Information processing device, factor analysis method, and factor analysis program

Abstract

To support abnormality cause investigation.SOLUTION: An information processor has: an acquisition part which acquires channel measurement data for each channel to be measured; and a calculation part which calculates an abnormality contribution rate indicating the degree of contribution to an abnormality for each channel based on a score obtained for each parameter from a difference between a parameter extracted from a plurality of pieces of channel measurement data acquired for each channel and a classification border used by a machine learning model for classification into an abnormal or normal class based on the parameter.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an information processing device, a factor analysis method, and a factor analysis program. [Background technology]

[0002] As one of the technologies for detecting equipment abnormalities and their precursors, a diagnostic model is provided that diagnoses the presence or absence of abnormalities based on measurement data obtained by measuring the status of each measurement object, such as equipment used in the equipment for one process, at the end of a batch corresponding to that process in the product manufacturing process. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2009-116427 Summary of the Invention [Problem to be solved by the invention]

[0004] However, the above diagnostic model is limited in its ability to provide information on whether or not an abnormality has occurred in a batch, making it difficult to obtain clues for identifying the cause of an abnormality when it occurs.

[0005] For example, according to the above diagnostic model, when an abnormality occurs, it is unclear which equipment in the facility is causing the abnormality, so on-site workers and others have to start by considering where to start their investigation, which increases the workload of on-site workers and others.

[0006] The present invention aims to assist in identifying the cause of an abnormality. [Means for solving the problem]

[0007] An information processing device according to one aspect of the present invention includes an acquisition unit that acquires channel measurement data for each channel to be measured, and a calculation unit that calculates an abnormality contribution rate indicating the degree to which each channel contributes to an abnormality based on a score obtained for each parameter from the difference between parameters extracted from the plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an abnormal or normal class based on the parameters.

[0008] In a cause analysis method according to one aspect of the present invention, a computer executes a process of acquiring channel measurement data for each channel to be measured, and calculating an abnormality contribution rate indicating the degree to which each channel contributes to an abnormality based on a score obtained for each parameter from the difference between a parameter extracted from the plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an abnormal or normal class based on the parameter.

[0009] A cause analysis program according to one aspect of the present invention causes a computer to execute a process of acquiring channel measurement data for each channel to be measured, and calculating an abnormality contribution rate indicating the degree to which each channel contributes to an abnormality based on a score obtained for each parameter from the difference between a parameter extracted from the plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an abnormal or normal class based on the parameter. [Effects of the Invention]

[0010] According to one embodiment, it is possible to assist in identifying the cause of an abnormality. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a block diagram illustrating an example of a functional configuration of an information processing device. [Figure 2] FIG. 2 is a schematic diagram showing an example of measurement data. [Figure 3]FIG. 3 is a schematic diagram showing an example of the parameter HS. [Figure 4] FIG. 4 is a schematic diagram showing an example of calculation of the abnormality contribution rate. [Figure 5] FIG. 5 is a diagram showing an example of displaying the abnormality contribution rate. [Figure 6] FIG. 6 is a diagram showing an example of displaying the abnormality contribution rate. [Figure 7] FIG. 7 is a diagram showing an example of displaying the abnormality contribution rate. [Figure 8] FIG. 8 is a diagram showing an example of displaying the abnormality contribution rate. [Figure 9] FIG. 9 is a diagram showing a display example of the abnormality contribution rate. [Figure 10] FIG. 10 is a diagram showing an example of the HS trend screen. [Figure 11] FIG. 11 is a flowchart showing the procedure of the abnormality contribution rate calculation process. [Figure 12] FIG. 12 is a flowchart showing the procedure of the output control process. [Figure 13] FIG. 13 is a diagram illustrating an example of a hardware configuration. DETAILED DESCRIPTION OF THE INVENTION

[0012] Hereinafter, embodiments of an information processing device, a factor analysis method, and a factor analysis program according to the present application will be described with reference to the accompanying drawings. Each embodiment merely illustrates one example or aspect, and the range of values, functions, and usage scenarios are not limited by such examples. Furthermore, each embodiment can be appropriately combined within the scope of not causing contradictions in the processing content.

[0013] <Overall structure> Fig. 1 is a block diagram showing an example of the functional configuration of an information processing device 10. The information processing device 10 shown in Fig. 1 provides a factor analysis function for analyzing the cause of an abnormality in order to assist in the investigation of the cause of the abnormality.

[0014] 1, an information processing device 10 can be communicatively connected to a sensor 20. For example, communication conforming to an industrial wireless standard may be performed between the information processing device 10 and the sensor 20. This is merely an example, and the communication performed between the information processing device 10 and the sensor 20 is not limited to a specific communication standard such as an industrial communication standard, and may not be limited to wired or wireless communication.

[0015] The sensor 20 is an example of a measurement device that measures a state of an object. By way of example only, the sensor 20 may be implemented by a measurement device that is incorporated into a control loop, such as measurement, control calculation, and operation.

[0016] For example, one or more sensors 20 may be installed for each measurement target called a "channel." The term "channel" used here refers to a measurement target such as equipment used in a facility for manufacturing a given product, and may include physical quantities such as temperature, pressure, flow rate, pH, speed, acceleration, and valve opening.

[0017] The time series data of the measurement values measured by the sensors 20 installed in this way for each channel to be measured is transmitted from the sensors 20 to the information processing device 10. Hereinafter, the time series data of the measurement values for one channel may be referred to as "channel measurement data." Furthermore, the channel measurement data for N channels, channel ch1 to channel chN (natural number), may be referred to as "measurement data."

[0018] It should be noted that it is not necessary for one sensor 20 to be installed for one channel, and a plurality of channel measurement data corresponding to a plurality of channels may be measured by one sensor 20.

[0019] The information processing device 10 is an example of a computer that provides the above-mentioned factor analysis function. As a mere example, the above-mentioned factor analysis function is provided as a function of a recording device, a so-called recorder, that records the above-mentioned measurement data, but is not limited to this. For example, the information processing device 10 may be realized as a server that provides the above-mentioned factor analysis function on-premise. Alternatively, the information processing device 10 may be realized as a PaaS (Platform as a Service) or SaaS (Software as a Service) application, thereby providing the above-mentioned factor analysis function as a cloud service.

[0020] <Configuration of information processing device 10> Next, an example of the functional configuration of the information processing device 10 according to this embodiment will be described. Fig. 1 shows a block diagram related to the factor analysis function of the information processing device 10. As shown in Fig. 1, the information processing device 10 has a display input unit 11, a communication control unit 12, a storage unit 13, and a control unit 15. Note that Fig. 1 only shows a selection of functional units related to the above-mentioned factor analysis function, and the information processing device 10 may also be provided with functional units other than those shown.

[0021] The display input unit 11 is a functional unit that inputs various operations and displays various information. As one example, the display input unit 11 can be realized by a touch panel in which an input device and a display device are integrated. This is merely one example, and the input function of various operations and the display function of various information do not necessarily have to be integrated and implemented; the input unit and the display unit may each be provided separately.

[0022] The communication control unit 12 is a functional unit that controls communication with other devices such as the sensor 20. As just one example, the communication control unit 12 may be realized by a network interface card. As one aspect, the communication control unit 12 can receive channel measurement data from the sensor 20. Note that bidirectional communication does not necessarily have to be performed between the information processing device 10 and the sensor 20, and serial communication from the sensor 20 to the information processing device 10 may be performed.

[0023] The storage unit 13 is a functional unit that stores various types of data. As just one example, the storage unit 13 is realized by internal, external, or auxiliary storage of the information processing device 10. For example, the storage unit 13 stores a measurement data log 13A, a diagnostic model 13B, an HS (Health Score) log 13C, and an abnormality contribution rate log 13D. The measurement data log 13A, the diagnostic model 13B, the HS log 13C, and the abnormality contribution rate log 13D will be described together when reference, generation, or registration is performed.

[0024] The control unit 15 is a functional unit that performs overall control of the information processing device 10. For example, the control unit 15 can be realized by a hardware processor. As shown in FIG. 1, the control unit 15 has a measurement data acquisition unit 15A, an abnormality determination unit 15B, an abnormality contribution rate calculation unit 15C, and an output control unit 15D. Note that the control unit 15 may also be realized by hardwired logic or the like.

[0025] The measurement data acquiring unit 15A is a processing unit that acquires measurement data. As just one example, the measurement data acquiring unit 15A acquires channel measurement data from each of the sensors 20 corresponding to the channel, for each channel to be measured by a device or the like used in the equipment corresponding to the process, in batch units corresponding to one process of the product manufacturing process.

[0026] The term "batch" here refers to a section corresponding to one step in the manufacturing process of a product, and can be defined, for example, by its beginning and end in time. For example, in the tire manufacturing process, the section between the beginning and end of the vulcanization process to increase the strength of the tire is called a batch.

[0027] For example, at the end of a batch, the measurement data acquiring unit 15A acquires channel measurement data for a section corresponding to the batch from each of the sensors 20 corresponding to each channel. At this time, the measurement data acquiring unit 15A can transmit measurement values from each sensor 20 in real time and accumulate measurement values for the section corresponding to the batch, or can transmit all of the channel measurement data for the section corresponding to the batch at the end of the batch.

[0028] After the channel measurement data for N channels has been acquired in this manner, the channel measurement data for N channels is compiled into a single data file as measurement data and then added to and saved in the measurement data log 13A stored in the storage unit 13. As a result, measurement data including the channel measurement data for N channels is saved in batches as the measurement data log 13A in the storage unit 13. When saving in the storage unit 13 in this manner, the measurement data acquisition unit 15A can save the channel measurement data in association with names previously assigned to the channels, such as channel identification information or tag names.

[0029] The abnormality determination unit 15B is a processing unit that determines whether or not there is an abnormality in a batch using the diagnostic model 13B. Here, the diagnostic model 13B is realized by a trained machine learning model, merely as an example. Hereinafter, a support vector machine, a so-called SVM (Support Vector Machine), will be exemplified as an example of the diagnostic model 13B, but the diagnostic model may also be realized by other machine learning models such as a neural network.

[0030] More specifically, when measurement data is acquired by the measurement data acquisition unit 15A, the abnormality determination unit 15B extracts one or more parameters from the measurement data for each channel measurement data included in the measurement data. FIG. 2 is a schematic diagram showing an example of measurement data. In FIG. 2, one batch of measurement data including channels ch1 to chN is illustrated enclosed in a solid-line frame. As shown in FIG. 2, one measurement data includes channel measurement data for N channels, and M (a natural number) parameters are extracted for each channel measurement data. Any method, such as feature selection or feature extraction, may be applied to extract such parameters. As an example, by applying statistical processing, such as averaging or variance processing, to one batch of channel measurement data, it is possible to extract the mean, variance, or other features corresponding to one batch of channel measurement data. In this way, M parameters are extracted for one channel.

[0031] Thereafter, the anomaly determination unit 15B inputs the N channel-specific and M parameter-specific parameters to the diagnostic model 13B read from the storage unit 13. The diagnostic model 13B to which the parameters have been input in this manner uses the N×M parameters as an input vector and outputs, as HS, the distance between the input vector and the classification boundary CB of the hyperplane that classifies the presence or absence of an anomaly in the batch. At this time, if HS output by the diagnostic model 13B is equal to or greater than zero, the anomaly determination unit 15B determines that the batch is not abnormal, whereas if HS is less than zero, it determines that the batch is abnormal.

[0032] Such a diagnostic model 13B can be generated by the following training. For example, machine learning is performed using N×M parameters as training samples and a set of training data labeled with a correct class, for example, "normal" or "abnormal," as a data set. For example, in the case of SVM, parameters of a discriminant function that maximizes the distance between a support vector and the classification boundary CB, i.e., the so-called margin, are trained. The "support vector" referred to here corresponds to the feature vector of the training data located near the boundary between the set of training data labeled with the normal class and the set of training data labeled with the abnormal class. A classifier that classifies the normal or abnormal class depending on whether the sign of the value output by such a discriminant function is positive or negative is stored in the storage unit 13 as the diagnostic model 13B.

[0033] In addition to the above HS, the diagnostic model 13B can output parameters HS for each type of parameter. Fig. 3 is a schematic diagram showing an example of the parameters HS. For convenience of explanation, Fig. 3 shows a two-dimensional feature space in which the number of parameter types M is "2" consisting of parameters 1 and 2, and also shows a classification boundary CB used by the diagnostic model 13B within the feature space.

[0034] 3, when an input vector V1 is input, the diagnostic model 13B outputs 1.0 as HS. When an input vector V2 is input, the diagnostic model 13B outputs 0.1 as HS. When an input vector V3 is input, the diagnostic model 13B outputs −1.5 as HS. When an input vector V4 is input, the diagnostic model 13B outputs −0.1 as HS. For these input vectors V1 to V4, it can be determined that there is no abnormality in the batch corresponding to the input vector V1 and the batch corresponding to the input vector V2, and that there is an abnormality in the batch corresponding to the input vector V3 and the batch corresponding to the input vector V4.

[0035] In addition to such HS, the diagnostic model 13B calculates the distance for each type of parameter that forms the distance between the feature vector corresponding to one or more parameters and the classification boundary CB. As just one example, the diagnostic model 13B can output, for each evaluation axis corresponding to the type of parameter, the distance of the component corresponding to the evaluation axis in HS, which is the distance between the input vector and the classification boundary CB, as the parameter HS. For example, in the case of the input vector V3, the diagnostic model 13B calculates the distance of the component corresponding to the parameter type "parameter 1" as the parameter HS. p1 Furthermore, the diagnostic model 13B can output the distance of the component corresponding to the parameter type “parameter 2” as the parameter HS p2 In this way, the diagnostic model 13B can output the parameters HS p1~pM can be output.

[0036] In this way, the HS and parameters HS output by the diagnostic model 13B are added to and saved in the HS log 13C stored in the storage unit 13. As a result, the storage unit 13 saves the HS of one entire batch and the parameters HS for each channel and each parameter type as the HS log 13C for each batch.

[0037] The abnormality contribution rate calculation unit 15C is a processing unit that calculates the degree to which each channel contributes to an abnormality for each channel. As just one example, the abnormality contribution rate calculation unit 15C can calculate the abnormality contribution rate of each channel based on an integrated value obtained by integrating the parameter HS of each channel.

[0038] FIG. 4 is a schematic diagram showing an example of calculating the abnormality contribution rate. FIG. 4 shows, as an example only, a table of the calculation results of the parameters Hs when the number of channels N is 4 and the number of types of parameters M extracted per channel is 5. In the example shown in FIG. 4, the abnormality contribution rate calculation unit 15C adds up the five parameters Hs of the channel ch1, ie, parameters 1 to 5. That is, the abnormality contribution rate of the channel ch1 is calculated as "0.505536" by calculating "0.01224 + 0.22005 - 0.043275 - 0.037099 + 0.35362." Similarly, the abnormality contribution rate calculation unit 15C adds up the five parameters Hs of the channel ch2 to calculate the abnormality contribution rate of the channel ch2 as "0.463747." Furthermore, the abnormality contribution rate calculation unit 15C calculates the abnormality contribution rate of channel ch3 as "-0.37696" and the abnormality contribution rate of channel ch4 as "-0.33223." For example, in the example of the abnormality contribution rates shown in Fig. 4, the smaller the value of the abnormality contribution rate, in other words, the more negative the sign of the abnormality contribution rate and the larger the absolute value, the higher the degree to which the channel contributes to the abnormality.

[0039] The abnormality contribution rate calculated for each channel in this manner is added to and stored in the abnormality contribution rate log 13D stored in the storage unit 13. As a result, the abnormality contribution rate for each channel is stored in the storage unit 13 as the abnormality contribution rate log 13D in batch units.

[0040] Although an example has been given here in which the integrated value obtained by integrating the parameter HS of each channel is calculated as the abnormality contribution rate of each channel, the abnormality contribution rate may be calculated by other calculation methods, such as a value obtained by further processing the integrated value, without being limited to the integrated value. For example, the abnormality contribution rate may be normalized so that the value increases as the degree of contribution to the abnormality increases, or may be normalized so that the value falls within a specific numerical range, for example, 0 to 1 or 0 to 100.

[0041] The output control unit 15D is a processing unit that executes various output controls. As just one example, the output control unit 15D controls the output to the display input unit 11. Here, display output is taken as an example of the output controlled by the output control unit 15D, but it goes without saying that other outputs such as print output and audio output may also be controlled.

[0042] In one aspect, when the output control unit 15D receives a request to display the abnormality contribution rate via the display input unit 11, it displays the abnormality contribution rate by channel for the batch specified in the display request from the abnormality contribution rates included in the abnormality contribution rate log 13D.

[0043] As an example, the output control unit 15D can associate and display the abnormality contribution rate of each channel with identification information of the channel associated with the channel, for example, with the channel number. FIG. 5 is a diagram showing an example of a display of the abnormality contribution rate. FIG. 5 shows an example of a display in which the abnormality contribution rate shown in FIG. 4 is displayed. In the example shown in FIG. 5, for each of channel numbers "0001" to "0004" corresponding to channels ch1 to chN, a bar graph corresponding to the numerical value of the abnormality contribution rate is displayed in association with the channel number. With this display, the abnormality contribution rate can be presented in a manner that allows field workers and the like to identify the channels.

[0044] As another example, the output control unit 15D can associate tag information associated with a channel, such as the abnormality contribution rate of each tag name, and display the associated tag information. FIGS. 6 and 7 are diagrams illustrating display examples of the abnormality contribution rates. FIGS. 6 and 7 also illustrate display examples of the abnormality contribution rates shown in FIG. 4. In the example illustrated in FIG. 6, for each of the tag names "Temperature" to "Flow Rate" corresponding to channels ch1 to chN, the numerical values of the abnormality contribution rates corresponding to the tag names are displayed in association with a bar graph corresponding to the numerical values. When displaying the numerical values of the abnormality contribution rates and the corresponding bar graphs for each channel in this manner, the output control unit 15D can display the numerical values of the abnormality contribution rates outside the bar graph frame as shown in FIG. 6, or can display the numerical values of the abnormality contribution rates within the bar graph frame as shown in FIG. 7. These tag names can be displayed as any character string based on user settings by a related person, such as a field worker or an operator. While FIG. 7 illustrates an example in which the numerical values of the abnormality contribution rates are displayed outside the bars within the bar graph frame, the numerical values of the abnormality contribution rates may also be displayed inside the bars. Although FIGS. 6 and 7 show examples in which character strings are displayed as tag names, icons or the like assigned to the tags may also be displayed.

[0045] As another example, the output control unit 15D can sort the abnormality contribution rates by channel in ascending or descending order and display the abnormality contribution rates in the order of the sorted channels. FIG. 8 is a diagram showing an example of how the abnormality contribution rates are displayed. FIG. 8 also shows an example of how the abnormality contribution rates shown in FIG. 4 are displayed. In the example shown in FIG. 8, the abnormality contribution rates are displayed in ascending order of the abnormality contribution rate, that is, in ascending order. That is, by sorting the abnormality contribution rates "0.505536," "0.463747," "-0.37696," and "-0.33223" of the four channels ch1 to ch4 in ascending order, the channels are sorted in the order of channel ch3, channel ch4, channel ch2, and channel ch1. In this sorted order, i.e., according to the tag names "displacement volume," "flow rate," "pressure," and "temperature," the abnormality contribution rates "-0.37696," "-0.33223," "0.463747," and "0.505536" corresponding to the respective tag names are displayed. This display allows field workers and the like to quickly grasp which channels have a high or low degree of contribution to the abnormality.

[0046] As another example, the output control unit 15D can change the display format of a channel whose abnormality contribution rate satisfies a specific condition among the abnormality contribution rates for each channel to a different display format from that of other channels. FIG. 9 is a diagram illustrating an example of displaying the abnormality contribution rates. FIG. 9 also illustrates a display example of the abnormality contribution rates shown in FIG. 4. Furthermore, FIG. 9 illustrates an example in which a channel whose abnormality contribution rate is below a threshold, e.g., zero, is displayed in a different display format from a channel whose abnormality contribution rate exceeds zero. As shown in FIG. 9, among the four channels, the tag names "Displacement Volume" and "Flow Rate" whose abnormality contribution rate is below zero are displayed in a different display format from the tag names "Pressure" and "Temperature" whose abnormality contribution rate exceeds zero. That is, by changing the hatching of the bar graph showing the abnormality contribution rate, the abnormality contribution rate of tag names whose abnormality contribution rate is below zero is highlighted. This display also allows field workers and others to quickly identify channels that contribute highly to abnormalities. Note that while Fig. 9 shows an example of changing the hatching of the bar graph, it is also possible to highlight the anomaly contribution rate of channels that have a high degree of contribution to the anomaly by changing the color of the bar graph or the font of the anomaly contribution rate value.

[0047] Here, a display request for the anomaly contribution rate can be received on a GUI (Graphical User Interface) related to the HS. FIG. 10 is a diagram showing an example of an HS trend screen. As shown in FIG. 10, the HS trend screen 30 displays the chronological transition of the HS. That is, on the HS trend screen 30, each batch is diagrammed into a block and displayed in chronological order, and the HS value for each batch is displayed within the block (within a square frame). A block selection operation on this HS trend screen 30 can be received as the above-mentioned display request. For example, when a selection operation of the block corresponding to the batch on March 9, 2022 (a square surrounding -0.8) is received, the channel-by-channel anomaly contribution rate corresponding to the batch on March 9, 2022, among the anomaly contribution rates included in the anomaly contribution rate log 13D, can be displayed. By displaying the anomaly contribution rate using such a guiding line, it is possible to confirm how the anomaly contribution rate changes over time.

[0048] In another aspect, the output control unit 15D can perform pull-type information provision that displays the abnormality contribution rate in response to a display request, as described above, and can also perform push notification that automatically notifies the user from a program that realizes the above-mentioned factor analysis function. For example, the output control unit 15D can display the abnormality contribution rate for each channel when the HS included in the HS log 13C satisfies a specific condition.

[0049] As an example, when the HS log 13C is updated, the output control unit 15D determines whether the latest HS is equal to or less than a threshold value Th1, for example, "1." If the latest HS is equal to or less than the threshold value Th1, the output control unit 15D calculates an approximate line corresponding to the distribution of past HS history, for example, a specific number of HSs from the latest, by regression analysis. If the slope a of the approximate line calculated in this way is equal to or less than a threshold value Th2, for example, -1, the output control unit 15D displays the latest anomaly contribution rate among the anomaly contribution rates included in the HS log 13C by channel. This makes it possible to provide information that contributes to the investigation of the cause of not only when an anomaly has occurred but also when there is a sign of an anomaly. Here, a sign of an anomaly occurrence refers, for example, to a state in which the latest HS is between 0 and 1 and no anomaly has occurred, but the slope of the approximate line is equal to or less than -1 and is on a decreasing trend. This makes it possible to predict that the future HS will be equal to or less than 0 and an anomaly will occur. Although the example in which the threshold value Th1 and the threshold value Th2 are used as constraints imposed on the display of the abnormality contribution rate has been given here, only one of them may be imposed as a constraint.

[0050] <Processing flow> Next, a flow of processing by the information processing device 10 according to this embodiment will be described. Here, (1) Anomaly contribution rate calculation processing executed by the information processing device 10 will be described first, followed by (2) Output control processing.

[0051] (1) Calculation of abnormality contribution rate 11 is a flowchart showing the procedure of the anomaly contribution rate calculation process. This process can be executed at the end of a batch, as an example. As shown in FIG. 11, the measurement data acquisition unit 15A acquires channel measurement data for each of N channels to be measured, thereby acquiring measurement data for one batch (step S101).

[0052] Subsequently, loop processing 1 and loop processing 2 are executed to repeat the processing of step S102 described below a number of times corresponding to the N channels acquired in step S101 and a number of times corresponding to the number M of types of parameters to be input to the diagnostic model 13B for each of the N channels. Note that, although Fig. 11 shows an example in which the processing of step S102 is repeatedly executed, the processing of step S102 does not necessarily have to be executed serially, and may be executed in parallel for each of the N channels and M types.

[0053] That is, the abnormality determination unit 15B extracts a parameter corresponding to the parameter type j from the channel measurement data of the channel i (step S102).

[0054] Step S102 is executed repeatedly, with the initial value of index j set to 1, until it is incremented to M. By this loop process 2, M types of parameters can be extracted for one channel. Note that although an example in which M types of parameters are extracted for one channel has been given here, the number of types of parameters for each channel i does not necessarily have to be the same, and may be different.

[0055] Furthermore, step S102 is repeated, with the index i initially set to 1, until it is incremented to N. By this loop process 1, M types of parameters can be extracted for each of N channels. Note that, although an example has been given in which the number of channels N and the number of parameter types M are plural, both the number of channels N and the number of parameter types M may be 1, or only one of them may be 1.

[0056] Thereafter, the abnormality determination unit 15B inputs the parameters for the N channels and the M parameter types to the diagnostic model 13B read from the storage unit 13 (step S103). Then, the abnormality contribution rate calculation unit 15C acquires the parameters HS output by the diagnostic model 13B for the N channels and the M parameter types (step S104).

[0057] Then, loop process 3 is executed in which the process of step S105 below is repeated a number of times corresponding to the number of channels N. Note that, although Fig. 11 shows an example in which the process of step S105 is repeatedly executed, the process of step S105 does not necessarily have to be executed serially, and may be executed in parallel for each of the N channels.

[0058] That is, the abnormality contribution rate calculation unit 15C calculates the abnormality contribution rate of the channel i based on the integrated value obtained by integrating the parameter HS of the channel i (step S105).

[0059] Such step S105 is repeatedly executed with the initial value of index i set to 1 until it is incremented to N. By such loop processing 3, the abnormality contribution rate can be calculated for each of N channels.

[0060] Thereafter, the abnormality contribution rate calculation unit 15C adds and stores the abnormality contribution rates calculated for each of the N channels in the abnormality contribution rate log 13D stored in the storage unit 13 (step S106), and ends the process.

[0061] (2) Output control processing 12 is a flowchart showing the procedure of the output control process. This process is merely an example and can be repeatedly executed while the information processing device 10 is powered on.

[0062] 12, when a request to display the abnormality contribution rate is accepted (Yes in step S301), the output control unit 15D executes the following process: The output control unit 15D causes the display input unit 11 to display, among the abnormality contribution rates included in the abnormality contribution rate log 13D, the abnormality contribution rate by channel for the batch specified in the display request (step S302), and proceeds to step S301.

[0063] On the other hand, if the request to display the abnormality contribution rate has not been accepted (No at step S301), the output control unit 15D determines whether the HS log 13C has been updated (step S303).

[0064] At this time, if the HS log 13C is updated (Yes in step S303), the output control unit 15D determines whether the latest HS is equal to or less than a threshold value Th1, for example, "1" (step S304).

[0065] If the latest HS is equal to or less than the threshold value Th1 (Yes in step S304), the output control unit 15D calculates an approximate straight line corresponding to the history of past HS, for example, the distribution of HS up to a specific number from the latest, by regression analysis (step S305).

[0066] If the slope a of the approximate line calculated in this way is equal to or smaller than the threshold value Th2, for example, −1 (Yes in step S306), the output control unit 15D executes the following process: The output control unit 15D displays the latest abnormality contribution rates for each channel among the abnormality contribution rates included in the HS log 13C (step S307), and proceeds to step S301.

[0067] <One aspect of the effect> As described above, the information processing device 10 according to this embodiment provides a factor analysis function that calculates the anomaly contribution rate for each channel. By presenting the anomaly contribution rate for each channel calculated by this factor analysis function, it is possible to grasp at a glance which channel is causing the anomaly. Therefore, when an anomaly occurs and it is unclear where to start investigating, it becomes clear which channel or areas should be prioritized for investigation. In other words, this can serve as an indicator for investigation when an anomaly occurs, and can reduce the man-hours required for the investigation.

[0068] <Numbers, etc.> The matters described in the above embodiment, such as the number of channels and sensors, the number of parameters and the method for extracting them, and the specific examples of the method for calculating HS and the parameters HS, are merely examples and can be changed. Also, the processing order of the flowcharts described in the embodiment can be changed within a consistent range.

[0069] <System> The information including the processing procedures, control procedures, specific names, various data, and parameters shown in the above documents and drawings can be changed as desired unless otherwise specified. For example, any one or more of the functional units among the measurement data acquisition unit 15A, the abnormality determination unit 15B, the abnormality contribution rate calculation unit 15C, and the output control unit 15D may be configured as separate devices.

[0070] Furthermore, the components of each device shown in the figure are functional concepts and do not necessarily have to be physically configured as shown. In other words, the specific form of distribution and integration of each device is not limited to that shown. In other words, all or part of them can be functionally or physically distributed and integrated in any unit depending on various loads, usage conditions, etc. Note that each configuration may also be a physical configuration.

[0071] Furthermore, each processing function performed by each device can be realized, in whole or in part, by a CPU (Central Processing Unit) and a program analyzed and executed by the CPU, or can be realized as hardware using wired logic.

[0072] <Hardware> Next, an example of the hardware configuration of the computer described in the embodiment will be described. Fig. 13 is a diagram illustrating an example of the hardware configuration. As shown in Fig. 13, an information processing device 10 includes a communication device 10a, an HDD (Hard Disk Drive) 10b, a memory 10c, and a processor 10d. The components shown in Fig. 13 are connected to each other via a bus or the like.

[0073] The communication device 10a is a network interface card or the like, and communicates with other servers. The HDD 10b stores programs and DBs that operate the functions shown in FIG.

[0074] Processor 10d reads a program that executes the same processing as the processing unit shown in Fig. 1 from HDD 100b or the like and loads it into memory 100c, thereby operating a process that executes the functions described in Fig. 1 or the like. For example, this process executes the same functions as the processing unit of information processing device 10. Specifically, processor 10d reads a program that has the same functions as measurement data acquisition unit 15A, abnormality determination unit 15B, abnormality contribution rate calculation unit 15C, output control unit 15D, etc. from HDD 10b or the like. Then, processor 10d executes a process that executes the same processing as measurement data acquisition unit 15A, abnormality determination unit 15B, abnormality contribution rate calculation unit 15C, output control unit 15D, etc.

[0075] In this way, the information processing device 10 operates as an information processing device that executes a factor analysis method by reading and executing a program. The information processing device 10 can also realize functions similar to those of the above-described embodiment by reading the program from a recording medium using a medium reading device and executing the read program. Note that the program in these other embodiments is not limited to being executed by the information processing device 10. For example, the present invention can also be applied in the same way to cases where another computer or server executes the program, or where these execute the program in cooperation with each other.

[0076] The above program can be distributed via a network such as the Internet. The above program can also be recorded on any recording medium and executed by a computer by reading it from the recording medium. For example, the recording medium can be a hard disk, a flexible disk (FD), a CD-ROM, a magneto-optical disk (MO), a digital versatile disk (DVD), or the like. [Explanation of symbols]

[0077] 10. Information processing equipment 11 Display and input section 12 Communication control section 13 Storage section 13A Measurement data log 13B Diagnostic Model 13C HS Log 13D Anomaly Contribution Log 15 Control Unit 15A Measurement data acquisition section 15B Abnormality judgment section 15C Abnormality contribution calculation section 15D Output control section 20 sensors

Claims

1. An acquisition unit that acquires channel measurement data for each channel to be measured; a calculation unit that calculates an anomaly contribution rate indicating the degree to which each channel contributes to an anomaly based on a score obtained for each parameter from the difference between a parameter extracted from a plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an anomaly or normal class based on the parameter; and and the calculation unit calculates the anomaly contribution rate for each channel based on an integrated value obtained by integrating scores calculated for each parameter extracted from channel measurement data corresponding to the channel; Information processing device.

2. the calculation unit calculates the score for each parameter by calculating a distance for each type of parameter that forms a distance between the feature vector corresponding to the parameter and the classification boundary; The information processing device according to claim 1 .

3. an output control unit that displays the abnormality contribution rate for each channel on a display unit; The information processing device according to claim 1 .

4. the output control unit associates the abnormality contribution rate of the channel with each piece of channel identification information associated with the channel, and causes the display unit to display the associated information. The information processing device according to claim 3 .

5. the output control unit associates the abnormality contribution rate of the channel with each piece of tag information associated with the channel and causes the display unit to display the associated information. The information processing device according to claim 3 .

6. the output control unit causes the display unit to display the abnormality contribution rates in the order of channels sorted in ascending or descending order of the abnormality contribution rates. The information processing device according to claim 3 .

7. the output control unit changes the display format of the channel whose abnormality contribution rate satisfies a specific condition to a display format different from the display format of other channels and displays the changed format on the display unit. The information processing device according to claim 3 .

8. Acquire channel measurement data for each channel to be measured; Calculating an anomaly contribution rate indicating the degree to which each channel contributes to an anomaly based on a score obtained for each parameter from the difference between a parameter extracted from the plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an anomaly or normal class based on the parameter; The computer executes the processing, the calculating process includes a process of calculating the anomaly contribution rate for each channel based on an integrated value obtained by integrating scores calculated for each parameter extracted from channel measurement data corresponding to the channel. Factor analysis methods.

9. Acquire channel measurement data for each channel to be measured; Calculating an anomaly contribution rate indicating the degree to which each channel contributes to an anomaly based on a score obtained for each parameter from the difference between a parameter extracted from the plurality of channel measurement data acquired for each channel and a classification boundary used by a machine learning model that classifies data into an anomaly or normal class based on the parameter; Have the computer execute the process, the calculating process includes a process of calculating the anomaly contribution rate for each channel based on an integrated value obtained by integrating scores calculated for each parameter extracted from channel measurement data corresponding to the channel. Factor analysis program.

Images