Information processing device, analysis system, analysis method, and program
The information processing device analyzes equipment operations by extracting sequences from logs and calculating error associations, addressing the challenge of customizing event analysis systems for each industrial facility, thereby improving operational safety and efficiency.
Patent Information
- Application Number
- PCT/JP2023/019153
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-23
- Publication Date
- 2024-11-28
AI Technical Summary
Existing event analysis systems in industrial facilities require customization for each plant or line, as they struggle to analyze causal relationships between equipment operations and errors effectively, especially when changes in process data are not directly related to alarms, and the classification of operation events must be predefined.
An information processing device that acquires operation logs, extracts operation sequences based on time intervals, calculates the degree of association between these sequences and subsequent errors, and outputs analysis results, allowing for the analysis of various equipment operations without predefining event classifications.
Enables the analysis of equipment operations and error identification across different types of equipment, improving operational safety by determining erroneous operations and enhancing operational conditions without the need for customization based on specific plant or line configurations.
Smart Images

Figure JP2023019153_28112024_PF_FP_ABST
Abstract
Description
Information processing device, analysis system, analysis method and program
[0001] The present disclosure relates to an information processing device, an analysis system, an analysis method, and a program.
[0002] In industrial facilities where a variety of equipment is in operation, there is a technology for monitoring events, including equipment operation or alarm generation, and analyzing the causal relationships between events in order to improve operational safety or operational status (see, for example, Patent Literature 1). Systems that issue warnings based on the results of such analysis are also in use.
[0003] The event analysis device described in Patent Document 1 analyzes causal relationships by combining events, including alarms generated by multiple devices in a plant and operations on the devices, with process data from multiple devices. It is explained that this makes it possible to perform analysis that also includes fluctuations in process data that appear due to operations that are not causally related to the alarms.
[0004] Japanese Patent Application Laid-Open No. 2018-128855
[0005] The event analysis device described in Patent Document 1 performs a process of aggregating the same type of operation event into a single operation event data when the same type of operation event is repeatedly generated within a short period of time. Furthermore, when detecting changes in the data trend, the event analysis device also classifies the data into fluctuation intervals and aggregates the data. Such classification of the same type of operation event or fluctuation interval must be defined in advance, which poses a problem of requiring customization of the event analysis device according to the plant or production line.
[0006] The present disclosure has been made in consideration of the above circumstances, and aims to provide an information processing device, an analysis system, an analysis method, and a program that are capable of analyzing a variety of equipment or operations.
[0007] To achieve the above object, an information processing device of the present disclosure acquires an operation log including records of multiple operations of a target facility and an operation log including records of errors of the target facility, and when a time interval between operations is equal to or longer than a predetermined first time, extracts from the operation log a series of operations up to the operation immediately preceding the time interval to create an operation sequence. Furthermore, the information processing device calculates a degree of association between the operation sequence and an error that occurs after the completion of the last operation in the operation sequence, and outputs an analysis result based on the degree of association.
[0008] According to the present disclosure, an operation sequence showing a series of operations up to a certain time interval is created, and the degree of association with an error that occurred after the operation sequence is calculated, thereby enabling analysis of a variety of equipment or operations.
[0009] Block diagram showing an example of the overall configuration of an analysis system according to the first embodiment Flowchart showing analysis processing of a block diagram showing an example of the hardware configuration of an information processing device Flowchart showing operation sequence creation processing Diagram showing an example of an operation log Flowchart showing relevance calculation processing Diagram showing an example of creation of an operation sequence Flowchart showing independence determination processing Diagram showing an example of a diagram relevance table Diagram showing an example of an operation log Diagram showing an example of an actual measurement value used in independence determination Diagram showing an example of an relevance table Diagram showing an example of a theoretical value used in independence determination Flowchart showing operation sequence list processing of an operation error determination Diagram showing an example of an alarm output Block diagram showing an example of the overall configuration of an analysis system according to the second embodiment Flowchart showing learning processing Diagram showing an example of a correct label for learning data
[0010] First Embodiment A first embodiment for carrying out the present disclosure will now be described in detail with reference to the drawings, in which the same or corresponding parts are designated by the same reference numerals.
[0011] Fig. 1 is a block diagram showing an example of the overall configuration of an analysis system 100 according to embodiment 1. The analysis system 100 is a system that analyzes the relationship between operations and errors in order to prevent erroneous operations on a target facility 10 and improve the operating status of the target facility 10. As shown in Fig. 1 , the analysis system 100 includes the target facility 10 that are communicatively connected to each other, an input display device 20 that receives operation inputs on the target facility 10 and displays information related to the target facility 10, and an information processing device 30 that analyzes operations on the target facility 10 and errors that occur in the target facility 10.
[0012] 1, the analysis system 100 includes one target facility 10 and one input display device 20, but the number of target facilities 10 and the number of input display devices 20 may be arbitrary. The analysis system 100 may include two or more target facilities 10 or two or more input display devices 20.
[0013] The target equipment 10 is any equipment operating in an industrial facility, such as a robot cell. The target equipment 10 is operated based on operation signals from the input display device 20, and accumulates an operation log 11 in a storage unit, which includes a record of errors that occur as a result of the operation.
[0014] The input display device 20 has a function of receiving operation inputs from an operator, outputting operation signals to the target equipment 10, and displaying the operating status, errors, alarms, etc. of the target equipment 10 received from the target equipment 10 and the information processing device 30. The input display device 20 is, for example, a programmable display device in which a control program for the target equipment 10 is installed. The input unit of the input display device 20 may include physical buttons or may be a touch panel. The input display device 20 accumulates a record of operations performed on the input unit by the operator in a storage unit as an operation log 21.
[0015] Furthermore, the input display device 20 has a function of determining whether or not an input operation is likely to be an erroneous operation, and outputting an alarm when it is determined that there is a high possibility of an erroneous operation.
[0016] The information processing device 30 is any computer that has a function of analyzing operations on the target equipment 10 and errors that occur in the target equipment 10, based on the operation log 11 acquired from the target equipment 10 and the operation log 21 acquired from the input display device 20. The information processing device 30 is a general-purpose computer, tablet terminal, smartphone, or the like, on which an analysis program is installed.
[0017] Fig. 2 is a block diagram showing an example of the hardware configuration of the information processing device 30. In the example of Fig. 2, the information processing device 30 includes a processor 301, a primary storage device 302, a secondary storage device 303, an input unit 304, a display 305, and a communication interface 306, which are connected to each other via a bus 310.
[0018] The processor 301 includes, for example, a CPU (Central Processing Unit). The processor 301 loads a control program stored in the secondary storage device 303 into the primary storage device 302 and executes the program.
[0019] The primary storage device 302 includes, for example, a RAM (Random Access Memory), functions as a work memory for the processor 301 , and temporarily stores the control program that the processor 301 reads from the secondary storage device 303 .
[0020] The secondary storage device 303 includes, for example, an EEPROM (Electrically Erasable and Programmable Read Only Memory), a flash memory, etc. The secondary storage device 303 stores programs executed by the processor 301, or data, parameters, etc. used for analysis.
[0021] The input unit 304 includes, for example, a keyboard or a pointing device, accepts input from the administrator, and passes the input signal to the processor 301. The display 305 is any display device, such as a liquid crystal display or an organic EL (Electro Luminescence) display. The input unit 304 and the display 305 may be an integrated touch panel.
[0022] The communication interface 306 is an interface of any communication method for transmitting and receiving data to and from the target equipment 10 and the input display device 20, and includes a communication device such as Ethernet (registered trademark) or wireless LAN.
[0023] For example, the processor 301 and primary storage device 302 of the information processing device 30 cooperate to realize each function shown in Fig. 1 . That is, the information processing device 30 functions as a log acquisition unit 31 that acquires the operation log 11 and the operation log 21 from the target equipment 10 and the input display device 20, respectively, an operation sequence creation unit 32 that extracts a series of operations from the operation log 21 to create an operation sequence, and an association calculation unit 33 that calculates the association between the operation sequence and an error included in the operation log 11. Furthermore, the information processing device 30 functions as an independence determination unit 34 that determines the independence between the operation sequence and the error, and an analysis unit 35 that analyzes the operation and the error based on the association or independence. Moreover, either or both of the primary storage device 302 and the secondary storage device 303 function as the memory unit 40 of the information processing device 30.
[0024] The log acquisition unit 31 acquires the operation log 11 including records of errors from the target equipment 10, and also acquires the operation log 21 including records of operations on the target equipment 10 from the input display device 20, and stores them in the storage unit 40. When the time interval between one operation and the next operation is equal to or longer than a certain time interval, the operation sequence creation unit 32 extracts from the operation log 21 a series of operations up to the operation immediately preceding that time interval, and creates an operation sequence.
[0025] The relevance calculation unit 33 calculates the relevance between the operation sequence created by the operation sequence creation unit 32 and an error included in the operation log 11 that occurred within a certain time period after the end of the last operation of the operation sequence. The calculated relevance is associated with the operation sequence and stored in the relevance table 41 in the storage unit 40.
[0026] The independence determination unit 34 determines the independence between the operation sequence created by the operation sequence creation unit 32 and an error that has occurred in the target equipment 10 and is included in the operation log 11, and updates the degree of relevance calculated by the relevance calculation unit 33 according to the determined independence. The analysis unit 35 analyzes the operations and errors based on the degree of relevance calculated by the relevance calculation unit 33 or updated by the independence determination unit 34. Specifically, the analysis unit 35 determines whether a series of operations indicated in the operation sequence are erroneous operations based on the degree of relevance, and outputs a list of operation sequences (operation sequence list 42) including the determination result to the input display device 20.
[0027] The input display device 20 also has a hardware configuration including a processor and a primary storage device similar to those of the information processing device 30. The processor and primary storage device of the input display device 20 work together to function as an operation error determination unit 22. The operation error determination unit 22 compares a newly detected operation sequence with the operation sequence list 42 acquired from the information processing device 30, and determines that there is a high possibility of an operation error if the newly detected operation sequence is the same as an operation sequence determined to be an operation error, or if the operation sequence has not been detected in the past.
[0028] The operation of the analysis system 100 having the above-described configuration will be described with reference to the flowchart in Fig. 3. Fig. 3 is a flowchart of the analysis process executed by the information processing device 30. The analysis process may be executed at any timing, for example, when an execution instruction is given by the administrator of the information processing device 30, when the operation log 11 or the manipulation log 21 is received, or at a predetermined execution date and time.
[0029] First, the log acquisition unit 31 acquires the operation log 11 from the target equipment 10, acquires the operation log 21 from the input display device 20 (step S1), and stores them in the storage unit 40. Next, the operation sequence creation unit 32 extracts a set of operations that are estimated to have meaning as a series of operations from the operation log 21 acquired in step S1, and creates an operation sequence (step S2; operation sequence creation step). The operation sequence creation process will be described in detail using Figures 4 to 6. Figure 4 is a flowchart of the operation sequence process, Figure 5 is an example of the operation log 21, and Figure 6 is an example of the creation of an operation sequence.
[0030] In the flowchart of Fig. 4, first, the operation sequence creation unit 32 acquires the operation log 21 stored in the storage unit 40 (step S101). As illustrated in Fig. 5, the operation log 21 includes the date, time, identification information of the input display device 20, identification information of the screen, and identification information of the operation. Here, the identification information of the input display device 20, the identification information of the screen, and the identification information of the operation are simply represented by a list of characters or symbols, and the identification information does not have any meaning such as use or function. For example, the identification information of the screen does not include the meaning that "screen 1 is the initial screen." By not assigning meaning to each piece of identification information, it becomes possible to apply it to any equipment or operation.
[0031] The operation sequence creation unit 32 determines the time interval between operations from the date and time of the operation log 21 acquired in step S101, starting from the earliest operation in chronological order (step S102). 1 If the time interval between operations is less than the first time t (step S102; No), it is estimated that the operation is in progress, and the process proceeds to step S107 without extracting the operation sequence. 1 If the time interval is equal to or greater than this (step S102; Yes), the series of operations up to the previous operation in the time interval is estimated to be a meaningful group of operations, and is extracted as an operation sequence (step S103). The first operation in this series of operations is the operation following the operation sequence extracted in the previous step S103.
[0032] For the operation sequence extracted in step S103, the operation sequence creation unit 32 determines whether or not the same operation sequence exists in the operation sequence list 42 of the storage unit 40 (step S104). If the same operation sequence does not exist (step S104; No), the operation sequence creation unit 32 assigns a new index to the operation sequence extracted in step S103 and adds it to the operation sequence list 42 of the storage unit 40 (step S105). If the same operation sequence exists (step S104; Yes), the number of indexes for that operation sequence is incremented by 1 (step S106).
[0033] In the example of FIG. 5, the first time t 1 is set to 20 seconds, the time interval between the second and third operations is 20 seconds or more, so the operation sequence creation unit 32 extracts the first and second operations as a first operation sequence. Also, the time interval between the fifth and sixth operations is 20 seconds or more, so the operation sequence creation unit 32 extracts the third to fifth operations as a second operation sequence. Note that in this embodiment, the time interval between operations within an operation sequence is not taken into consideration.
[0034] 6 shows an operation sequence extracted from the operation log 21 of FIG. 5 . As shown in FIG. 6 , an operation sequence in which the first operation "Display 1 + Screen 1 + Button 1" and the second operation "Display 1 + Screen 2 + Button 1" in FIG. 5 are performed in this order is assigned a new index "1" and added to the operation sequence list 42 of the storage unit 40. Also, an operation sequence in which the third operation "Display 1 + Screen 1 + Button 1," the fourth operation "Display 1 + Screen 2 + Button 3," and the fifth operation "Display 1 + Screen 3 + Button 2" in FIG. 5 are performed in this order is assigned a new index "2" and added to the operation sequence list 42 of the storage unit 40. Hereinafter, when an operation sequence is generated that has the same operation content and procedure as a previously saved operation sequence, a new index is not assigned, and 1 is added to the number of occurrences of the operation sequence. Hereinafter, the operation sequence with index x will be referred to as operation sequence x.
[0035] Thereafter, if there are no unprocessed operations in the operation log 21 acquired in step S101 (step S107; No), the process ends. If there are unprocessed operations (step S107; Yes), the process returns to step S102 and continues.
[0036] After the operation sequence creation process in Fig. 4 is completed, the process returns to the flowchart in Fig. 3 and executes the relevance calculation process (step S3; relevance calculation step). The relevance calculation process will be described in detail with reference to Figs. 7 to 9. Fig. 7 is a flowchart of the relevance calculation process, Fig. 8 is an example of an operation log 11 including an error for which the relevance is to be calculated, and Fig. 9 is an example of how the relevance is calculated.
[0037] In the flowchart of Fig. 7, the relevance calculation unit 33 first acquires one operation sequence created in the operation sequence creation process (step S2 in Fig. 3) shown in Fig. 4 (step S201). Next, in the operation log 11 acquired from the target equipment 10 in step S1 in Fig. 3 and stored in the storage unit 40 of the information processing device 30, 2 The log in the file is acquired (step S202).
[0038] Next, the relevance calculation unit 33 calculates the second time t 2 The degree of association between an error included in the operation log 11 within a second time t after the end of the last operation in the operation sequence is calculated. 2 The relevance of each error that occurred within the specified time is calculated, and if the relevance of the same operation sequence index and the same error code has already been calculated, the relevance calculated this time is added to the relevance value stored at that time (step S203).
[0039] The operation log 11 shown in Fig. 8 records errors that occurred in the target equipment 10 when the operations shown in the operation log 21 in Fig. 5 were performed. As shown in Fig. 8, the operation log 11 includes a date, a time, and an error code. Here, the error code is simply represented by a list of letters or symbols, and does not have any meaning such as the type of target equipment or the content of the error.
[0040] Second time t 2 When the time is set to 1 minute, errors with error codes 1 to 3 (hereinafter referred to as errors 1 to 3, and the same applies below) occurred within 1 minute from 8:00:02, the end time of the last operation in operation sequence 1 shown in Figure 5. Also, errors 1 and 4 occurred within 1 minute from 8:01:20, the end time of the last operation in operation sequence 2 shown in Figure 5.
[0041] An example of calculating the degree of association between each error and each operation sequence will be described below. For example, when the time difference between the end time of operation sequence x and the occurrence time of error (y, i) is represented as t(x, y, i), the degree of association Assoc(x, y) may be defined by the following formula (1). Here, since the same error y may occur multiple times, error (y, i) indicates that the error y occurs the i-th time.
[0042]
[0043] FIG. 9 is a graph showing an example of the relevance between operation sequence x and error (y, i), which is the i-th component below Σ in equation (1). The horizontal axis of the graph in FIG. 9 represents the time difference t(x, y, i) between the end time of operation sequence x and the occurrence time of error (y, i), and the vertical axis represents the relevance of error (y, i). As shown in FIG. 9, the relevance calculated by equation (1) can be said to be the sum of index values that decrease with an increase in the time t(x, y, i) from the end of the last operation of operation sequence x to the occurrence of error (y, i). Note that the relevance may be defined by other indexes.
[0044] In the example shown in FIGS. 5 and 8, the second time t 2 When is set to 1 minute, errors 1 to 3 within 1 minute from the end of operation sequence 1 and errors 1 and 4 within 1 minute from the end of operation sequence 2 are subject to calculation of the relevance.
[0045] 10 shows a relevance table 41 containing the results of calculating the relevance expressed by equation (1) for the operation log 21 and the operation log 11 shown in FIGS. 5 and 8. In detail, as shown in FIG. 10, the relevance is calculated for each of error 1, which occurred at 8:00:04 within one minute after operation sequence 1 ended at 8:00:02, error 2, which occurred at 8:00:05, and error 3, which occurred at 8:00:06. In addition, the relevance is calculated for error 1, which occurred at 8:01:30 within one minute after operation sequence 2 ended at 8:01:20, and error 4, which occurred at 8:01:31.
[0046] Here, we show an example where each error occurs only once, but the relevance increases each time the same error occurs for the same operation sequence, that is, each time i increases. In other words, the higher the frequency of error occurrence, the higher the relevance, which can indicate the strength of the causal relationship between the operation sequence and the error.
[0047] 7 , the relevance calculation unit 33 calculates the relevance for each error in the operation sequence acquired in step S201 in step S203, calculates the relevance for each index of the operation sequence and for each error code by adding them up, and then stores the calculated relevance in the relevance table 41 in the storage unit 40 (step S204). The relevance calculation unit 33 then determines whether or not there is an unprocessed operation sequence (step S205), and ends the process if there is no unprocessed operation sequence. If there is an unprocessed operation sequence, the process returns to step S201, and continues the process of calculating the relevance for the next operation sequence.
[0048] After the relevance calculation process of Fig. 7 is completed, the process returns to the flowchart of Fig. 3, and the independence determination unit 34 executes the independence determination process (step S4; independence determination step). The independence determination process will be described in detail using Figs. 11 and 12A to 12C. Fig. 11 is a flowchart of the independence determination process, and Figs. 12A to 12C are specific examples for explaining the independence determination process.
[0049] In the flowchart of Fig. 11, first, the independence determination unit 34 acquires one operation sequence created in the operation sequence creation process (step S2 in Fig. 3) shown in Fig. 4 (step S301). Next, in the operation log 11 acquired from the target equipment 10 in step S1 in Fig. 3 and stored in the storage unit 40 of the information processing device 30, 2 The logs within the period are acquired (step S302).
[0050] Next, the independence determination unit 34 determines the independence between the operation sequence and an error of one error code selected from the errors included in the operation log 11. Any method may be used to determine the independence, but in this embodiment, a case where the independence is determined by a chi-square test will be described. In steps S303 to S305, the independence determination between the operation sequence x and the error y will be described.
[0051] First, the independence determination unit 34 calculates the test statistic of the chi-square test for the operation sequence x and the error y (step S303), and if the test statistic is equal to or less than the corresponding value of the significance level (step S304; Yes), it determines that there is independence, and updates the relevance stored in the relevance table 41 in the storage unit 40 to 0 (step S305). On the other hand, if the test statistic exceeds the corresponding value of the significance level (step S304; No), it determines that there is no independence and that there is a causal relationship, and proceeds to step S306 without updating the relevance. The chi-square test will be described using a specific example.
[0052] The chi-square test determines the independence of operation sequence x and error y based on the frequency of their occurrence. Here, "independence" refers to a state in which the frequency of overlapping occurrence of operation sequence x and error y is necessarily due to the high frequency of their occurrence. In the chi-square test, the predetermined significance level is, for example, 1% or 5%.
[0053] 12A to 12C, we will explain how to determine the independence of the combination of index 1 of an operation sequence and error code 2. Fig. 12A shows the actual measurement value of the number of times any error code appeared within one minute of the end of any operation sequence in the target period, when the total number of times that any error code appeared was 1,001. When an error appeared within one minute of the end of an operation sequence, the index of that operation sequence and the error code of that error are said to be related, and when an error did not appear within one minute of the end of the operation sequence, the error code of that error and the index of that operation sequence are said to be unrelated.
[0054] As shown in Figure 12A, out of a total of 1001 detections, there were 3 times when index 1 of the operation sequence was associated with error code 2. There were 48 times when index 1 of the operation sequence was associated with an error code other than error code 2, and there were 25 times when error code 2 was associated with an index other than index 1 of the operation sequence.
[0055] The theoretical values calculated based on the table of actual measurements shown in Fig. 12A are shown in Fig. 12B. For example, the total number of times that any operation sequence index and any error code are associated is 1001, of which 28 times the error code was 2 and 51 times the operation sequence index was 1. From this, a probability calculation is performed, and the theoretical value of the number of times that operation sequence index 1 and error code 2 are associated is 28 x 51 / 1001 = 1.42657. Calculations for other combinations are performed in the same way, and the theoretical values shown in Fig. 12B are obtained.
[0056] Next, the independence determining unit 34 calculates the test statistic of the chi-square test using the following formula (2).
[0057]
[0058] The test statistic calculated by applying equation (2) to the theoretical values shown in Figure 12B and the corresponding measured values in Figure 12A is 1.88118. A chi-square test in a 2x2 distribution table follows a chi-square distribution with one degree of freedom, and the chi-square value corresponding to a significance level of 5% for a chi-square distribution with one degree of freedom is 3.87. Since the test statistic calculated using equation (2) is 1.88118, which is less than 3.87, it is determined that index 1 in the operation sequence and error code 2 are independent.
[0059] 11, as described in the example above, the independence determination unit 34 calculates the test statistic of the chi-square test for the error of error code 1 (step S303), and since the test statistic is equal to or less than the chi-square value corresponding to the significance level (step S304; Yes), it is determined that index 1 of the operation sequence and error code 2 are independent, and the relevance is updated to 0, which is lower than the value 0.95 calculated by the relevance calculation unit 33 (step S305). Fig. 12C is a diagram showing the relevance table 41 including the relevance after being updated from Fig. 10.
[0060] After the processing of steps S304 and S305, if there are any unprocessed error codes (step S306; Yes), the process returns to step S303, and the independence determination unit 34 executes the chi-square test processing of steps S303 to S305 for the errors in the unprocessed error codes.
[0061] If there is no unprocessed error code (step S306; No), it is determined whether there is an unprocessed operation sequence (step S307). If there is an unprocessed operation sequence (step S307; Yes), the process returns to step S301, and steps S301 to S305 are executed for the unprocessed operation sequence. If there is no unprocessed operation sequence (step S307; No), the process ends.
[0062] In this way, in the flowchart of Figure 3, the relevance calculation unit 33 calculates the relevance (step S3), and after the independence determination unit 34 determines the independence and updates the relevance (step S4), the analysis unit 35 analyzes the operation and error based on the relevance and outputs the analysis result to the input display device 20 (step S5; analysis step).
[0063] Specifically, if the relevance is equal to or greater than a predetermined threshold, the analysis unit 35 determines that the operation sequence related to the error code is an operation error. On the other hand, if the relevance is less than the predetermined threshold, the analysis unit 35 determines that the operation sequence related to the error code is a normal operation and not an operation error. The threshold may be any value, but for example, if the relevance ranges from 0 to 1, the threshold may be 0.5. When the relevance shown in FIG. 12C is used, when the threshold is 0.5, the relevance between index 1 of the operation sequence and error code 1 is 0.97, which is greater than 0.5, and therefore the analysis unit 35 determines that operation sequence 1 is an operation error. The analysis unit 35 outputs an operation sequence list 42 including the determination result to the input display device 20.
[0064] 13 shows an example of the operation sequence list 42 including the determination result. The operation sequence list 42 includes an index of the operation sequence and a determination result of whether the operation is an erroneous operation or normal operation. The operation sequence list 42 may further include the operation content of the operation sequence and an error code.
[0065] The manipulation error determination unit 22 of the input display device 20, which has acquired the operation sequence list 42 including the determination results, executes a manipulation error determination process. FIG. 14 is a flowchart of the manipulation error determination process. First, it is determined whether or not an operation sequence has been detected (step S501). The procedure for detecting the operation sequence is the same as steps S101 to S103 of the operation sequence creation process shown in FIG. 4. That is, the manipulation error determination unit 22 determines whether or not the time interval between operations is equal to the first time t 1 If this is the case, the series of operations up to that time interval is detected as an operation sequence.
[0066] The manipulation error determination unit 22 waits until it detects an operation sequence (step S501; No). When it detects an operation sequence (step S501; Yes), the manipulation error determination unit 22 refers to the operation sequence list 42 acquired from the information processing device 30, and if it determines that the detected operation sequence is not included in the operation sequence list 42 and is an operation sequence that has not been seen before (step S502; Yes), it outputs an alarm indicating a possible manipulation error (step S504).
[0067] On the other hand, if the operation sequence detected in step S501 is included in the operation sequence list 42 and the same operation sequence has occurred in the past (step S502; No), it is determined whether the operation sequence has been determined to be an operation error (step S503). If the operation sequence has been determined to be an operation error (step S503; Yes), an alarm is output (step S504). For example, if an operation sequence identical to the operation sequence of index 1 is detected in step S501, the operation error determination unit 22 refers to the operation sequence list 42 in FIG. 13 to determine that it is an operation error and outputs an alarm (step S504).
[0068] The method for displaying the alarm is arbitrary, but for example, as shown in FIG. 15, after the first operation (display 1 + screen 1 + button 1) and second operation (display 1 + screen 2 + button 1) that are the same as operation sequence 1 are performed, a message saying "An operation that may be an erroneous operation has been observed" may be displayed.
[0069] If the operation sequence detected in step S501 is the same as the operation sequence that has not been determined to be an operation error (step S503; No), and if there is no end command from the operator of the input display device 20 (step S505; No), the process returns to step S501 and continues. If there is an end command from the operator (step S505; Yes), the process ends.
[0070] As described above, in the analysis system 100 according to this embodiment, the log acquisition unit 31 of the information processing device 30 acquires the operation log 21 and the operation log 11 related to the target equipment 10, and the operation sequence creation unit 32 calculates the time interval between operations as t1 The time interval t 1 An operation sequence is created that indicates a series of operations up to the operation immediately preceding the last operation in the operation sequence. The relevance calculation unit 33 calculates the relevance between the operation sequence and an error that occurred after the final operation in the operation sequence is completed, and the independence determination unit 34 determines the independence between the operation sequence and the error and updates the relevance according to the independence. The analysis unit 35 determines that an operation in the operation sequence is an operation error if the relevance is higher than a threshold value. This makes it possible to analyze operations and errors for a variety of equipment or operations and determine whether an operation is an error, regardless of the type of equipment or operation.
[0071] Second Embodiment A second embodiment of the present disclosure will be described in detail with reference to Figures 16 to 18. Note that identical or corresponding parts in the figures are denoted by the same reference numerals. Figure 16 is a block diagram showing an example of the overall configuration of an analysis system 200 according to the second embodiment. Similar to the first embodiment, the analysis system 200 is a system for preventing erroneous operation of the target equipment 10 and improving the operating status.
[0072] The analysis system 200 includes a target facility 10, an input display device 20, and an information processing device 50, similar to the first embodiment. The configurations of the target facility 10 and the input display device 20 are similar to those of the first embodiment. In addition to the configuration of the information processing device 30 according to the first embodiment, the information processing device 50 includes a learning unit 36, and further includes a storage unit inside or outside the information processing device 50 that stores learning data 51.
[0073] The learning unit 36 performs learning using the learning data 51 in order to optimize at least one of the parameters 43 used in the operation sequence creation unit 32, the relevance calculation unit 33, the independence determination unit 34, and the analysis unit 35 of the information processing device 50. Specifically, the parameter 43 is a parameter 43 obtained by learning the first time t 1 and a second time t after the end of the operation sequence for obtaining the error for which the relevance is to be calculated. 2 The significance level for determining whether or not the operations are independent in the chi-square test for determining independence, and the threshold value of the degree of association for determining whether or not an operation is an operation error in the operation error determination.
[0074] For at least one of these parameters 43, a value that allows more accurate manipulation error determination is found through machine learning by the learning unit 36. Any conventional machine learning method may be used.
[0075] The learning data 51 includes the operation log 21, the operation log 11, and a correct label for the operation error determination. The correct label of the learning data 51 indicates whether or not there is an association between a given operation sequence and a given error. For example, the correct label has a value of "1" if there is an association, and a value of "0" if there is no association. FIG. 17 is a diagram showing examples of correct labels of the learning data 51. The example of FIG. 17 indicates that operation sequence 1 and error 1, and operation sequence 2 and error 2, are associated, and operation sequence 1 and error 2, and operation sequence 2 and error 1 are not associated. The learning unit 36 uses machine learning to determine values of various parameters 43 such that the difference between the value of the correct label (0 or 1) and the operation error determination result determined from the operation log 21 and the operation log 11 is equal to or less than a certain value, and updates the parameters 43 to the obtained values.
[0076] 18 is a flowchart of the learning process executed by the learning unit 36. The following description will be given with reference to the flowchart of FIG.
[0077] First, the learning unit 36 acquires the initial values of the various parameters 43 from the storage unit 40 (step S601). Next, using the parameters 43 acquired in step S601, the learning unit 36 calculates the degree of association from the operation log 21 and the operation log 11 to determine whether an operation error has occurred, and calculates the error between the result and the correct label (step S602).
[0078] The error in step S602 is calculated using, for example, the following formula (3): In formula (3), Value is the error between the manipulation error determination result and the correct label, and when the index of the manipulation sequence is i and the error code is j, the manipulation error determination result for each combination is Result(i, j), and the correct label is Correct(i, j).
[0079]
[0080] If the error value Value calculated by equation (3) is equal to or less than the threshold value (step S603; Yes), the parameter 43 is not updated and the process proceeds to step S605. If the error value Value exceeds the threshold value (step S603; No), the parameter 43 is updated based on the error value Value (step S604). The threshold value is a predetermined value, for example, 0.01.
[0081] Thereafter, if the error calculation has been performed a certain number of times or more (step S605; Yes), the process ends. If the number of error calculations is less than the certain number (step S605; No), the process returns to step S602 and continues. In the next step S602, an operation error determination is performed using the updated parameters 43, and an error calculation with respect to the correct label is performed.
[0082] The parameters 43 can be updated in step S604 using any conventional method, such as a steepest descent method, back propagation, or other methods used in the context of neural networks.
[0083] Various parameters 43 obtained as a result of the machine learning by the learning unit 36 are output to the operation sequence creation unit 32, the relevance calculation unit 33, the independence determination unit 34, and the analysis unit 35, and are used in each process. 1 The relevance calculation unit 33 generates an operation sequence using the second time t 2 The independence determination unit 34 determines the independence using the significance level obtained by learning and updates the degree of association. The analysis unit 35 determines an operation error in the operation sequence using the threshold value of the degree of association obtained by learning, and outputs an operation sequence list 42 including the operation error determination result to the input display device 20.
[0084] As described above, in the analysis system 200 according to this embodiment, the information processing device 50 further includes a learning unit 36, which performs machine learning using learning data 51 including correct labels for manipulation error determination, and determines parameters 43 to be used in creating operation sequences, calculating the degree of association, determining independence, and determining manipulation error, which can output a correct answer for manipulation error determination with a higher probability. The information processing device 50 uses the parameters 43 obtained through learning to create operation sequences, calculate the degree of association, determine independence, and determine manipulation error. This eliminates the need to input parameters required for analysis processing in advance, and enables accurate manipulation error determination regardless of the configuration or specifications of the equipment.
[0085] The hardware configurations and flowcharts shown in the first and second embodiments are merely examples and can be modified and applied as desired. For example, in the first embodiment, the target equipment 10 accumulates the operation log 11, the input display device 20 accumulates the operation log 21, and the information processing device 30 acquires each of the accumulated logs. However, the method of acquiring each log is not limited to this. For example, each time an error occurs in the target equipment 10, the target equipment 10 may transmit error information to the information processing device 30, and the information processing device 30 may accumulate the operation log 11 in the storage unit 40. Furthermore, each time the input display device 20 detects an operation, the input display device 20 may transmit operation information to the information processing device 30, and the information processing device 30 may accumulate the operation log 21 in the storage unit 40.
[0086] Furthermore, in the first embodiment, the information processing device 30 transmits the operation sequence list 42 determined to be an operation error to the input display device 20, and the input display device 20 refers to the list and outputs an alarm if there is a possibility of an operation error, but the alarm output method is not limited to this. For example, the input display device 20 may transmit operation information to the information processing device 30 every time it detects an operation, and the information processing device 30 may determine the possibility of an operation error and send alarm information to the input display device 20. In this case, the input display device 20 outputs the received alarm information.
[0087] Furthermore, in the calculation of the relevance in the first embodiment, the i component of the relevance between the same operation sequence x and the same error (y, i) is added up by the number of occurrences. However, other statistical values such as the average or maximum number of occurrences may also be used as the relevance between the operation sequence x and the error y.
[0088] In the first embodiment, the information processing device 30 directly connected to the target facility 10 and the input display device 20 performs the analysis process, but the configuration of the analysis system is not limited to this. For example, the target facility 10 or the input display device 20 may have the functions of the information processing device 30.
[0089] Alternatively, a server connected to multiple sets of target equipment 10 and input display devices 20 via a network may accumulate the operation logs 11 and the operation logs 21 and perform an analysis process thereon. In this case, the server transmits an operation sequence list 42 including an operation error determination result obtained as a result of the analysis process to the input display device 20, thereby enabling the input display device 20 to determine an operation error. Furthermore, the learning process in the second embodiment may be performed by the server.
[0090] In the first embodiment, the relevance is updated based on the determination result by the independence determination unit 34 for the result of the relevance calculated by the relevance calculation unit 33. However, it is also possible to omit either the relevance calculation unit 33 or the independence determination unit 34. That is, the analysis unit 35 may determine whether the operation sequence is an operation error based on either the result calculated by the relevance calculation unit 33 or the result determined by the independence determination unit 34.
[0091] The allocation of the functions realized by the processor 301 in the information processing devices 30 and 40 in the first and second embodiments is an example, and the allocation may be changed as desired. Furthermore, the functions realized by the processor 301 can be realized by a computer system having a normal communication function, rather than by a dedicated system.
[0092] Furthermore, a computer capable of realizing each function may be configured by distributing a program for executing the operations of the first and second embodiments stored on a computer-readable recording medium such as a CD-ROM (Compact Disc Read-Only Memory), a DVD (Digital Versatile Disc), an MO (Magneto Optical Disc), or a memory card, and installing the program on a computer. When each function is realized by sharing the work between an OS (Operating System) and an application, or by cooperation between the OS and an application, only the parts other than the OS may be stored on the recording medium.
[0093] The present disclosure allows various embodiments and modifications without departing from the broad spirit and scope of the present disclosure. Furthermore, the above-described embodiments are intended to explain the present disclosure and do not limit the scope of the present disclosure. That is, the scope of the present disclosure is defined by the claims, not the embodiments. Various modifications made within the scope of the claims and the meaning of equivalent disclosures are considered to be within the scope of the present disclosure.
[0094] 100, 200 Analysis system, 10 Target equipment, 11 Operation log, 20 Input display device, 21 Operation log, 22 Error operation determination unit, 30, 50 Information processing device, 31 Log acquisition unit, 32 Operation sequence creation unit, 33 Association calculation unit, 34 Independence determination unit, 35 Analysis unit, 36 Learning unit, 40 Memory unit, 41 Association table, 42 Operation sequence list, 43 Parameters, 51 Learning data, 301 Processor, 302 Primary storage device, 303 Secondary storage device, 304 Input unit, 305 Display, 306 Communication interface, 310 Bus.
Claims
1. An information processing device comprising: a log acquisition unit that acquires an operation log including records of multiple operations of a target equipment, and an operation log including records of errors of the target equipment; an operation sequence creation unit that, when a time interval between operations is equal to or longer than a predetermined first time, extracts from the operation log a series of operations up to the operation just before the time interval, and creates an operation sequence; a relevance calculation unit that calculates the relevance between an error included in the operation log that occurs after completion of the last operation of the operation sequence, and the operation sequence; and an analysis unit that outputs an analysis result based on the relevance.
2. The information processing device according to claim 1, wherein the relevance calculation unit calculates the relevance using an index value that decreases with an increase in the time from the end of the last operation in the operation sequence to the occurrence of the error.
3. The information processing device of claim 2, wherein when the operation log includes multiple records in which errors having the same error code occurred within a predetermined second hour from the end of the last operation in the same operation sequence, the correlation calculation unit calculates the sum of the index values calculated for each error as the correlation.
4. The information processing device of claim 2, wherein when the operation log includes multiple records in which errors having the same error code occurred within a predetermined second hour from the end of the last operation in the same operation sequence, the correlation calculation unit calculates the average of the index values calculated for each error as the correlation.
5. An information processing device as described in any one of claims 1 to 4, further comprising an independence judgment unit that judges the independence between the operation sequence created by the operation sequence creation unit and the error included in the operation log acquired by the log acquisition unit, and changes the degree of association between the operation sequence and the error that is judged to be independent to a value lower than the value calculated by the degree of association calculation unit, wherein the analysis unit outputs an analysis result based on the degree of association calculated by the degree of association calculation unit or the degree of association updated by the independence judgment unit.
6. The information processing device according to claim 5, wherein the independence determination unit makes the determination by a chi-square test with a predetermined significance level.
7. An information processing device according to any one of claims 1 to 6, wherein the analysis unit determines that the operation sequence related to the relevance is an operation error when the relevance is equal to or greater than a predetermined threshold, and outputs a list of the operation sequences including the operation error determination result as the analysis result.
8. An information processing device according to any one of claims 1 to 7, further comprising a learning unit that determines at least one of the parameters used in the processing of the operation sequence creation unit, the relevance calculation unit, and the analysis unit through machine learning.
9. An analysis system comprising: an information processing device according to any one of claims 1 to 8; a target equipment communicatively connected to the information processing device and outputting the operation log to the information processing device; and an input display device communicatively connected to the information processing device and the target equipment, accepting operation input to the target equipment, outputting the operation log to the information processing device, and displaying information based on the analysis result output by the analysis unit of the information processing device.
10. The analysis system according to claim 9, wherein when the input display device detects a series of operations identical to the operation sequence in which the relevance calculated by the relevance calculation unit is equal to or greater than a predetermined value, the input display device determines that the operations are erroneous operations and outputs an alarm.
11. The analysis system according to claim 9 or 10, wherein the input display device outputs an alarm when it detects a series of operations that differs from the operation sequence previously created by the operation sequence creation unit.
12. An analysis method comprising: an operation sequence creation step of extracting, from the operation log of a target equipment, a series of operations up to an operation that occurred a predetermined time interval of at least a first hour, and creating an operation sequence; a correlation calculation step of calculating a correlation between an error that occurs after completion of the last operation in the operation sequence, and the operation sequence itself; and an analysis step of performing an analysis based on the correlation.
13. A program for causing a computer to function as: a log acquisition unit that acquires an operation log including a record of operations of a target equipment; an operation sequence creation unit that extracts from the operation log a series of operations up to an operation that occurred a predetermined time interval of at least a first hour, and creates an operation sequence; a relevance calculation unit that calculates the relevance between an error that occurred after completion of the last operation in the operation sequence and the operation sequence itself; and an analysis unit that performs analysis based on the relevance.
Citation Information
Patent Citations
Real-time log control system and method, cloud computing system and server
CN108038049A
Peripheral device
JP2011028355A
Information processor, and information processing system
JP2014085885A
Log information analysis device and log information analysis method
JP2017199126A
Procedure identification device, calculation model generation device, procedure identification method, procedure identification program, calculation model generation method, calculation model generation program, learning data generation device and calculation program
JP2021015321A
