Relationship identification device, relationship identification method, and relationship identification program

Abstract

In the present invention, an analysis unit (103) uses each of a plurality of devices as a target device and uses each of one or more patterns of device change, which is a change in the target device, as a target pattern to generate generation data indicating a device change occurrence timing for the target pattern of the target device. A relationship identification unit (104) identifies, on the basis of the occurrence data generated by the analysis unit (103), a set of device changes having a causal relationship, from among the device changes for the plurality of devices.

Description

Relationship identification device, relationship identification method, and relationship identification program

[0001] The present disclosure relates to a technique for identifying relationships between changes between devices.

[0002] Patent Document 1 describes a technique for extracting inter-device dependencies for the purpose of assisting in debugging a sequence program executed by a control device. In Patent Document 1, a device is a variable that stores a control signal value. A dependency is a relationship in which the value of one device determines the value of another device. In Patent Document 1, inter-device dependencies are extracted based on log data of devices in a program. Specifically, Patent Document 1 calculates the time difference between the date and time when one device changes and the date and time when another device changes, and extracts the dependencies based on the average or standard deviation of the calculated time difference.

[0003] International Publication No. 2021 / 210071

[0004] The technology described in Patent Document 1 does not take into consideration cases where one device changes in response to multiple patterns of device changes. Here, a pattern refers to a device change that uniquely determines the impact on other devices. Note that in Patent Document 1, a device means a variable. However, a device is not limited to a variable, and may be any type of equipment. In Patent Document 1, because a device is a variable, a device change is a change in value. However, if a device is an equipment, a device change is an equipment operation.

[0005] For example, suppose there are multiple operating patterns, and after device A changes, device B may or may not change. In this case, the change interval between device A and device B may be short or extremely long (or the change interval may not be calculable). With the technology described in Patent Literature 1, in such cases, device A and device B would be determined to have no relationship or to have a partial relationship. For applications that present relationships to users for debugging support, even extracting a partially recognized relationship is useful. However, when extracting relationships for systems that predict device change intervals or detect anomalies, a strict relationship in which changes correspond to each other is required. Therefore, extracting relationships using the technology described in Patent Literature 1 is insufficient. In the above example, the relationship should be determined after distinguishing between multiple operating patterns. Specifically, the change in device A that contributes to the change in device B is classified as "change A1," and the change in device A that does not contribute to the change in device B is classified as "change A2." In this case, it can be determined that there is always a causal relationship between change A1 and device B, and that there is always no causal relationship between change A2 and device B. This allows the user to accurately understand the operation of the control device when used for debugging support purposes. Furthermore, for systems that predict device change intervals or detect anomalies, the ability to associate changes between devices each time enables appropriate prediction of device change intervals or detection of anomalies.

[0006] The present disclosure aims to make it possible to appropriately identify the relationship between device changes.

[0007] The relationship identification device according to the present disclosure includes an analysis unit that generates occurrence data indicating the timing of occurrence of device changes for the target patterns of a plurality of devices, each of which is a target device and each of which is a target pattern, and a relationship identification unit that identifies, based on the occurrence data generated by the analysis unit, sets of device changes that have a causal relationship among the device changes for the plurality of devices.

[0008] In the present disclosure, the timing of occurrence of device changes is analyzed for each device change pattern, and a set of device changes having a causal relationship is identified. As a result, the device changes are distinguished by pattern, and a set of device changes having a causal relationship is identified. Therefore, it is possible to appropriately identify the relationship between device changes.

[0009] 1 is a functional configuration diagram of a relationship identification device 100 according to a first embodiment. FIG. 2 is a hardware configuration diagram of the relationship identification device 100 according to the first embodiment. FIG. 3 is a flowchart of overall processing by the relationship identification device 100 according to the first embodiment. FIG. 4 is a diagram showing an example of a control device having a plurality of patterns of device changes according to the first embodiment. FIG. 5 is a diagram showing device changes of the control device shown in FIG. 4 according to the first embodiment. FIG. 6 is an explanatory diagram of classification according to the order in which device changes occur according to the first embodiment. FIG. 7 is an explanatory diagram of classification according to the states of other devices before and after a device change occurs according to the first embodiment. FIG. 8 is an explanatory diagram of a method of outputting a relationship according to the first embodiment. FIG. 9 is a configuration diagram of a relationship identification device 100 according to a second embodiment. FIG. 10 is a flowchart of overall processing by the relationship identification device 100 according to the second embodiment. FIG. 11 is an explanatory diagram of classification of device changes of output devices by a control program according to the second embodiment. FIG. 12 is an explanatory diagram of classification of device changes of input devices by a control program according to the second embodiment. FIG. 13 is an explanatory diagram of classification according to control design information according to the second embodiment. FIG. 14 is an explanatory diagram of classification according to classification conditions according to the second embodiment.

[0010] Embodiment 1. ***Description of Configuration*** The functional configuration of a relationship identification device 100 according to embodiment 1 will be described with reference to Fig. 1 . The relationship identification device 100 according to embodiment 1 is connected to a PLC 200. PLC stands for Programmable Logic Controller. The PLC 200 is connected to a plurality of devices 300 and controls each of the devices 300.

[0011] The relationship identification device 100 includes, as functional components, a data collection unit 101, a pattern classification unit 102, an analysis unit 103, a relationship identification unit 104, and a relationship output unit 105. The data collection unit 101 collects device log data from the PLC 200. The pattern classification unit 102 classifies device changes of each device into one or more patterns based on the collected log data. The analysis unit 103 analyzes device changes from the collected log data for each pattern of each device. The relationship identification unit 104 identifies a set of device changes having a causal relationship based on the analysis results. The relationship output unit 105 outputs the identified results. In the first embodiment, the device may be the device 300, a piece of equipment constituting the device 300, or a variable in a program. A device change is an operation when the device is the device 300 or an equipment, and a change in value when the device is a variable.

[0012] With reference to FIG. 2 , a hardware configuration of the relationship identification device 100 according to the first embodiment will be described. The relationship identification device 100 is a computer. The relationship identification device 100 includes a processor 10, a memory 11, an input device 12, a display device 13, a storage device 14, and a communication device 15. The processor 10 is an IC that executes various processes. IC stands for Integrated Circuit. A specific example of the processor 10 is a CPU. CPU stands for Central Processing Unit. The memory 11 is a device that temporarily stores data. A specific example of the memory 11 is a RAM. RAM stands for Random Access Memory. The input device 12 is a device that inputs data. A specific example of the input device 12 is a keyboard. The display device 13 is a device that displays data. A specific example of the display device is an LCD. LCD is an abbreviation for Liquid Crystal Display. The storage device 14 is a device that stores data. A specific example of the storage device 14 is an SSD. SSD is an abbreviation for Solid State Drive. The communication device 15 is a device that communicates with devices external to the relationship identification device 100. A specific example of the communication device 15 is a network card.

[0013] ***Description of Operation*** The operation of relationship identification device 100 according to embodiment 1 will be described with reference to Figures 3 to 8. The operation procedure of relationship identification device 100 according to embodiment 1 corresponds to the relationship identification method according to embodiment 1. Furthermore, a program that realizes the operation of relationship identification device 100 according to embodiment 1 corresponds to the relationship identification program according to embodiment 1.

[0014] The overall processing of the relationship identification device 100 according to the first embodiment will be described with reference to Fig. 3. (Step S101: Data Collection Processing) When each device 300 is operating, the data collection unit 101 collects device log data from the PLC 200. The data collection unit 101 stores the collected log data in the memory 11.

[0015] (Step S102: Pattern Classification Process) The pattern classification unit 102 classifies the device changes for each device into one or more patterns based on the log data collected in step S101.

[0016] (Step S103: Analysis Process) The analysis unit 103 analyzes the log data collected in step S101. At this time, the analysis unit 103 distinguishes and analyzes each of the one or more patterns classified in step S102. That is, the analysis unit 103 sets each of the multiple devices as a target device, and sets each of one or more patterns of device changes, which are changes in the target device, as a target pattern. Then, the analysis unit 103 analyzes the device changes for the target pattern of the target device. The analysis unit 103 analyzes the timing of device changes and generates occurrence data indicating the timing of device changes. Specifically, the analysis unit 103 calculates, for the target pattern of the target device, statistics such as the number of device changes, the interval between device changes, and the average or standard deviation of the interval between device changes. Then, the analysis unit 103 includes the number of occurrences, the interval between device changes, the statistics, etc. in the occurrence data.

[0017] (Step S104: Relationship Identification Process) The relationship identification unit 104 identifies a set of device changes having a causal relationship among the device changes for the multiple devices indicated by the log data, based on the occurrence data generated in step S103. A causal relationship is a relationship in which when a device change of a source pattern occurs in a source device, a device change of a destination pattern occurs in a destination device.

[0018] (Step S105: Relationship Output Process) The relationship output unit 105 outputs the relationship identified in step S104.

[0019] An example of a device with multiple device change patterns will be described with reference to FIGS. 4 and 5 . As shown in FIG. 4 , device Y1, which drives conveyor 1, changes ON twice in one cycle. The first time, by carrying a workpiece on conveyor 1, it contributes to an OFF change of device X1 corresponding to sensor 1 and an ON change of device X2 corresponding to sensor 2. The second time, by transferring a workpiece from conveyor 1 to conveyor 2, it contributes to an OFF change of device X2. The device change at this time is as shown in FIG. 5 . The first ON change (1) and the second ON change (2) of device Y1 correspond to other device changes as indicated by the dashed arrows. For devices with multiple device change patterns, such as device Y1, in step S102 the pattern classification unit 102 classifies the device as separate patterns, such as device Y1 ON change (1) and device Y1 ON change (2). Then, in step S103, the analysis unit 103 calculates the number of changes, the change intervals relative to other device changes, and the statistics of the change intervals for each pattern.

[0020] A method for classifying device changes into one or more patterns in step S102 will be described. Here, two classification methods, classification method 1 and classification method 2, will be described as methods for classifying device changes into one or more patterns.

[0021] Classification method 1 will be described with reference to FIG. 6 . Classification method 1 is a method for distinguishing device change patterns based on the order in which device changes occur in one cycle of operation. One cycle of operation is a series of operations repeatedly performed by the device 300. In other words, one cycle of operation is one periodic period of operation performed by the device 300. In the example of FIG. 6 , the Y1 device changes to ON twice in one cycle. Therefore, the pattern classification unit 102 distinguishes between (1) the first change to ON and (2) the second change to ON, which are device changes of the Y1 device, and classifies them into different patterns.

[0022] The pattern classification unit 102 can divide the operation cycle of the device by the following methods: (A) The pattern classification unit 102 identifies repetition of device changes by pattern recognition based on log data, and divides each repetition into one cycle; (B) The pattern classification unit 102 has the user set the time for one operation cycle, and divides each set time into one cycle; (C) The pattern classification unit 102 has the user specify a device that changes only once per cycle, and divides one cycle based on the device change of the specified device.

[0023] Classification method 2 will be described with reference to FIG. 7 . Classification method 2 is a method for distinguishing device change patterns based on the states of other devices before and after a device change occurs. In FIG. 7 , the Y1 device changes to ON twice. The values ​​representing the states of other devices before and after these two changes are as follows: Y1ON change (1) is M1 = 01, M2 = 00, M3 = 11, .... Y1ON change (2) is M1 = 00, M2 = 01, M3 = 11, .... Therefore, the states of other devices before and after Y1ON change (1) and Y1ON change (2) are different. Therefore, the pattern classification unit 102 distinguishes between Y1ON change (1) and Y1ON change (2) and classifies them into different patterns.

[0024] If the states of all other devices are used to classify the patterns, an infinite number of patterns may be generated. Therefore, the pattern classification unit 102 may use only the states of devices that affect the subsequent change intervals of other devices to classify the patterns. The devices to be used to classify the patterns may be extracted using decision tree analysis or the like.

[0025] A method for identifying a pair of device changes having a causal relationship in step S104 will be described. The relationship identifying unit 104 sets each combination of two devices as a target combination. The relationship identifying unit 104 determines that a causal relationship exists when a specific pattern of device changes corresponds to each other in every cycle between the devices in the target combination. Alternatively, the relationship identifying unit 104 determines that a causal relationship exists when the average or standard deviation of the change intervals of the specific pattern of device changes between the devices in the target combination is equal to or less than a preset threshold. When devices are divided into cause side and result side, the relationship identifying unit 104 sets each result side device as a target device. The relationship identifying unit 104 may determine that a causal relationship exists between each pattern of device changes of the target device and the device change pattern of the cause side device with the smallest average or standard deviation of the change intervals.

[0026] A method for outputting the relationships identified in step S105 will be described with reference to FIG. 8 . The relationship output unit 105 displays patterns of related device changes on a screen to a person in charge of debugging for the purpose of supporting debugging. Alternatively, the relationship output unit 105 passes relationship information to a system that predicts device change intervals or detects anomalies. As shown in FIG. 8 , the relationship information is information that indicates devices, changes, and patterns (conditions) for both the cause side and the result side. Note that in FIG. 8 , if a device has only one device change pattern, a "-" is set in the pattern column.

[0027] ***Effects of First Embodiment*** As described above, the relationship identification device 100 according to the first embodiment analyzes the timing of occurrence of a device change for each device change pattern and identifies a set of device changes having a causal relationship. This allows the device changes to be distinguished by pattern and then a set of device changes having a causal relationship to be identified. This makes it possible to appropriately identify the relationship between device changes.

[0028] For example, as shown in FIG. 5 , assume that the ON changes of device Y1, which have two patterns, are not distinguished. In this case, the ON changes of device Y1 do not correspond to changes of device X1 or device X2 every time, and therefore are determined to have no causal relationship. It is possible to assume that there is a partial causal relationship. However, this is inappropriate when identifying relationships for systems that predict device change intervals or detect anomalies, as a strict relationship in which changes correspond every time is required. In contrast to this example, the relationship identification device 100 according to embodiment 1 distinguishes between the two patterns of ON changes of device Y1. That is, the first ON change (1) of device Y1 and the second ON change (2) are treated as separate patterns. In this case, the ON change (1) of device Y1 and the OFF change of device X1, the ON change (1) of device Y1 and the ON change of device X2, and the ON change (2) of device Y1 and the OFF change of device X2 are each associated every time. Therefore, it is possible to determine that there is a causal relationship between these device changes and to properly calculate the change interval, which also makes it possible to predict the change interval or detect an abnormality.

[0029] ***Other Configurations*** <Modification 1> In the first embodiment, each functional component is realized by software. However, in Modification 1, each functional component may be realized by hardware. The differences between Modification 1 and the first embodiment will be described below.

[0030] When each functional component is realized by hardware, the relationship identification device 100 includes an electronic circuit instead of the processor 10 and the memory 11. The electronic circuit is a dedicated circuit that realizes the functions of each functional component and the memory 11.

[0031] Possible electronic circuits include a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, a logic IC, a GA, an ASIC, and an FPGA. GA stands for Gate Array. ASIC stands for Application Specific Integrated Circuit. FPGA stands for Field-Programmable Gate Array. Each functional component may be realized by a single electronic circuit, or each functional component may be distributed across multiple electronic circuits.

[0032] <Modification 2> As a modification 2, some of the functional components may be realized by hardware, and other functional components may be realized by software.

[0033] The processor 10, the memory 11, and the electronic circuit are collectively referred to as a processing circuit. That is, the functions of the respective functional components are realized by the processing circuit.

[0034] Embodiment 2. Embodiment 2 differs from embodiment 1 in that the pattern classification unit 102 classifies patterns by referring to input information other than log data. In embodiment 2, this difference will be explained, and explanations of the same points will be omitted.

[0035] ***Description of Configuration*** The configuration of a relationship identification device 100 according to embodiment 2 will be described with reference to Fig. 9. The relationship identification device 100 differs from the relationship identification device 100 shown in Fig. 1 in that the pattern classification unit 102 refers to input information from the input device 12.

[0036] ***Description of Operation*** The overall processing of the relationship identification device 100 according to the second embodiment will be described with reference to Fig. 10. The processing of step S111 is the same as the processing of step S101 in Fig. 3. The processing of steps S113 to S115 is the same as the processing of steps S103 to S105 in Fig. 3.

[0037] (Step S112: Pattern Classification Process) The pattern classification unit 102 classifies the device changes for each device into one or more patterns based on the log data collected in step S101 and the input information input from the input device 12.

[0038] A method for classifying device changes into one or more patterns in step S112 will be described. Here, three classification methods, Classification Method 3 to Classification Method 5, will be described as methods for classifying device changes into one or more patterns.

[0039] Classification method 3 will be described with reference to FIGS. 11 and 12 . Classification method 3 is a method of distinguishing device change patterns based on a control program that controls multiple devices as input information. When the target device is an output device, the pattern classification unit 102 classifies the device changes for the target device into one or more patterns based on the combination of devices used in the control program as conditions for changing the target device. When the target device is an input device, the pattern classification unit 102 classifies the device changes for the target device into one or more patterns based on the combination of devices used in the control program as targets for change by the target device.

[0040] For an output device such as a Y device, the pattern classification unit 102 distinguishes the device change pattern based on the device used as a condition for changing the device in the control program. FIG. 11 shows an example in which the control program is a ladder program. In this example, there are two patterns for the contacts that turn on the Y1 device: the M1 device and the M2 device. Therefore, the ON change of the Y1 device can be divided into a case in which the M1 device changes ON, a case in which the M2 device changes ON, and a case in which both the M1 device and the M2 device change ON simultaneously. The pattern classification unit 102 classifies each of these three cases into a different pattern.

[0041] For an input device such as an X device, the pattern classification unit 102 distinguishes the device change pattern based on the device used as the target of the change in the control program. FIG. 12 shows an example in which the control program is a ladder program. In this example, there are two patterns for the coils that the X1 device turns ON: the Y1 device and the Y2 device. Therefore, the ON change of the X1 device can be distinguished into a case where the M1 device is ON and the Y1 device simultaneously changes ON, and a case where the M2 device is ON and the Y2 device simultaneously changes ON. In addition, there are also cases where both the Y1 device and the Y2 device change ON, and cases where neither changes ON. Therefore, the pattern classification unit 102 classifies these four cases into separate patterns.

[0042] Classification method 4 will be described with reference to FIG. 13 . Classification method 4 is a method of distinguishing device change patterns based on control design information, which is input information for controlling multiple devices. When the target device is an output device, the pattern classification unit 102 classifies the device changes for the target device into one or more patterns based on the combination of devices used as conditions for changing the target device in the control design information. When the target device is an input device, the pattern classification unit 102 classifies the device changes for the target device into one or more patterns based on the combination of devices used as targets for change by the target device in the control design information.

[0043] 13 shows an example of a case where the control design information is a flowchart. In this example, there are two patterns in which the Y1 device changes to ON: pattern (1) in which Y1 changes to ON when the M1 device turns ON, and pattern (2) in which Y1 changes to ON when the M2 device turns ON. The pattern classification unit 102 classifies these two patterns into different patterns.

[0044] Classification method 5 will be described with reference to FIG. 14 . Classification method 5 is a method in which classification conditions for classifying device change patterns input by a user are used as input information, and device change patterns are distinguished based on the classification conditions. FIG. 14 shows an example of a classification condition input screen. In this example, an ON change of device Y1 is specified as the device change for which patterns are to be distinguished. Three classification conditions are specified as a method of distinction: when M1 is ON, when M1 is OFF and M2 is ON, and other cases. The pattern classification unit 102 classifies patterns that match the three classification conditions into different patterns according to these classification conditions.

[0045] ***Effects of Embodiment 2*** As described above, the relationship identification device 100 according to Embodiment 2 distinguishes device change patterns based on classification conditions, using input information entered by a user, etc. This enables more appropriate distinction of device changes.

[0046] In a control program, if the conditions for changing an output device are different, the intended behavior is likely to be different. Also, if the device changed by an input device is different, the equipment state indicated by the input is likely to be different. Therefore, it makes sense to distinguish device change patterns from this information. Similarly, control design information can obtain information such as the conditions for changing an output device and the effects of changing an input device, making it possible to distinguish appropriate device change patterns just as when using a control program. The method in which the user directly inputs device change patterns makes it possible to more reliably distinguish appropriate device change patterns.

[0047] The relationship identification device 100 according to the first embodiment is superior to the relationship identification device 100 according to the second embodiment in the sense that it can distinguish device change patterns basically using only log data without using input information. The relationship identification device 100 according to the first embodiment requires the user to input device information indicating the cycle time or cycle in some cases. This has the effect of reducing the number of steps required for the user to input control programs or control design information, etc., and enabling implementation even when there are no control programs or control design information, etc.

[0048] The term "unit" in the above description may be read as a "circuit," "step," "procedure," "process," or "processing circuit."

[0049] The embodiments and modifications of the present disclosure have been described above. Some of these embodiments and modifications may be combined and implemented. Furthermore, one or more of them may be implemented partially. Note that the present disclosure is not limited to the above embodiments and modifications, and various modifications are possible as needed.

[0050] 100 Relationship identification device, 200 PLC, 300 device, 101 data collection unit, 102 pattern classification unit, 103 analysis unit, 104 relationship identification unit, 105 relationship output unit, 10 processor, 11 memory, 12 input device, 13 display device, 14 storage device, 15 communication device.

Claims

1. A relationship identification device comprising: an analysis unit that generates occurrence data indicating the occurrence timing of device changes for the target patterns of a plurality of devices, each of which is a target device and each of which is a target pattern of one or more patterns of device changes that are changes in the target devices; and a relationship identification unit that identifies, based on the occurrence data generated by the analysis unit, sets of device changes that have a causal relationship among the device changes for the plurality of devices.

2. The relationship identification device according to claim 1, further comprising a pattern classification unit that classifies device changes for the target device into one or more patterns based on the order in which the device changes occur in one cycle of operation, and the analysis unit generates the occurrence data using each of the one or more patterns classified by the pattern classification unit for the target device as a target pattern.

3. The relationship identification device according to claim 1, further comprising a pattern classification unit that classifies device changes for the target device into one or more patterns based on the states of other devices before and after each device change for the target device occurs, and the analysis unit generates the generated data using each of the one or more patterns classified by the pattern classification unit for the target device as a target pattern.

4. The relationship identification device according to claim 1, further comprising a pattern classification unit that classifies device changes for the target device into one or more patterns based on at least one of a combination of devices used as a condition for changing the target device and a combination of devices used as a target for the change made by the target device in a control program that controls the multiple devices, and the analysis unit generates the generated data using each of the one or more patterns classified by the pattern classification unit for the target device as a target pattern.

5. The relationship identification device according to claim 1, further comprising a pattern classification unit that classifies device changes for the target device into one or more patterns based on at least one of a combination of devices used as a condition for changing the target device and a combination of devices used as a target for change by the target device in control design information that specifies the control of the plurality of devices, and the analysis unit generates the generated data using each of the one or more patterns classified by the pattern classification unit for the target device as a target pattern.

6. The relationship identification device according to claim 1, further comprising a pattern classification unit that classifies device changes for the target device into one or more patterns based on classification conditions for classifying patterns specified by a user, and the analysis unit generates the generated data using each of the one or more patterns classified by the pattern classification unit for the target device as a target pattern.

7. A relationship identification method in which a computer sets each of a plurality of devices as a target device, sets each of one or more patterns of device changes that are changes in the target devices as a target pattern, generates occurrence data indicating the timing of the device changes for the target patterns of the target devices, and identifies, based on the occurrence data, sets of device changes that have a causal relationship among the device changes for the plurality of devices.

8. A relationship identification program that causes a computer to function as a relationship identification device that performs an analysis process in which each of a plurality of devices is a target device, and each of one or more patterns of device changes that are changes in the target devices is a target pattern, to generate occurrence data that indicates the occurrence timing of the device changes for the target patterns of the target devices, and a relationship identification process in which, based on the occurrence data generated by the analysis process, a set of device changes that have a causal relationship among the device changes for the plurality of devices.

Images