 that connects the facility to be diagnosed 130 and the device diagnosis device 100.

  The equipment diagnosis apparatus 100 includes a processor 10 that performs arithmetic processing, a memory 11 that stores programs and data, a storage apparatus 12 that stores programs and data, an interface 13 connected to a network 16, and a keyboard, mouse, or touch panel. Is a computer that includes an input device 14 configured as described above and an output device 15 configured as a display or a touch panel.

  Programs of the log collection unit 101, the pre-diagnosis processing unit 110, the element division unit 103, the equipment diagnosis unit 120, and the diagnosis result display unit 106 are loaded into the memory 11 and executed by the processor 10.

  The processor 10 operates as a functional unit that provides a predetermined function by performing processing according to a program of each functional unit. For example, the processor 10 functions as the log collection unit 101 by performing processing according to the log collection program. The same applies to other programs. Furthermore, the processor 10 also operates as a functional unit that provides the functions of a plurality of processes executed by each program. A computer and a computer system are an apparatus and a system including these functional units.

  The storage device 12 stores an equipment operation log DB (database) 102, an input sensor DB 105, and an equipment element DB 106, and is accessed from the functional unit. The facility operation log DB 102 stores operation information of the diagnosis facility 130. The operation information is composed of sensor data of the diagnosis facility 130, for example.

  The facility element DB 106 stores the relationship between the elements of the diagnosis facility 130 and the sensors. The input sensor DB 105 stores information on the input sensors arranged for each element of the diagnosis facility 130.

  FIG. 1B is a block diagram illustrating an example of a functional part of the equipment diagnosis apparatus 100. In the first embodiment, an example in which a rolling facility is used as the diagnosis facility 130 and abnormality relating to temperature is diagnosed will be described. In the equipment diagnosis apparatus 100, a log collection unit 101 that collects equipment operation logs from the equipment to be diagnosed 130 collects sensor data and stores it in the equipment operation log DB 102.

  The element dividing unit 103 divides the constituent elements of the diagnosis facility 130 (hereinafter simply referred to as elements) as described later, and stores them in the facility element DB 104.

  The pre-diagnosis processing unit 110 includes a data division unit 111, an abnormality detection unit 112, and an input sensor estimation unit 113. As described later, an input sensor is provided for each element based on the facility operation log DB 102 and the facility element DB 104. Estimated and stored in the input sensor DB 105.

  The facility diagnosis unit 120 includes a data division unit 121, an abnormality detection unit 122, and an abnormality filter unit 123. As will be described later, the facility operation log DB 102 is analyzed to detect an abnormality and identify an element in which the abnormality has occurred. Do. The diagnosis result display unit 106 outputs the diagnosis result of the facility diagnosis unit 120 to the output device 15 or an external computer.

<Rolling equipment>
FIG. 2 is a diagram illustrating an example of a configuration of the rolling equipment 200 that is the equipment to be diagnosed 130 according to the first embodiment.

  When the material is input to the inlet 201 of the rolling facility 200, the material is rolled and comes out from the outlet 202 of the rolling facility 200. From the inlet 201 to the outlet 202, the temperature of the material gradually decreases. If the material temperature changes abnormally in the rolling equipment, it is considered that an abnormality has occurred in the rolling equipment.

  In order to detect this abnormality, an example in which four sensors 203-1 to 203-4 for measuring temperature are installed in the rolling equipment 200 is shown. In the following description, the reference numeral 203 without “−” is used for the entire sensor. The facility diagnostic apparatus 100 evaluates the relationship between the outputs of the sensors 203.

  In the illustrated rolling equipment 200, the entire equipment is divided into a first half 207 and a second half 208, a sensor 203-1 (sensor A) is installed at the position of the inlet 201 of the first half 207, and at the boundary with the second half 208 serving as the outlet of the first half 207. The sensor 203-2 (sensor B) is installed, the sensor 203-3 (sensor C) is installed in the middle of the second half 208, and the sensor 203-4 (sensor D) is installed at the outlet 202.

  In the first embodiment, the measured value of the sensor 203-2 is the measured value at the outlet of the first half 207 and the measured value at the inlet of the second half 208.

<Equipment operation log DB>
The log collection unit 101 uses SCADA (Supervisory Control And Data Acquisition) that operates in the diagnosis facility 130 to collect data of each sensor 203 of the diagnosis facility 130 as an facility operation log. The facility operation log is time-series data in which sensor values are recorded at predetermined time intervals for the sensor 203 in the diagnosis target facility 130. In addition, about the collection of the data of the sensor 203 from the to-be-diagnosed apparatus 130, what is necessary is just to apply a well-known or well-known technique, Therefore It does not elaborate in a present Example.

  FIG. 3 is a diagram illustrating an example of the configuration of the facility operation log DB 102. The equipment operation log DB 102 stores the equipment operation log collected by the log collection unit 101. The facility operation log DB 102 stores time series data (sensor data) including columns 1022 to 1025 for storing measured values of the sensors A to D 203-1 to 203-4 in one entry with the time 1021 as an index. The identifiers of the sensors A to D 203-1 to 203-4 are given to the names of the columns 1022 to 1025, and the data of the specific sensor 203 can be searched from the time and the identifier.

  The facility operation log DB 102 may be one facility operation log DB 102 for a plurality of diagnosed facilities 130, or may be prepared for each diagnosis facility 130. Further, the facility operation log DB 102 may be installed at a position distant from the facility diagnostic apparatus 100.

<Element division part>
The element dividing unit 103 divides the diagnosis facility 130 into one or a plurality of elements. The division of the elements of the diagnosis facility 130 may be determined by the user of the facility diagnosis apparatus 100 or the administrator of the diagnosis facility 130 based on domain knowledge, or may be automatically divided based on the facility operation log. Also good.

  In the case of the rolling equipment 200 of FIG. 2, an example in which the first half 207 and the second half 208 are divided into a plurality of elements for each process is shown. The divided elements are registered in the facility element DB 104.

  FIG. 4 is a diagram illustrating an example of the element division screen 400 when the user determines the element division of the element division unit 103. The element division screen 400 includes an output device element division screen 400, a sensor list 401 that displays a list of identifiers of the sensors 203, an element (first half) 402, and an element (second half) 403 that displays the identifiers of the sensors 203 for each element. , An element registration button 404 and an element addition button 405 are included.

  The sensor list 401 is an area for displaying a list of identifiers of the sensors 203 installed in the diagnosis facility 130. An element (first half) 402 and an element (second half) 403 display a list of registration statuses of the sensor 203 related to each element. As for the sensor 203 related to each element, one sensor 203 may be registered as a sensor related to a plurality of elements.

  For example, the sensor B (203-2) is the sensor 203 located at the outlet of the first half 207 of the process and the sensor 203 at the inlet of the second half 208 of the process. In such a case, the sensor 203-2 may be registered as the sensor 203 relating to both the “first half” and “second half” elements.

  The sensor registration button 404 for an element is a button for registering the sensor 203 related to the element. When the user selects the sensor 203 from the sensor list 401, selects an addition destination element and presses the button 404, the selected sensor is registered as a related sensor of the selected element.

  The element addition button 405 is a button for registering a new element. When the user presses an element addition button 405, a new element is registered. The added element is stored in the facility element DB 104.

<Equipment element DB>
The facility element DB 104 stores the relationship between the element determined by the element dividing unit 103 and the sensor 203.

  FIG. 5 is a diagram illustrating an example of the configuration of the facility element DB 104. The facility element DB 104 is a table having, as an index, an element 1041 that stores an element identifier, and a column that includes a sensor 1042 that stores an identifier of the sensor 203 assigned to each element.

  In the element dividing unit 103, when the sensor A 203-1 and the sensor B 203-2 are assigned to the “first half” 207 of the element, and the sensor B 203-2, the sensor C 203-3, and the sensor D 203-4 are assigned to the element “second half” 208, Information of each element and the sensor 203 is stored as shown in FIG.

<Pre-diagnosis processing unit>
The pre-diagnosis processing unit 110 refers to the equipment operation log DB 102 and the equipment element DB 104 to estimate the input sensor of each element. The input sensor is a sensor that measures an input from the outside of each element. When an abnormality occurs in the element due to this input sensor, the equipment diagnosis apparatus 100 determines that the abnormality is an abnormality that has occurred outside the element. To do. As a result, it is possible to easily identify the cause of the abnormality and improve the diagnostic accuracy.

  The pre-diagnosis processing unit 110 includes a data dividing unit 111, an abnormality detecting unit 112, and an input sensor estimating unit 113.

  The data division unit 111 is a process of dividing the data of the sensor 203 for each element with reference to the equipment operation log DB 102 and the equipment element DB 104. This process is started after the data dividing unit 111 receives designation of a period from the input device 14. The data dividing unit 111 acquires data at the time 1021 corresponding to the specified period from the equipment operation log DB 102 and classifies the data for each sensor 1042 of the element 1041.

  In the first embodiment, an example is shown in which the time when the data of the sensor 203 is measured is used as an index. However, a serial number corresponding to the time may be used as an index for time-series data.

<Abnormality detection unit>
The abnormality detection unit 112 analyzes the data of the equipment operation log DB 102 classified for each element by the data division unit 111 to determine whether there is an abnormality. If the data is abnormal, the abnormality occurrence time and the cause of the abnormality are detected. The specified sensor 203 is specified. The sensor 203 that caused the abnormality is assumed to be an abnormality cause sensor. FIG. 6 is a flowchart illustrating an example of processing of the abnormality detection unit 112.

  In this process, first, in step 601, the abnormality detection unit 112 calculates the degree of abnormality for the data of the sensor 203 in the element classified for each element, performs abnormality diagnosis, and if an abnormality occurs, the abnormality is detected. The sensor 203 having a high degree is specified.

  Next, in step 602, the abnormality detection unit 112 calculates the contribution degree of each sensor 203 with respect to the abnormality degree of the data of the sensor 203 in which the abnormality has occurred. The contribution degree indicates the degree to which the data of each sensor 203 has an influence on the data of the sensor 203 in which an abnormality has occurred.

  Through the above processing, the sensor 203 in which an abnormality has occurred, the degree of abnormality, and the degree of contribution are calculated from the equipment operation log DB 102 corresponding to the specified period.

  The processing of element abnormality degree calculation (step 601) and contribution degree calculation step 602 for the abnormality degree of each sensor 203 can be realized by a known or well-known abnormality diagnosis algorithm. The requirement of this algorithm is that it is an abnormality diagnosis algorithm that can calculate the abnormality time and the degree of abnormality of each sensor 203 at that time.

Specifically, multivariate statistical process management utilizing principal component analysis and abnormality diagnosis algorithms such as the Mahalanobis-Taguchi method can be applied. In the algorithm for multivariate process management, Hotteling T ² statistics may be calculated using an equipment operation log as an input, and the contribution of T ² statistics of each sensor may be used as the degree of abnormality of each sensor. In the Mahalanobis Taguchi method, the Mahalanobis distance may be calculated using the equipment operation log as an input, and this may be used as the degree of abnormality of each sensor. Further, the equipment operation log may be divided by time windows, the amplitude for each sensor 203 in each time window may be calculated, and this amplitude may be used as the degree of abnormality of each sensor.

  FIG. 17 shows a flowchart when the abnormality detection unit 112 performs abnormality diagnosis by principal component analysis and multivariate statistical process management. For the principal component analysis (PCA) and multivariate statistical process control (MSPC), refer to the document "Multivariate Statistical Process Control" (http://manabukano.brilliant-future.net/ research / report / Report2005_MSPC.pdf).

  The abnormality detection unit 112 normalizes the data of the facility operation log DB 102 classified for each element (610), and performs dimension compression by principal component analysis (PCA) (611). Next, the abnormality detection unit 112 calculates a Q statistic according to the following equation (1) disclosed in the above-mentioned document “Multivariate Statistical Process Management”.

Next, the abnormality detection unit 112 calculates the T ² statistic as the degree of abnormality for each time according to the following equation (2) disclosed in the above-mentioned document “multivariate statistical process management”.

Then, the abnormality detecting section 112 determines that monitors both the Q statistic and T ² statistic, or even one data exceeds a predetermined control limit is an abnormal value.

  The abnormality detection unit 112 calculates the contribution degree of the data of each sensor 203 determined as an abnormal value by the following equation (3).

  The above equation (3) is a technique disclosed in the document “Total PLS Based Contribution Plots for Fault Diagnosis” (http://www.sciencedirect.com/science/article/pii/S1874102908600944).

  FIG. 18 is a graph when the abnormality detection unit 112 performs abnormality diagnosis by principal component analysis and multivariate statistical process management.

  The state after normalization in step 610 of FIG. 17 plots the data of the sensor 203 for each time 1021, as shown by a graph 631 in FIG. In the figure, circles and stars are data of the sensor 203. In the figure, variable 1 indicates time, for example, and variable 2 indicates temperature, for example.

Next, when dimension compression is performed on the normalized data using principal component analysis in step 611 in FIG. 17, principal components 1 and 2 are calculated as shown by a graph 632. The result of calculating the T ² statistic for the data subjected to dimensional compression becomes a graph 633 in FIG. 18, and data of the degree of abnormality exceeding the control limit from the origin can be specified as abnormal data.

  FIG. 7 is a graph showing an example of data in the facility operation log DB 102 regarding the sensors A and B. The graph 701 in the figure represents the time transition of the data (sensor value) of the sensor A (203-1) in time series, and the graph 702 in the figure represents the time of the data (sensor value) of the sensor B (203-2). This is a time series of changes. In the figure, the horizontal axis represents time, and the vertical axis represents temperature.

  FIG. 7 shows that an abnormal temperature change occurred in the graph 701 of the sensor A from time t0 to t1, and thereafter an abnormal temperature change occurred in the graph 702 of the sensor B from time t2 to t3. Further, it is shown that the maximum value d1 of the amplitude in the graph 701 of the sensor A is larger than the maximum value d2 of the amplitude of the sensor B.

  FIG. 8 is an abnormality detection log obtained by inputting the data of the facility operation log DB 102 shown in FIG. 7 to the abnormality diagnosis algorithm shown in FIG.

  The abnormality detection log is a graph 800 showing time series data representing the time transition of the degree of abnormality of the elements in the first half 207 of the rolling equipment 200, and a graph showing time series data representing the time transition of the degree of abnormality of each sensor A and B. 801, 802. In the figure, the horizontal axis represents time, and the vertical axis represents the degree of abnormality. Note that the degree of abnormality of the element is the maximum value of the degree of abnormality of the sensor B203 in the element. In the illustrated example, the maximum value at each time among the data of the sensor A 203-1 and the data of the sensor B 203-2 in the element “first half” 207 is the degree of abnormality of the element “first half” 207.

  Graphs 800 to 802 indicate that the degree of abnormality is higher as the abnormality degree value is larger. FIG. 8 shows that a large abnormality occurred in the sensor A of the graph 801 from time t0 to t1, and then a large abnormality occurred in the sensor B of the graph 802 from time t2 to t3. Further, the abnormality degree 811 of the sensor A is larger than the abnormality degree 812 of the sensor B in the normal period from the time t1 to the time t2.

  Further, when there are a plurality of sensors 203 in the element, the abnormality detection unit 112 can identify the sensor 203 having a large contribution as the abnormality cause sensor.

<Input sensor estimation unit>
The input sensor estimation unit 113 estimates the input sensor of each element based on the processing result of the abnormality detection unit 112.

  FIG. 9 is a flowchart illustrating an example of processing performed by the input sensor estimation unit 113. This process is executed after the process of the abnormality detection unit 112 is completed. The processing performed by the input sensor estimation unit 113 includes a step 901 for extracting a normal period, a step 902 for calculating the degree of abnormality with an average value within the normal period, and a step 903 for estimating an input sensor.

  In step 901 of normal period extraction, the input sensor estimation unit 113 extracts a period in which the operation state of the rolling equipment 200 is normal. In step 901, the user may give a normal period based on the facility operation log DB 102. Further, the input sensor estimation unit 113 may calculate the average and variance of the values of each sensor 203 using the equipment operation log DB 102, and the period during which the data value of the sensor 203 is in a preset range may be set as a normal period. .

  Furthermore, a period in which the degree of abnormality calculated by the abnormality detection unit 112 is lower than a predetermined threshold may be set as a normal period. Alternatively, the input sensor estimation unit 113 divides the time series data of the facility operation log DB 102 by time windows, calculates the variance for each sensor 203 in each time window, and sets the period during which the variance value is lower than a predetermined threshold value. It may be a normal period. Further, the variance for each sensor 203 in the time window may be handled as the degree of abnormality of each sensor B203.

  In step 902 in which the degree of abnormality is calculated with the average value of the normal period, the input sensor estimation unit 113 calculates the degree of abnormality of the past element and the degree of abnormality of each sensor 203 using the past facility operation log DB 102. Further, the average value of the normal period calculated in step 901 is calculated for the degree of abnormality of each sensor 203.

  In step 903 of estimating the input sensor of each element, the input sensor estimation unit 113 estimates the input sensor of each element based on the degree of abnormality of each sensor 203 in the normal period calculated in step 902. And the input sensor estimation part 113 stores the estimation result of the input sensor of each element in input sensor DB105.

  Basically, a sensor 203 having a higher degree of abnormality or an average value of the degree of abnormality is estimated as an input sensor. For example, when there are two sensors 203 related to a certain element, the sensor having the highest degree of abnormality is set as the input sensor. Also, in the case of an element that can be regarded as one input sensor, for example, a one-input facility, the sensor 203 having the highest degree of abnormality is used as the input sensor. In an element in which the number of input sensors of the element can be specified in advance, the user sets the number of input sensors to n for each element, and in step 903, the top n of the sensors 203 having a high degree of abnormality are estimated as input sensors. May be. Furthermore, the user may set a threshold value for each element in advance, and one or a plurality of sensors 203 whose abnormalities exceed the threshold value may be estimated as input sensors.

  In addition, when a difference is not seen in the abnormality degree of the some sensor 203, it is not necessary to estimate an input sensor. Furthermore, regarding the result (input sensor DB 105) estimated in step 903 for estimating the input sensor, the user may check the estimation result and cancel or correct it as necessary.

  As described above, the method of estimating the input sensor based on the degree of abnormality can be used more easily than the method of specifying the input sensor based on the time delay. Furthermore, the input sensor can be estimated even using time-series data having a data interval that is so long that a time delay cannot be expressed.

<Input sensor DB>
The input sensor DB 105 stores input sensor information of each element, which is a processing result of the diagnosis preprocessing unit 110. FIG. 10 is a diagram illustrating an example of the configuration of the input sensor DB 105. The input sensor DB 105 is a table having the element 1051 as an index and the input sensor 1052 as a column. When the pre-diagnosis processing unit 110 obtains the result that the input sensor of the element “first half” is the sensor A and the input sensor of the element “second half” is the sensor B, this information is stored as shown in FIG.

<Equipment diagnosis department>
The equipment diagnosis unit 120 is a process for detecting an abnormality occurrence time and an abnormal element of the equipment 130 to be diagnosed. The equipment diagnosis unit 120 includes a data division unit 121, an abnormality detection unit 122, and an abnormality filter unit 123. The data division unit 121 and the abnormality detection unit 122 are the same processes as the data division unit 111 and the abnormality detection unit 112 of the pre-diagnosis processing unit 110, respectively.

  The equipment diagnosis unit 120 reads the data of the sensor 203 from the equipment operation log DB 102 for the period specified by the input device 14 or the like, and the data division unit 121 refers to the equipment element DB 104 to obtain the data of the sensor 203 as the equipment to be diagnosed. Divide into 130 elements.

  Next, the equipment diagnosis unit 120 inputs the data divided for each element to the abnormality detection unit 112, and detects abnormality for each element based on the degree of abnormality by the abnormality diagnosis algorithm of FIG. 6 and FIG. To do.

  The abnormality filter unit 123 filters the output result of the abnormality detection unit 122 based on the input sensor information for each element stored in the input sensor DB 105. That is, when an abnormality is detected by a certain element, the abnormality occurrence time and abnormality cause sensor information are sent to the abnormality filter unit 123.

  When the abnormality cause sensor matches the input sensor of the element, the abnormality filter unit 123 determines that the abnormality is caused by a factor outside the element, and does not determine that the element is abnormal. Thereby, when an abnormality occurs in a plurality of elements, the true abnormality factor can be narrowed down, and the abnormality diagnosis accuracy can be improved.

  FIG. 19 is a flowchart illustrating an example of processing performed in the abnormality filter unit 123. This process is started when the process of the abnormality detection unit 122 is completed. The abnormality filter unit 123 receives the detection result of the abnormality detection unit 122 (620). Next, the abnormality filter unit 123 repeatedly executes Steps 621 to 626 for each element for all elements set in the facility element DB 104.

  The abnormality filter unit 123 refers to the facility element DB 104, identifies the sensor 1042 of the currently selected element 1041, and determines whether any of the sensors 1042 is an abnormality cause sensor. If there is an abnormality cause sensor in the element, the process proceeds to step 623, and if there is no abnormality cause sensor in the element, the next element is selected and the above process is repeated. If the last element has been reached, the process proceeds to step 627.

  In step 623, the abnormality filter unit 123 acquires the identifier of the abnormality cause sensor, and determines whether the input sensor 1052 of the element 1051 of the input sensor DB 105 matches the identifier of the abnormality cause sensor.

  If the abnormality cause sensor is the input sensor of the element, the abnormality filter unit 123 proceeds to step 624 and classifies the element as an abnormality-related element. On the other hand, if the abnormality cause sensor is not the input sensor of the element, the abnormality filter unit 123 proceeds to step 625 and classifies the element as the abnormality cause element. Then, the next element is selected and the above process is repeated. If the last element has been reached, the process proceeds to step 627.

  In step 627, the abnormality filter unit 123 outputs the classification result of the element in which the abnormality is detected to the diagnosis result display unit 106, and ends the process.

  As described above, when the abnormality cause sensor is the input sensor of the element, the abnormality filter unit 123 determines that an abnormality has occurred due to the influence from the outside (previous stage or upstream), and the element is abnormal. Judged as a related element. The abnormality-related element indicates that the value of the sensor 203 indicates an abnormality, but the element to be inspected or repaired is another element.

  Further, if the abnormality cause sensor is not the input sensor of the element, the abnormality filter unit 123 determines that an abnormality has occurred in the element, and determines that the element is an abnormality cause element. The abnormality cause element indicates that the element should be inspected or repaired.

  When there are a plurality of elements having an abnormality cause sensor, the abnormality filter unit 123 uses an abnormality cause element to be inspected or repaired and an element that is abnormal in the data of the sensor 203 but suspected of external influence as an abnormality related element. By categorizing, it is possible to provide priorities for the elements to be inspected or repaired.

<Diagnosis result display>
The diagnosis result display unit 106 outputs the result of the equipment diagnosis unit 120. 11 and 12 are examples of a diagnostic result display screen output by the diagnostic result display unit 106. FIG.

  FIG. 11 is a display example of the diagnosis result display unit 106 when no abnormality has occurred in each element. In FIG. 11, the diagnosis result display screen 1100 displays the sensors 203 of the first half 207 and the second half 208 as elements of the rolling equipment 200. Sensor A 203-1 is displayed above the element as the input sensor in the first half 207, and sensor B 203-2 B is displayed above the element as the input sensor in the second half 208. In the figure, since the sensor C203-3 and the sensor D203-4 are not determined as input sensors by the diagnosis preprocessing unit 110, they are displayed in a parallel relationship downstream of the sensor B203-2B.

  In the element “first half” 207 of the rolling equipment 200, the only input sensor is the sensor A 203-1, and therefore, an abnormality from the outside propagates to the sensor B 203-2 A through the sensor A 203-1 in the element “first half” 207. Therefore, the causal relationship 1113 between the sensor A 203-1 and the sensor B 203-2 can be derived.

  In the element “second half” 208, the input sensor is only the sensor B 203-2B. Therefore, in the element “second half” 208, an abnormality from the outside propagates to the sensor C 203-3 and the sensor D 203-4 through the sensor B 203-2B. Therefore, the causal relationships 1124 and 1125 between the sensor B 203-2B, the sensor C 203-3, and the sensor D 203-4 can be derived. Further, the sensor B 203-2 A is a non-input sensor in the element “first half” 207, and the sensor B 203-2 B is an input sensor in the element “second half” 208. This information suggests a causal relationship 1130 that the abnormality of the element “first half” 207 propagates to the element “second half” 208.

  FIG. 12 is a display example of the diagnosis result display screen 1100 when a sudden temperature change occurs in the sensor B 203-2 and the abnormality detection unit 122 detects an abnormality in both the “first half” and “second half”. . The diagnosis result display unit 106 displays an alert on the sensor B 203-2A in the first half 207 and the sensor B 203-2B in the second half 208 because an abnormality has occurred due to the data of the sensors B 203-2A and 203 on the diagnosis result display screen 1100. To do.

  The diagnosis result display unit 106 displays (outputs) the abnormality cause sensor and the elements displayed on the diagnosis result display screen 1100 according to the classification result of the abnormality filter unit 123. If the element including the abnormality cause sensor is an abnormality cause element (first abnormality element), the diagnosis result display unit 106 displays, for example, an element display area and an abnormality cause sensor display area with an alert display color or blinking. Display with changes such as.

  On the other hand, if the element including the abnormality cause sensor is an abnormality related element (second abnormality element), for example, the display area of the element is set as a normal display, and the display area of the abnormality cause sensor is displayed as a warning. And so on.

  The display form of the abnormality cause element and the abnormality related element may be any display form that can identify whether the element including the abnormality cause sensor is the abnormality cause element or the abnormality related element. In addition, the diagnosis result display unit 106 may be configured to display an abnormal cause sensor in an identifiable manner and to display an input sensor in an irrelevant element.

  However, in the element “second half” 208, since the sensor B 203-2B is an input sensor, it is classified as an abnormality-related element, and the display area of the sensor B 203-2B is weaker than the sensor B 203-2A in the first half 207 (attention) It is displayed in a display color indicating (arousal).

  On the other hand, in the element “first half” 208, since the sensor B 203-2A is a non-input sensor, the “first half” 207 is classified as an abnormality-related element, and the display area of the element “first half” 207 is a display color indicating an alert. In addition, the display area of the sensor B 203-2 of the abnormality cause sensor is displayed in a display color indicating a warning to notify the user that there is an abnormality in the element “first half” 207. Thereby, the user can narrow down the cause of abnormality easily.

  According to the present invention, the equipment diagnosis apparatus 100 can improve equipment diagnosis accuracy by narrowing down the cause of an abnormality. In particular, even when abnormalities are detected in multiple elements at the same time, the cause of the abnormality seems to be the cause of the abnormality, including the cause of the abnormality that should be repaired or inspected immediately and the cause sensor of the abnormality. By classifying the elements into abnormality-related elements, it is possible to narrow down the elements that cause the abnormality and thus improve the diagnostic accuracy. Abnormalities include rapid changes in temperature, failure and deterioration in equipment, performance recovery due to maintenance, and fluctuations in external factors such as outside air temperature and material temperature.

  The facility diagnosis apparatus 100 may be separated from the facility to be diagnosed 130 as long as the operation information of the facility to be diagnosed 130 can be collected. For example, the diagnostic apparatus 100 may be installed in a factory where the diagnosis facility 130 is installed, or may be provided as one of cloud services. By providing it as a cloud service, it is possible to provide the service in a short period of time.

  In addition, the facility diagnosis apparatus 100 may perform facility diagnosis of the facility 130 to be diagnosed offline. That is, the operation information of the equipment 130 to be diagnosed may be conveyed by a storage medium or the like, input to the equipment diagnosis apparatus 100, stored in the equipment operation log DB 102, and equipment diagnosis may be performed. Thereby, pre-diagnosis before the start of the diagnostic service can be realized.

  The facility diagnosis apparatus 100 of the present invention can also be applied to abnormality diagnosis of power generation facilities. In the second embodiment, an example of application to power generation equipment is shown as the equipment to be diagnosed of the equipment diagnosis apparatus 100, and a bearing vibration cause analysis of the generator is performed. Other configurations are the same as those in the first embodiment. The generator bearings are vibrated due to aging and grease defects. If the vibration is left unattended, bearing failure may occur, leading to a reduction in generator operating time. In order to prevent such a situation, the bearings of the power generation equipment can be monitored, and if there is a sign of abnormality, measures can be taken early.

  FIG. 13 is a block diagram illustrating an example of a configuration of a power generation facility 1300 that is a diagnosis target facility according to the second embodiment. The power generation facility 1300 includes a generator 1310, a speed increaser 1320, a generator shaft 1311, a main shaft 1321, a generator shaft bearing 1314, and a main shaft bearing 1324.

  The rotation of the main shaft 1321 drives the generator shaft 1311 via the speed increaser 1320, and the generator 1310 connected to the generator shaft 1311 is driven to generate power. The main shaft 1321 is coupled with a windmill (not shown) to transmit driving force.

  The main shaft 1321 and the generator shaft 1311 are provided with a rotational speed sensor A1312, a rotational speed sensor B1322, an acceleration sensor A1313, and an acceleration sensor B1323 for detecting the rotational speed. Further, an acceleration sensor C1303 for detecting vibration from the environment is installed outside the power generation facility 1300.

  FIG. 14 is a diagram illustrating an example of the configuration of the facility element DB 104 in the present embodiment. The elements of the power generation facility 1300 are shown as an example in which the element dividing unit 103 is divided into a generator 1310 and a speed increaser 1320 and a connection unit that connects both in the same manner as in the first embodiment. The connecting portion is an element connected by a gear (not shown) of the main shaft 1321 and a gear (not shown) of the generator shaft 1311.

  In consideration of propagation of external vibration to the generator 1310, the sensors related to the generator 1310 and the acceleration sensor A1313 (acceleration A in the figure) of the generator shaft 1311 and the rotation speed sensor A1312 of the generator 1310 (in the figure) Rotation speed A) and outer acceleration sensor C1303 (acceleration C in the figure).

  The sensors related to the speed increaser 1320 consider the fact that external vibration propagates to the speed increaser 1320, and the acceleration sensor B 1323 (acceleration B in the drawing) of the main shaft 1321 and the rotation speed sensor B 1322 of the speed increaser 1320. (The rotational speed A in the figure) and the outer acceleration sensor C1303 (acceleration C in the figure). Sensors related to the connection portion are a rotation speed sensor B1322 (rotation speed B in the figure) of the speed increaser 1320 and a rotation speed sensor A1312 (rotation speed A in the figure) of the generator 1310.

  FIG. 15 is a diagram illustrating an example of the configuration of the input sensor DB 105 according to the second embodiment. This configuration was calculated by the pre-diagnosis processing unit 110 using the time series data of the power generation facility 1300 as an input.

  The processing contents of the pre-diagnosis processing unit 110 are the same as those in the first embodiment. As the input sensor 1052 of the element “generator” 1310 of the power generation facility 1300, the rotational speed sensor A1312 (the rotational speed A in the figure) and the acceleration sensor C1303 are used. Is selected, the rotation speed sensor B1322 (rotation speed B in the figure) and the acceleration sensor C1303 are selected as the input sensor 1052 of the element “speed increaser” 1320, and the input sensor 1052 of the “connection unit” is not provided. Show.

  FIG. 16 is an image of a screen 1600 output by the diagnosis result display unit 106 when the information of the input sensor DB 105 shown in FIG. 15 is given. By giving the information of FIG. 15, the equipment diagnosis unit 120 determines that only excessive fluctuation of the data of the acceleration sensor A1313 (acceleration A in the figure) due to the bearing 1314 is abnormal, and the rotation speed sensor A1312 of the generator 1310. Excessive fluctuations in the data of the rotation speed A (in the figure) and excessive fluctuations in the data of the acceleration sensor A1313 (acceleration A in the figure) due to the data of the acceleration sensor C1303 (acceleration C in the figure) are not determined as abnormal. , Diagnostic accuracy can be improved.

  It should be noted that an abnormality caused by excessive fluctuations in the data of the rotational speed sensor A 1312 (the rotational speed A in the figure) is not detected by the element “generator” 1620 but may be detected by the element “connection unit” 1630. it can.

  As described above, the facility diagnosis apparatus 100 according to the present invention can be applied to the power generation facility 1300 in addition to the rolling facility 200 of the first embodiment, and can accurately diagnose the abnormality of facilities and apparatuses composed of a plurality of elements. Can be implemented. In particular, even when abnormality cause sensors occur in a plurality of elements, it is possible to identify an abnormality cause element that should be urgently checked or repaired by classifying it into an abnormality cause element and an abnormality related element by the abnormality filter unit 123. It becomes.

<Summary>
As described above, in the present invention, facilities and devices are divided into a plurality of elements, and each element is individually diagnosed to identify an abnormality in the facility and the cause of the abnormality. The element dividing unit divides the facility into a plurality of elements. The user may divide or automatically divide based on the equipment operation log.

  The diagnosis preprocessing unit estimates an input sensor of each element. The diagnosis pre-processing unit is based on the characteristic that the vibration of the input is attenuated in the output in known or publicly known equipment or devices in which the output is generated based on the input, and the abnormality generated outside the equipment propagates from the input to the output. Assuming that the effect decreases, the input sensor of each element is estimated. Specifically, the degree of abnormality of each sensor of the element in the normal period is evaluated, and a sensor having a large degree of abnormality is estimated as an input sensor.

  The equipment diagnosis unit performs an abnormality detection process for each element of the equipment, and when an element abnormality is detected, the equipment diagnosis unit classifies the element into an abnormality cause element and an abnormality related element according to the position within the element of the abnormality cause sensor. When the cause of abnormality is an input sensor of an element, it is classified as an abnormality-related element, and although the sensor value indicates an abnormality, it is determined that the cause of the abnormality has occurred outside the element. Note that it may be determined that the element is a false positive and removed from the abnormality detection result.

  In this way, by classifying elements including abnormality cause sensors into abnormality cause elements and abnormality related elements, and setting high priority for repair and inspection for abnormality cause elements, the diagnosis accuracy of facilities and equipment can be improved, It is possible to increase the operating rate of the equipment.

  Further, according to the present invention, even when an abnormality is detected with a plurality of elements during facility diagnosis, the elements causing the abnormality can be narrowed down, and the diagnostic accuracy can be improved. The present invention can also be applied to equipment in which it is difficult to specify the input or output of an element from a design drawing. Further, the present invention can be applied to facilities in which the data interval of time series data is long and a time delay between inputs or outputs cannot be detected. Thereby, a highly accurate equipment diagnosis can be provided for a wide range of equipment.

  In addition, this invention is not limited to an above-described Example, Various modifications are included. For example, the above-described embodiments are described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Further, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment, and the configuration of another embodiment can be added to the configuration of one embodiment. In addition, any of the additions, deletions, or substitutions of other configurations can be applied to a part of the configuration of each embodiment, either alone or in combination.

  Each of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. In addition, each of the above-described configurations, functions, and the like may be realized by software by the processor interpreting and executing a program that realizes each function. Information such as programs, tables, and files for realizing each function can be stored in a memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

  Further, the control lines and information lines indicate what is considered necessary for the explanation, and not all the control lines and information lines on the product are necessarily shown. Actually, it may be considered that almost all the components are connected to each other.

10 processor 11 memory 11
12 Storage device 100 Equipment diagnosis device 102 Equipment operation log DB
104 Equipment element DB
105 Input sensor DB
106 Diagnosis result display unit 110 Pre-diagnosis processing unit 113 Input sensor estimation unit 120 Equipment diagnosis unit 122 Abnormality detection unit 123 Abnormal filter unit 130 Diagnosed equipment

Claims (14)

A facility diagnostic apparatus having a processor and a memory and diagnosing an abnormality in a facility from a value measured by a sensor,
One or more elements constituting the facility, and facility element information that sets a relationship between sensors installed in the element,
Time-series sensor data including the sensor identifier and the value;
Among the sensors in the element, input sensor information that sets the input sensor, and
An abnormality detection unit that identifies the sensor that has detected an abnormality from the value of the sensor data as an input of the sensor data as an abnormality cause sensor, and identifies the element that includes the abnormality cause sensor with reference to the facility element information;
It is determined whether or not the abnormality cause sensor is an input sensor of the specified element with reference to the input sensor information. If the abnormality cause sensor is not the input sensor, the element is designated as a first abnormality element. And when the abnormality cause sensor is the input sensor, an abnormality filter unit that classifies the element as a second abnormality element;
A facility diagnostic apparatus comprising: The equipment diagnostic apparatus according to claim 1,
A diagnostic pre-processing unit for setting the input sensor information;
The pre-diagnosis processing unit
A second abnormality detection unit that calculates the degree of abnormality of the sensor value using the sensor data as an input;
An input sensor that extracts a normal period in which the value falls within a predetermined range from the time-series sensor data, estimates the input sensor from the degree of abnormality of the sensor value in the normal period, and sets the input sensor information An estimation unit;
A facility diagnostic apparatus comprising: The facility diagnostic apparatus according to claim 2,
The second abnormality detector is
An equipment diagnostic apparatus characterized in that the time-series sensor data is divided into time windows, and the variance of the sensor values is defined as the degree of abnormality in each time window. The facility diagnostic apparatus according to claim 2,
The second abnormality detector is
A facility diagnosis apparatus that calculates Hotteling T ² statistics using the time-series sensor data as an input, and uses the contribution to the Hotteling T ² statistics for each sensor as an abnormality level. The equipment diagnostic apparatus according to claim 1,
A diagnostic result display unit for generating a diagnostic result display screen based on the classification result of the abnormal filter unit;
The diagnostic result display unit
Based on the facility element information, the element and the sensor installed on the element are displayed on the diagnosis result display screen, and if the sensor is an abnormality cause sensor, the abnormality cause sensor is included based on the classification result. An equipment diagnosis apparatus, characterized in that an element is displayed on the diagnosis result display screen so as to be identifiable as to whether the element is a first abnormal element or a second abnormal element. The equipment diagnostic apparatus according to claim 5,
The diagnostic result display unit
In the first abnormal element, the abnormality cause sensor is displayed on the diagnostic result display screen in an identifiable manner, and in the second abnormal element, the input sensor is displayed on the diagnostic result display screen in an identifiable manner. Equipment diagnostic device to do. The facility diagnostic apparatus according to claim 4,
The second abnormality detector is
A facility diagnosis apparatus characterized in that a contribution degree is calculated using the sensor data as an input, and a sensor having a large contribution degree is specified as an abnormality cause sensor. An equipment diagnosis method for diagnosing equipment abnormality from a value measured by a sensor using a computer having a processor and a memory,
1 step, wherein the calculator inputs time-series sensor data including the sensor identifier and the value;
A second step in which the computer specifies a sensor that has detected an abnormality from the value of the sensor data as an abnormality cause sensor;
A third step in which the computer specifies the element including the abnormality cause sensor with reference to facility element information in which a relationship between one or more elements constituting the facility and a sensor installed in the element is set; ,
The computer determines whether or not the abnormality cause sensor is an input sensor of the specified element with reference to input sensor information in which the input sensor is set among the sensors in the element, and the abnormality cause sensor If the input sensor is not the input sensor, the element is classified as a first abnormal element, and if the abnormality cause sensor is the input sensor, the fourth step of classifying the element as a second abnormal element;
A facility diagnosis method comprising: The facility diagnosis method according to claim 8,
The computer further includes a fifth step of setting the input sensor information;
The fifth step includes
An abnormality degree of the sensor value is calculated using the sensor data as input, a normal period in which the value falls within a predetermined range is extracted from the time-series sensor data, and an abnormality degree of the sensor value in the normal period is extracted. A facility diagnosis method characterized in that the input sensor is estimated and set in the input sensor information. The facility diagnosis method according to claim 9,
The fifth step includes
A facility diagnosis method, wherein the time-series sensor data is divided into time windows, and the variance of the sensor values in each time window is defined as an abnormality level. The facility diagnosis method according to claim 9,
The fifth step includes
A facility diagnosis method characterized in that a Hottelling T ² statistic is calculated using the time-series sensor data as an input, and a contribution to the Hotteling T ² statistic for each sensor is defined as an abnormality level. The facility diagnosis method according to claim 8,
The computer further includes a sixth step of generating a diagnostic result display screen based on the classification result;
The sixth step includes
Based on the facility element information, the element and a sensor installed in the element are displayed on the diagnosis result display screen. When the sensor is an abnormality cause sensor, the abnormality cause sensor is determined based on the classification result. A facility diagnosis method characterized by displaying on the diagnosis result display screen so as to be able to identify whether a contained element is a first abnormal element or a second abnormal element. The facility diagnosis method according to claim 12,
The sixth step includes
In the first abnormal element, the abnormality cause sensor is displayed on the diagnostic result display screen in an identifiable manner, and in the second abnormal element, the input sensor is displayed on the diagnostic result display screen in an identifiable manner. Equipment diagnosis method to do. The facility diagnosis method according to claim 11,
The fifth step includes
A facility diagnosis method characterized in that a contribution is calculated using the sensor data as an input, and a sensor having a large contribution is specified as an abnormality cause sensor.

Images