Information processing device, information processing method, and program

Abstract

To provide an information processing apparatus, an information processing method, and a program capable of more appropriately performing detection of a specific state of a monitoring target such as signs of an abnormality.SOLUTION: The information processing apparatus is provided with a candidate generation unit and a correction unit. The candidate generation unit generates a frequently occurring transition path around a state as a candidate for a first path, which is a transition path assumed to be a first specific state using first log information indicating the state of the monitoring object obtained within a certain period. The correction unit corrects the candidate by using a predetermined reference path and obtains the first path.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] An embodiment of the present invention relates to an information processing device, an information processing method, and a program. [Background technology]

[0002] Conveying devices for conveying media, such as automatic ticket gates, postal sorters, banknote sorters, multifunction peripherals (MFPs), and scanners, are known. Media include, for example, train tickets, mail, banknotes, and printed matter. Rollers for conveying media are incorporated as key components in such conveying devices.

[0003] When the rollers deteriorate, not only will they be unable to transport the media, but the media itself may be damaged. For this reason, abnormality detection is implemented to detect signs of abnormalities in each part of the transport device, including the rollers, and enable maintenance of the device. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Patent No. 2703414 [Patent Document 2] Japanese Patent Application Laid-Open No. 2007-217103 [Non-patent literature]

[0005] [Non-Patent Document 1] PrefixSpan: Mining Sequential Patterns by Prefix-Projected Growth. Jian Pei, Jiawei Han, Behzad Mortazavi-Asl, Helen Pinto, Qiming Chen, Umeshwar Dayal, Meichun Hsu. In Proc. of the 17th International Conference on Data Engineering, 2001. Summary of the Invention [Problem to be solved by the invention]

[0006] An object of the present invention is to provide an information processing device, an information processing method, and a program that can more appropriately detect a specific state of a monitored object, such as a sign of an abnormality. [Means for solving the problem]

[0007] According to an embodiment, an information processing apparatus includes a candidate generator and a corrector. The candidate generator uses first log information indicating a state of a monitoring target obtained within a certain period of time to generate frequently occurring transition paths for the state as candidates for a first path, which is a transition path assumed to be a first specific state. The corrector corrects the candidates using a predetermined reference path to obtain the first path. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram showing an example of the configuration of a transport device that is a target for abnormality sign detection. [Figure 2] FIG. 1 is a block diagram of an information processing system according to a first embodiment. [Figure 3] FIG. 4 is a diagram showing an example of the data structure of an operation log. [Figure 4] 10 is a flowchart of a learning process according to an embodiment. [Figure 5] 10 is a flowchart of abnormality sign detection according to an embodiment. [Figure 6] FIG. 10 is a diagram showing an example of a warning screen. [Figure 7] FIG. 10 is a block diagram of an information processing system according to a second embodiment. [Figure 8] FIG. 10 is a block diagram of an information processing system according to a third embodiment. [Figure 9] FIG. 1 is a hardware configuration diagram of an information processing system according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] A preferred embodiment of an information processing system (an example of an information processing device) according to the present invention will be described in detail below with reference to the accompanying drawings. Hereinafter, an example will be described in which the system is applied to a system in which a transport device that transports a medium is the monitored object and a specific state of the monitored object is detected, but the applicable system is not limited to this. Any device other than a transport device may be the monitored object. Furthermore, the specific state (second specific state) of the monitored object is, for example, an abnormal state, but is not limited to this.

[0010] As mentioned above, parts such as rollers in a conveying device deteriorate with use. Therefore, proper roller maintenance, such as replacement and cleaning, is required for long-term use. Furthermore, conveying control can involve complex processes. For example, even when there is a risk of a conveying abnormality occurring, the system can minimize the occurrence of abnormalities by controlling the conveying process, such as by slightly reversing the media during processing and then conveying it again. In abnormality detection, it is necessary to distinguish such complex processes from abnormalities.

[0011] Methods for detecting signs of anomalies include detecting anomalies based on state transition diagrams in design documents and methods that use machine learning models. For example, methods based on state transition diagrams can calculate an anomaly score that indicates the severity of the anomaly, or can detect anomalies by focusing only on limited causes of the anomaly. Methods that use high-performance machine learning models, such as deep learning, can treat internal processing as a black box, making it difficult to determine the basis for the detection results. In other words, regardless of which method is used, it can be difficult to detect anomalies in the behavior of the entire system while still providing evidence for the detection results.

[0012] The information processing system according to the following embodiment detects anomalies at an early stage of signs and quantifies them as signs scores. The information processing system according to the embodiment also constructs an anomaly sign detection model that detects sequential patterns of control events and abnormal events that serve as the basis for the signs, and uses the anomaly sign detection model to detect anomaly signs based on the degree of deviation from the sequential patterns under normal conditions.

[0013] 1 is a diagram showing an example of the configuration of a conveyance device 10 that is a target (monitoring target) for abnormality sign detection. As shown in Fig. 1, the conveyance device 10 includes medium passage sensors 11-1 to 11-4, conveyance rollers 12-1 to 12-12, and conveyance belts 13-1 to 13-6.

[0014] The arrows indicate the direction in which the media are transported. Fig. 1 shows an example in which the media are transported from left to right. Conveyor belts 13-1 to 13-6 transport the media. Fig. 1 shows, as examples of media, media 21-1 and 21-2 immediately after being inserted into an inlet, and media 22-1 and 22-2 ejected by conveyor belts 13-5 and 13-6, which are arranged on the side from which the media are ejected (the right side of Fig. 1).

[0015] The transport rollers 12-1 to 12-12 are rollers for rotating the transport belts 13-1 to 13-6, which will be described below. Conveyor rollers 12-1, 12-2: Conveyor belt 13-1 Conveyor rollers 12-3, 12-4: Conveyor belt 13-2 Conveyor rollers 12-5, 12-6: Conveyor belt 13-3 Conveyor rollers 12-7, 12-8: Conveyor belt 13-4 Conveyor rollers 12-9, 12-10: Conveyor belt 13-5 Conveyor rollers 12-11, 12-12: Conveyor belt 13-6

[0016] Media passing sensors 11-1 to 11-4 detect the passage of media at different positions on the media transport path. For example, media passing sensor 11-1 detects the passage of a medium inserted into the input port. Media passing sensor 11-2 detects the passage of a medium transported from conveyor belts 13-1 and 13-2 toward conveyor belts 13-3 and 13-4. Media passing sensor 11-3 detects the passage of a medium transported from conveyor belts 13-3 and 13-4 toward conveyor belts 13-5 and 13-6. Media passing sensor 11-4 detects the passage of a medium transported (discharged) from conveyor belts 13-5 and 13-6.

[0017] Each media passage sensor 11-n (n is 1, 2, 3, or 4) is realized by a photointerrupter or the like including a light-emitting unit 11-na and a light-receiving unit 11-nb. The media passage sensor 11-n detects that a medium has passed when, for example, the light emitted from the light-emitting unit 11-na is not received by the light-receiving unit 11-nb.

[0018] The medium passing sensor 11-n stores information indicating whether a medium is passing or not as a log (operation log) at regular intervals (for example, every second). The operation log corresponds to log information (first log information) indicating the status of the monitored object.

[0019] Conveying device 10 may also have the function of separating and conveying stacked media. In this case, conveying device 10 may further include another sensor that detects overlapping media. When such a sensor detects that two stacked media (media 21-1 and 21-2 in FIG. 1) have been inserted into the inlet, conveying device 10 reverses conveyor belt 13-3 to separate upper medium 21-1 from lower medium 21-2 when media 21-1 and 21-2 reach media passage sensor 11-2, and then separates the two media one by one and ejects them using conveyor belts 13-5 and 13-6.

[0020] The configuration of the conveying device 10 in FIG. 1 is an example. Any other configuration of conveying device may be used as long as at least a path (an example of a transition path) representing the transition of the position through which the medium passes (an example of the state of the conveying device) can be acquired from a log or the like. Furthermore, the state of the conveying device is not limited to the position through which the medium passes. A transition path is information representing the transition of the state of the monitored object.

[0021] The information processing system of each embodiment will be described below. The information processing system may be installed at the same location (base, office, building, etc.) as the transport device 10 to be monitored, or may be installed at a different location from the transport device 10 and configured to remotely monitor the transport device 10 via a network (such as the Internet).

[0022] (First embodiment) 2 is a block diagram showing an example of the configuration of the information processing system 100 according to the first embodiment. As shown in FIG. 2, the information processing system 100 includes a plurality of storage units (a normal log storage unit 121, a candidate storage unit 122, an exception path storage unit 123, a path storage unit 124, an operation log storage unit 141, a deviation path storage unit 142, and a score storage unit 143), a learning unit 110, a prediction unit 130, and a display 151.

[0023] The display 151 is a display device for displaying various types of information used in the information processing system 100. The display 151 is realized by, for example, a liquid crystal display, a touch panel, or the like.

[0024] Each storage unit can be configured from any commonly used storage medium such as a flash memory, a memory card, a RAM (Random Access Memory), an HDD (Hard Disk Drive), or an optical disk.

[0025] Each storage unit may be implemented as a physically different storage medium, or as different storage areas of the same physically identical storage medium.Furthermore, each storage unit may be implemented as a plurality of physically different storage media.

[0026] The normal log storage unit 121 stores normal logs. The normal log represents operation logs obtained in the past for the transport device 10 during a certain period (e.g., one month) during which the transport device 10 is assumed to be operating normally. The normal log may be generated by a device external to the information processing system 100, or may be generated by the information processing system 100 (e.g., the learning unit 110).

[0027] Fig. 3 is a diagram showing an example of the data structure of an operation log (normal log). As shown in Fig. 3, the operation log includes a time, an event, and the number of media. An event indicates, for example, a location where a medium has passed. In the following, the locations where the passage of a medium is detected by media passage sensors 11-1 to 11-4, respectively, are designated as P1, P2, P3, and P4. The number of media indicates the number of media that have passed through the corresponding location.

[0028] Returning to FIG. 2, the candidate storage unit 122 stores candidates for normal paths generated by the learning unit 110 (candidate generation unit 111, described later). The normal path represents a transition path (first path) that is assumed to be in a first specific state. The first specific state is, for example, a normal state. The normal paths include paths that appear frequently among the transition paths and paths that do not appear frequently but are predetermined as normal. Details of the candidate generation process by the candidate generation unit 111 will be described later.

[0029] The exception path storage unit 123 stores predetermined reference paths. The reference paths are paths that the correction unit 112 refers to when correcting a candidate normal path. The reference paths represent, for example, paths that are added or excluded as exceptions to the candidate normal paths (hereinafter referred to as exception paths).

[0030] The path storage unit 124 stores the normal path obtained by correcting the normal path candidate using the exceptional path by the correction unit 112. The normal path corresponds to an abnormality sign detection model for detecting an abnormality sign.

[0031] The operation log storage unit 141 stores operation logs. The operation logs stored in the operation log storage unit 141 are used as input data for abnormality sign detection by the prediction unit 130. The operation logs stored in the operation log storage unit 141 may be different logs from the operation logs that served as the basis for the normal logs used in the learning process by the learning unit 110, or may include the same logs.

[0032] The deviation path storage unit 142 stores deviation paths. A deviation path indicates a path extracted by the prediction unit 130 (extraction unit 131, described later) from the operation log as a path that deviates from a normal path.

[0033] The score storage unit 143 stores a sign score that indicates the degree of a sign of an abnormality, calculated by the prediction unit 130 (score calculation unit 132, described later) using the deviation path.

[0034] The learning unit 110 executes a learning process to generate and correct candidates for normal paths to obtain normal paths, mainly using a normal log storage unit 121, a candidate storage unit 122, an exception path storage unit 123, and a path storage unit 124. The learning unit 110 includes a candidate generation unit 111 and a correction unit 112.

[0035] The candidate generation unit 111 generates candidates for normal paths. For example, the candidate generation unit 111 uses a normal log to generate transition paths that frequently appear among the transition paths extracted from the normal log as normal path candidates. For example, the candidate generation unit 111 extracts transition paths that represent transitions in the positions through which a medium passes from the normal log using a sequential pattern extraction method (for example, Non-Patent Document 1), and generates transition paths that occur at a frequency equal to or greater than a lower limit value among the extracted transition paths as normal path candidates. The candidate generation unit 111 stores the generated normal path candidates in the candidate storage unit 122.

[0036] In the example of the conveying device 10 shown in Figure 1, if there is only one medium, the most common transition path is "P1 → P2 → P3 → P4." If two media are loaded on top of each other, the two media are separated by a conveyance equivalent to "P2 → P3 → P2 → P3." Therefore, the most common transition path is "P1 → P2 → P3 → P2 → P3 → P4."

[0037] The correction unit 112 corrects the normal path candidates using the exception paths to obtain normal paths. The exception paths include excluded paths and additional paths. The excluded paths are paths that are predetermined as paths to be excluded from the normal path candidates. The additional paths are paths that are predetermined as paths to be added to the normal path candidates. For example, the correction unit 112 corrects the normal path candidates by excluding the excluded paths from the normal path candidates. Furthermore, the correction unit 112 corrects the normal path candidates by adding the additional paths to the normal path candidates.

[0038] The correction unit 112 stores the normal paths obtained by the correction in the path storage unit 124. If the normal path candidates do not include the excluded path and the additional path is not stored in the exception path storage unit 123, the correction unit 112 stores the normal path candidates as normal paths in the path storage unit 124 as they are.

[0039] The excluded paths and added paths are determined in advance based on, for example, accelerated tests (endurance tests, reliability tests) of the transport device 10, design knowledge, mechanical characteristics, maintenance knowledge, and the like, regardless of the occurrence frequency.

[0040] For example, a transition path such as "P1 → P2 → P1 → P2 → P3 → P4" indicates that the medium progressed to P2, then returned to P1 and continued processing. This corresponds to a situation where, for example, an attempt was made to move the medium using transport rollers 12-5 to 12-8 along the way, but due to the low frictional force on the medium surface, the medium was returned to position P1 and transported again to P2. Such a transition path does not necessarily indicate an abnormality, but can be interpreted as a normal path that is known in advance to occur depending on mechanical characteristics. Such paths are stored in advance in the exception path storage unit 123 as additional paths.

[0041] Conversely, if the medium remains at the transport position P2 for a while, such as in the sequence "P1 → P2 → P2 → P3 → P4," this can be interpreted as a sign of an abnormality that prevents the medium from being transported properly due to roller wear or the like. If such a transition path occurs frequently, it may be stored in the candidate storage unit 122 as a candidate for a frequently occurring path, but since it indicates a sign of an abnormality, it must be excluded from the candidates. Therefore, in this embodiment, transition paths that indicate such signs of an abnormality are stored in advance in the exception path storage unit 123 as excluded paths.

[0042] The prediction unit 130 performs abnormality sign detection mainly using the operation log storage unit 141, the deviation path storage unit 142, and the score storage unit 143. The abnormality sign detection is a process of extracting deviation paths that do not match normal paths (deviations from normal paths) and calculating a sign score based on the deviation paths to detect abnormality signs. The prediction unit 130 includes an extraction unit 131, a score calculation unit 132, a determination unit 133, and an output control unit 134.

[0043] The extraction unit 131 extracts deviation paths that do not match normal paths from the operation log (second log information) obtained within a determination period for determining whether the transport device 10 to be monitored is in a specific state (for example, abnormal). For example, the extraction unit 131 extracts normal paths included in the path storage unit 124 from the operation log stored in the operation log storage unit 141, and stores the remaining transition paths other than the extracted normal paths as deviation paths in the deviation path storage unit 142 for a certain period of time (for example, every day).

[0044] For example, a transition path such as "P1 → P2 → P1 → P2 → P1 → P2 → P3 → P4" indicates that the medium repeatedly stops at P2, and is not stored as a normal path in the path storage unit 124. Therefore, the extraction unit 131 extracts this transition path as a deviation path and stores it in the deviation path storage unit 142.

[0045] The score calculation unit 132 calculates a predictive score using the extracted deviation path. For example, the score calculation unit 132 calculates a predictive score for each fixed period (e.g., one day) using the number of times inserted per fixed period (e.g., one day) as the denominator and the number of occurrences of sequential paths in the deviation path storage unit 142 per fixed period as the numerator, and stores the predictive score in the score storage unit 143. The number of times inserted indicates the number of times media have been inserted into the conveyance device 10.

[0046] The determination unit 133 determines whether or not the predictive score exceeds a predetermined warning threshold at regular intervals, and passes the determination result for each regular interval to the output control unit 134.

[0047] The method of determining a sign is not limited to the above. For example, the determination unit 133 may determine a sign of an abnormality using changes (such as gradients) in multiple sign scores calculated for multiple target periods. The multiple target periods are, for example, a period within the current determination period and a period within a past determination period (for example, one month ago). For example, the determination unit 133 calculates the amount of change (such as gradients) in the current sign score relative to the past sign score, and determines that there is a sign of an abnormality if the amount of change exceeds a threshold. If a short time difference between the current and past periods would result in an erroneous determination, the target past determination period may be determined so that the difference is longer.

[0048] The output control unit 134 controls the output of various information used in the information processing system 100. For example, the output control unit 134 outputs to the display 151 the predictive scores for each fixed period, the determination results by the determination unit 133, and information that visualizes some or all of the deviation paths.

[0049] Each of the above units (learning unit 110, prediction unit 130) is realized, for example, by one or more processors. For example, each of the above units may be realized by having a processor such as a CPU (Central Processing Unit) execute a program, i.e., by software. Each of the above units may be realized by a processor such as a dedicated IC (Integrated Circuit), i.e., by hardware. Each of the above units may be realized by a combination of software and hardware. When multiple processors are used, each processor may realize one of the units, or two or more of the units.

[0050] Next, a description will be given of the learning process executed by the learning unit 110. Fig. 4 is a flowchart showing an example of the learning process in the first embodiment.

[0051] The candidate generator 111 extracts frequently occurring transition paths as candidates for normal paths from the past normal logs stored in the normal log storage unit 121, and stores them in the candidate storage unit 122 (step S101).

[0052] The corrector 112 determines whether the normal path candidates include an exception path stored in the exception path storage unit 123 (step S102). If the normal path candidates include the exception path stored in the exception path storage unit 123 (step S102: Yes), the corrector 112 excludes the exception path from the normal path candidates (step S103).

[0053] If the normal path candidates do not include an excluded path (step S102: No) and after step S103, the correction unit 112 determines whether an additional path is stored in the exception path storage unit 123 (step S104). If an additional path is stored (step S104: Yes), the correction unit 112 adds the additional path to the normal path candidates (step S105).

[0054] If no additional path is stored (step S104: No), or after step S105, the learning unit 110 ends the learning process.

[0055] Next, a description will be given of abnormality sign detection executed by the prediction unit 130. Fig. 5 is a flowchart showing an example of abnormality sign detection in the first embodiment.

[0056] The extraction unit 131 extracts deviation paths that do not match normal paths from the operation logs stored in the operation log storage unit 141, and stores the deviation paths in the deviation path storage unit 142 (step S201). The score calculation unit 132 calculates a predictive score using the extracted deviation paths (step S202). The determination unit 133 determines a predictive score for an abnormality using the calculated predictive score (step S203). The output control unit 134 outputs a warning screen for visualizing the predictive score, the determination result by the determination unit 133, and the deviation paths to, for example, the display 151 (step S204).

[0057] 6 is a diagram showing an example of a warning screen. As shown in FIG. 6, the warning screen 600 includes a graph 601 showing changes in the predictive score and information 602 indicating the basis for the predictive error. The information 602 indicating the basis for the predictive error includes, for example, information indicating the extracted deviation path and information indicating the frequency of occurrence of the deviation path. This warning screen makes it possible to detect abnormalities in the behavior of the entire system while indicating the basis for the detection results.

[0058] In this way, the information processing system according to the first embodiment extracts normal paths from operation logs for periods of normal operation and stores them as anomaly sign detection models. The information processing system performs anomaly sign detection by determining deviation paths that deviate from normal paths from the operation logs for the period to be monitored. This enables more appropriate anomaly sign detection.

[0059] In this embodiment, deviations from various normal operating states are identified as abnormality signs, and the extent to which the abnormality signs have occurred can be quantified. That is, normal paths representing normal operating states are comprehensively extracted from past operation logs, and an abnormality score is calculated that indicates the extent to which the state sequence of a new operation log to be monitored deviates from the normal path. This makes it possible to calculate an abnormality score even for unknown abnormality signs, unlike methods based on state transition diagrams that detect limited causes of abnormalities, for example. Furthermore, by outputting the deviation path, it is possible to clearly indicate which part and what system behavior is causing the abnormality.

[0060] For example, because the predictive score and the basis for the predictive sign (such as the location of an abnormality) can be obtained simultaneously, it becomes easier to make decisions regarding maintenance inspections and part replacements, which results in improved equipment availability and more efficient maintenance.

[0061] Furthermore, because normal paths are extracted from past operation logs, even if there are individual differences in the operation of the conveyance devices 10, it is possible to extract normal paths corresponding to the operation of each conveyance device 10. By using the normal paths extracted in this way, it becomes possible to detect abnormalities with higher accuracy.

[0062] (Second embodiment) The information processing system according to the second embodiment calculates the sign score using not only the deviation path but also additional information, such as weather information, but also any other information related to the abnormal sign.

[0063] Fig. 7 is a block diagram showing an example of the configuration of an information processing system 100-2 according to the second embodiment. As shown in Fig. 7, the information processing system 100-2 includes a plurality of storage units (a normal log storage unit 121, a candidate storage unit 122, an exception path storage unit 123, a path storage unit 124, an operation log storage unit 141, a deviation path storage unit 142, a score storage unit 143, and a weather information storage unit 144-2), a learning unit 110, a prediction unit 130-2, and a display 151.

[0064] The second embodiment differs from the first embodiment in that a weather information storage unit 144-2 is added and in the function of a score calculation unit 132-2 in a prediction unit 130-2. The other configurations and functions are the same as those in FIG. 2, which is a block diagram of the information processing system 100 according to the first embodiment, so the same reference numerals are used and the description thereof will be omitted here.

[0065] The weather information storage unit 144-2 stores at least weather information for the location where the transport device 10 to be monitored is installed during the determination period for determining whether an abnormality has occurred. The weather information includes temperature, humidity, air pressure, and precipitation, but may also include any other weather-related information.

[0066] The score calculation unit 132-2 calculates the predictive score using the deviation path and the weather information. For example, the score calculation unit 132-2 calculates the predictive score by dividing a value obtained by dividing the number of occurrences by the number of insertions, as in the first embodiment, by the maximum humidity observed for a certain period at a weather station close to the location where the transport device 10 is installed.

[0067] For example, when humidity is high, there may be a situation where a warning sign is more likely to appear temporarily, but this does not affect the progress of deterioration of the conveyance device 10. According to this embodiment, by using a warning score corrected by humidity, it is possible to detect a warning sign of an abnormality taking such a situation into consideration.

[0068] Other than the change in the method of calculating the sign score, the flow of the learning process and the flow of abnormality sign detection are the same as those in the above embodiment (FIGS. 4 and 5), and therefore a description thereof will be omitted.

[0069] (Third embodiment) In the third embodiment, a configuration example for remotely monitoring a monitoring target will be described. Fig. 8 is a block diagram showing an example of the configuration of an information processing system 100-3 according to the third embodiment. As shown in Fig. 8, the information processing system 100-3 has a configuration in which a learning device 200-3 and a prediction device 300-3 are connected via a network 400-3.

[0070] The network 400-3 may be, for example, the Internet, or any other type of network. The network 400-3 may be a wireless network, a wired network, or a network in which both wireless and wired networks coexist.

[0071] The learning device 200-3 is a device that mainly has functions related to learning processing. The learning device 200-3 can be configured as a server device in a cloud environment, for example.

[0072] The learning device 200-3 includes a normal log storage unit 121, a candidate storage unit 122, an exception path storage unit 123, a path storage unit 124, a learning unit 110, and a communication control unit 201-3. Since the units other than the communication control unit 201-3 have the same functions as those in the above embodiment, the same reference numerals are used and the description thereof will be omitted.

[0073] The communication control unit 201-3 controls communication with external devices such as the prediction device 300-3. For example, the communication control unit 201-3 transmits a normal path obtained by the learning process to the prediction device 300-3 via the network 400-3.

[0074] The prediction device 300-3 is a device that mainly has a function related to abnormality sign detection. The prediction device 300-3 can be configured as a server device installed at the same location as the transport device 10, for example. The prediction device 300-3 may also be configured as a server device in a cloud environment.

[0075] The prediction device 300-3 includes a prediction unit 130, an operation log storage unit 141, a deviation path storage unit 142, a score storage unit 143, a path storage unit 145-3, a display 151, and a communication control unit 301-3. Since the units other than the path storage unit 145-3 and the communication control unit 301-3 have the same functions as those in the above embodiment, the same reference numerals are used and the description thereof will be omitted.

[0076] Path storage unit 145-3 stores the normal path transmitted from learning device 200-3. Note that prediction device 300-3 may not include path storage unit 145-3, and prediction unit 130 (extraction unit 131) may extract a deviation path by referring to path storage unit 124 of learning device 200-3.

[0077] The communication control unit 301-3 controls communication with an external device such as the learning device 200-3. For example, the communication control unit 301-3 receives a normal path obtained by the learning process from the learning device 200-3 via the network 400-3.

[0078] The flow of the learning process and the flow of the abnormality sign detection are the same as those in the above embodiment (FIGS. 4 and 5), and therefore a description thereof will be omitted.

[0079] In this way, the information processing system according to the third embodiment can achieve a configuration in which functions are distributed among a plurality of devices.

[0080] As described above, according to the first to third embodiments, abnormality sign detection can be more appropriately performed.

[0081] Next, the hardware configuration of the information processing system according to the first to third embodiments will be described with reference to Fig. 9. Fig. 9 is an explanatory diagram showing an example of the hardware configuration of the information processing system according to the first to third embodiments.

[0082] The information processing systems according to the first to third embodiments include a control device such as a CPU (Central Processing Unit) 51, a storage device such as a ROM (Read Only Memory) 52 and a RAM (Random Access Memory) 53, a communication I / F 54 that connects to a network and communicates, and a bus 61 that connects each part.

[0083] The programs executed in the information processing systems according to the first to third embodiments are provided in advance in the ROM 52 or the like.

[0084] The programs executed by the information processing systems according to the first to third embodiments may be configured to be provided as a computer program product by being recorded in an installable or executable file format on a computer-readable recording medium such as a CD-ROM (Compact Disk Read Only Memory), a flexible disk (FD), a CD-R (Compact Disk Recordable), or a DVD (Digital Versatile Disk).

[0085] Furthermore, the programs executed by the information processing systems according to the first to third embodiments may be stored on a computer connected to a network such as the Internet and provided by being downloaded via the network. Also, the programs executed by the information processing systems according to the first to third embodiments may be provided or distributed via a network such as the Internet.

[0086] The programs executed in the information processing systems according to the first to third embodiments can cause a computer to function as each part of the information processing system. In this computer, the CPU 51 can read the programs from a computer-readable storage medium onto a main storage device and execute the programs.

[0087] Although several embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments can be embodied in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, and are also included in the scope of the invention and its equivalents as defined in the claims.

[0088] A configuration example of the embodiment will be described below. (Configuration example 1) a candidate generation unit that uses first log information indicating a state of a monitoring target obtained within a certain period of time to generate frequently occurring transition paths for the state as candidates for a first path that is a transition path assumed to be a first specific state; a correction unit that corrects the candidate using a predetermined reference path to obtain the first path; An information processing device comprising: (Configuration example 2) an extraction unit that extracts a deviation path that does not match the first path from second log information indicating the state of the monitoring target obtained within a determination period for determining whether the monitoring target is in a second specific state; a score calculation unit that calculates a score indicating a degree of a sign of the second specific state using the deviation path; a determination unit that determines a sign of the second specific state using the score; an output control unit that outputs information that visualizes a part or all of the score, the determination result by the determination unit, and the deviation path; Further provided with The information processing device according to configuration example 1. (Configuration example 3) the score calculation unit calculates the score using the deviation path and weather information for the determination period. The information processing device according to configuration example 2. (Configuration Example 4) the score calculation unit calculates the score for each of a plurality of target periods; the determination unit determines the sign of the second specific state using changes in the scores calculated for the target periods. The information processing device according to configuration example 2 or 3. (Configuration Example 5) a learning device including the candidate generation unit and the correction unit; a prediction device including the extraction unit, the score calculation unit, the determination unit, and the output control unit; Equipped with The information processing device according to any one of configuration examples 2 to 4. (Configuration Example 6) the reference paths include an exclusion path that is predetermined as a path to be excluded from the candidates; the correction unit determines the first path by excluding the excluded path from the candidates. 6. The information processing device according to any one of configuration examples 2 to 5. (Configuration Example 7) the reference path includes a predetermined additional path as a path to be added to the candidate; the correction unit determines the first path by adding the additional path to the candidates. 7. The information processing device according to any one of configuration examples 2 to 6. (Configuration Example 8) the monitoring target is a transport device that transports a medium, The state represents a position through which the medium passes; The transition path represents a transition of positions through which the medium passes. 8. The information processing device according to any one of configuration examples 1 to 7. (Configuration Example 9) The first specific state is a normal state. 10. The information processing device according to any one of configuration examples 1 to 8. (Configuration Example 10) The second specific state is an abnormal state. The information processing device according to any one of configuration examples 2 to 4. (Configuration Example 11) An information processing method executed by an information processing device, a candidate generation step of generating, using first log information indicating a state of a monitoring target obtained within a certain period of time, a transition path that frequently appears for the state as a candidate for a first path that is a transition path assumed to be a first specific state; a correction step of correcting the candidate using a predetermined reference path to obtain the first path; An information processing method including: (Configuration Example 12) On the computer, a candidate generation step of generating, using first log information indicating a state of a monitoring target obtained within a certain period of time, a transition path that frequently appears for the state as a candidate for a first path that is a transition path assumed to be a first specific state; a correction step of correcting the candidate using a predetermined reference path to obtain the first path; A program to execute. [Explanation of symbols]

[0089] 10. Conveyor 100, 100-2, 100-3 Information Processing Systems 110 Learning Department 111 Candidate generation section 112 Correction section 121 Normal log storage section 122 Candidate storage section 123 Exception Path Memory 124 Path memory section 130, 130-2 Prediction Section 131 Extraction part 132, 132-2 Score calculation section 133 Judgment section 134 Output control section 141 Operation log storage unit 142 Deviation Path Memory Unit 143 Score memory section 144-2 Weather information storage section 145-3 Path memory section 151 Display 200-3 Learning Device 201-3 Communication control section 300-3 Prediction Device 301-3 Communication control section 400-3 Network

Claims

1. a candidate generation unit that uses first log information indicating a state of a monitoring target obtained within a certain period of time to generate frequently occurring transition paths for the state as candidates for a first path that is a transition path assumed to be a first specific state; a correction unit that corrects the candidate using a predetermined reference path to obtain the first path; an extraction unit that extracts deviation paths that do not match the first path from second log information indicating the state of the monitoring target obtained within a determination period for determining whether the monitoring target is in a second specific state; An information processing device comprising:

2. A score calculation unit that calculates a score indicating the degree of a sign of the second specific state using the deviation path; a determination unit that determines a sign of the second specific state using the score; an output control unit that outputs information that visualizes a part or all of the score, the determination result by the determination unit, and the deviation path; Further provided with The information processing device according to claim 1 .

3. the score calculation unit calculates the score using the deviation path and weather information for the determination period. The information processing device according to claim 2 .

4. the score calculation unit calculates the score for each of a plurality of target periods; the determination unit determines a sign of the second specific state using changes in the scores calculated for the target periods. The information processing device according to claim 2 .

5. a learning device including the candidate generation unit and the correction unit; a prediction device including the extraction unit, the score calculation unit, the determination unit, and the output control unit; Equipped with The information processing device according to claim 2 .

6. the reference paths include an exclusion path that is predetermined as a path to be excluded from the candidates; the correction unit determines the first path by excluding the excluded path from the candidates. The information processing device according to claim 2 .

7. the reference path includes a predetermined additional path as a path to be added to the candidate; the correction unit determines the first path by adding the additional path to the candidates. The information processing device according to claim 2 .

8. the monitoring target is a transport device that transports a medium, The state represents a position through which the medium passes; The transition path represents a transition of positions through which the medium passes. The information processing device according to claim 1 .

9. The first specific state is a normal state. The information processing device according to claim 1 .

10. The second specific state is an abnormal state. The information processing device according to claim 2 .

11. An information processing method executed by an information processing device, a candidate generation step of generating, using first log information indicating a state of a monitoring target obtained within a certain period of time, a transition path that frequently appears for the state as a candidate for a first path that is a transition path assumed to be a first specific state; a correction step of correcting the candidate using a predetermined reference path to obtain the first path; an extraction step of extracting a deviation path that does not match the first path from second log information indicating the state of the monitoring target obtained within a determination period for determining whether the monitoring target is in a second specific state; An information processing method including:

12. On the computer, a candidate generation step of generating, using first log information indicating a state of a monitoring target obtained within a certain period of time, a transition path that frequently appears for the state as a candidate for a first path that is a transition path assumed to be a first specific state; a correction step of correcting the candidate using a predetermined reference path to obtain the first path; an extraction step of extracting a deviation path that does not match the first path from second log information indicating the state of the monitoring target obtained within a determination period for determining whether the monitoring target is in a second specific state; A program to execute.

Images