Remote monitoring system and remote monitoring method

Abstract

To provide a remote monitoring system with which it is possible to detect abnormality or abnormality sign of a lane apparatus in real time with good accuracy.SOLUTION: A remote monitoring system for monitoring a lane apparatus installed on a toll gate lane from a distance comprises: a data server for storing master data including the installation environment and composition information of the lane apparatus; a first data collection unit for collecting log data related to operation or processing of the lane apparatus; an abnormality detection model training unit for training, by using training data including the log data and the master data, an abnormality detection model in which the log data is adopted as an explanatory variable and the presence of abnormality or a sign of abnormality of the lane apparatus is adopted as an objective variable; and a monitoring unit that, when the log data is inputted to the abnormality detection model and a detection result is outputted indicating that there is abnormality or a sign of abnormality of the lane apparatus, notifies monitoring personnel of the detection result.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present disclosure relates to a remote monitoring system and a remote monitoring method. [Background technology]

[0002] At toll booths on toll roads, lane equipment used to collect tolls (various sensors such as vehicle detectors and footboards, vehicle type identification devices, wireless communication devices, electronic signboards that display toll fees and whether or not a vehicle can pass, etc.) is installed on each lane.

[0003] In recent years, the use of the Electronic Toll Collection System (ETC (registered trademark)) has become so high that it is being considered to make toll booths unmanned and exclusively for ETC vehicles. At manned toll booths, toll collectors on duty can easily notice signs of abnormalities (breakdowns) in the lane equipment from the sound, appearance, smell, heat, etc. Furthermore, in the event of a breakdown, toll collectors can visually inspect the lane equipment and quickly take action such as arranging for repairs or replacement.

[0004] On the other hand, at unmanned toll booths, a monitor remotely monitors multiple lanes while watching video from surveillance cameras. However, it is difficult to detect signs of abnormalities such as sounds or smells using only video from surveillance cameras. Furthermore, if a malfunction occurs in lane equipment, a maintenance technician will first be contacted by the monitor and go to the site to visually confirm the malfunction and then arrange for repair or replacement. Therefore, at unmanned toll booths that use remote monitoring, the time from the occurrence of a malfunction to the response (repair or replacement) may be longer than at manned toll booths.

[0005] As a technique for reducing downtime of facilities such as toll booths, for example, Patent Document 1 describes creating a facility inspection plan using a model that has learned from failures that have occurred in the past. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Japanese Patent Application Publication No. 2023-069397 Summary of the Invention [Problem to be solved by the invention]

[0007] However, conventional technologies do not monitor the operating status of lane equipment, making it difficult to detect potential malfunctions (signs of abnormalities) in advance or to quickly detect malfunctions (anomalies) that have just occurred. Furthermore, when there is a malfunction in lane equipment or a decline in sensor accuracy, it is difficult to distinguish whether this is a sign of an abnormality in the lane equipment (e.g., deterioration, dirt, etc.) or whether it is due to the influence of the lane equipment's installation environment (e.g., surrounding structures, weather, road surface inclination, vibrations, etc.).

[0008] An object of the present disclosure is to provide a remote monitoring system and a remote monitoring method that can accurately detect abnormalities or signs of abnormalities in lane devices in real time. [Means for solving the problem]

[0009] According to one aspect of the present disclosure, a remote monitoring system is a remote monitoring system that monitors lane equipment installed in lanes of a toll gate from a remote location, and includes: a data server that stores master data including installation environment and configuration information of the lane equipment; a first data collection unit that collects log data related to the operation or processing of the lane equipment; an anomaly detection model learning unit that uses learning data including the log data and the master data to learn an anomaly detection model in which the log data is an explanatory variable and the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is a target variable; and a monitoring unit that inputs the log data into the anomaly detection model, and if a detection result indicating the presence of an abnormality or a sign of an abnormality in the lane equipment is output, notifies a monitor of the detection result.

[0010] According to one aspect of the present disclosure, a remote monitoring method is a method for remotely monitoring lane equipment installed in a lane of a toll gate, the method comprising the steps of: collecting log data related to the operation or processing of the lane equipment; using learning data including the log data and master data including installation environment and configuration information of the lane equipment recorded in a data server, learning an anomaly detection model in which the log data is an explanatory variable and the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is an objective variable; and inputting the log data into the anomaly detection model, and when a detection result indicating the presence of an abnormality or a sign of an abnormality in the lane equipment is output, notifying a monitor of the detection result. [Effects of the Invention]

[0011] According to the above aspect, it is possible to accurately detect abnormalities or signs of abnormalities in lane devices in real time. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a schematic diagram showing the overall configuration of a remote monitoring system according to a first embodiment. [Figure 2] FIG. 2 is a block diagram showing the functional configuration of a central device according to the first embodiment. [Figure 3] 1 is a block diagram showing the functional configuration of a failure analysis apparatus according to a first embodiment. [Figure 4] 1 is a block diagram showing a functional configuration of a maintenance planning device according to a first embodiment. [Figure 5] 10 is a flowchart showing an example of a learning phase of the central device according to the first embodiment. [Figure 6] FIG. 4 is a diagram illustrating a learning process of the central device according to the first embodiment. [Figure 7] 6 is a flowchart showing an example of an operation phase of the central device according to the first embodiment. [Figure 8] 5 is a flowchart showing an example of a learning phase of the failure analysis apparatus according to the first embodiment. [Figure 9]FIG. 3 is a diagram for explaining a learning process of the failure analysis device according to the first embodiment. [Figure 10] 5 is a flowchart showing an example of an operation phase of the failure analysis apparatus according to the first embodiment. [Figure 11] 5 is a flowchart showing an example of a learning phase of the maintenance planning device according to the first embodiment. [Figure 12] FIG. 2 is a diagram for explaining a learning process of the maintenance planning device according to the first embodiment. [Figure 13] 5 is a flowchart showing an example of an operation phase of the maintenance planning device according to the first embodiment. [Figure 14] FIG. 2 is a diagram illustrating an example of a hardware configuration of a central device, a failure analysis device, and a maintenance planning device. DETAILED DESCRIPTION OF THE INVENTION

[0013] First Embodiment Hereinafter, the embodiments will be described in detail with reference to the drawings.

[0014] (Overall configuration of the remote monitoring system) Fig. 1 is a schematic diagram showing the overall configuration of a remote monitoring system according to the first embodiment. As shown in Fig. 1, the remote monitoring system 1 includes a central device 2, a failure analysis device 3, a maintenance planning device 4, and a data server 5. These devices are communicatively connected via a network NW1, which is a private network of a road operator that operates a toll road.

[0015] 1 shows an example in which the failure analysis device 3 and the maintenance planning device 4 are configured as devices independent of the central device 2, but this is not limiting. In other embodiments, these devices may be configured as an integrated device. For example, the central device 2 may be configured to include the functions of one or both of the failure analysis device 3 and the maintenance planning device 4.

[0016] The central device 2 remotely monitors multiple lanes L (L1, L2, ..., Ln). In this embodiment, the central device 2 collects log data of lane equipment on each lane L via a network NW2, which is a private network of the road operator, and monitors for abnormalities (failures) or signs of abnormalities in the lane equipment.

[0017] The failure analysis device 3 analyzes abnormalities that occur in the lane equipment.

[0018] The maintenance planning device 4 creates a maintenance plan for lane equipment. The maintenance plan includes the frequency and schedule of maintenance and inspection of lane equipment.

[0019] The data server 5 is a server that stores data commonly used by the central device 2, the failure analysis device 3, and the maintenance planning device 4. The data server 5 has tollgate master data D51 and lane equipment master data D52. The tollgate master data D51 is master data that manages data that represents the structure and equipment configuration of a toll road (route). The tollgate master data D51 records the number of lanes at each tollgate, as well as the identification ID and installation date of the lane equipment installed in each lane. The lane equipment master data D52 is master data that manages data related to the configuration information of all lane equipment installed on the toll road. The lane equipment master data D52 records the identification ID, model, version number, and special specifications (salt damage countermeasures, cold weather countermeasures, etc.) of each lane equipment.

[0020] Lane L is, for example, a lane dedicated to ETC vehicles provided at an unmanned toll gate on a toll road (hereinafter simply referred to as an ETC lane). Lane L may also include a free-flow lane provided at a charging point on a free-flow toll road. In this embodiment, an example will be described in which lane L is an ETC lane. The central device 2 monitors, for example, the lanes L of all toll gates on the route of the toll road, or the lanes L of some toll gates included in a predetermined jurisdiction area.

[0021] Each lane L is equipped with multiple lane devices. The lane devices include, for example, wireless communication devices, lane control devices, cash processing machines, barriers, signboards, lane monitoring cameras, communication devices, and vehicle type identification devices. The wireless communication devices perform wireless communication related to charging processing with onboard devices installed in vehicles. The lane control devices control each lane device and perform charging processing. The lane control devices are also communicatively connected to a central unit 2 and transmit log data of the lane devices to the central unit 2. The cash processing machines accept cash payments of tolls from vehicle passengers (users) when ETC charging processing is not possible. The barriers are installed at the exits of the lane L and control vehicle passage by opening and closing. The signboards inform users of the operational status of the lane L (e.g., "closed," "ETC only"), etc. The lane monitoring cameras capture images of the lane L. The communication devices are intercoms that allow users to communicate with supervisors in remote locations. The vehicle type identification device identifies the vehicle types of vehicles traveling on the lane L. The vehicle type discrimination device has sensor equipment such as a vehicle detector that detects vehicles entering and exiting, an axle number detector (tread) that measures the number of axles and width of the vehicle, a vehicle height detector that measures the vehicle height, a vehicle number recognition camera that takes images of the vehicle's license plate, and a vehicle number recognition device that recognizes license plate information from the license plate image using OCR.

[0022] The central device 2 may also be communicably connected to an external system 6 via a network NW3 such as the Internet. The external system 6 provides, for example, weather data and traffic volume data for each time period for each district (prefecture, city, etc.).

[0023] (Functional configuration of the central device) 2 is a block diagram showing the functional configuration of the central device according to the first embodiment. As shown in FIG. 2, the central device 2 includes a first data collection unit 21, an anomaly detection model learning unit 22, and a monitoring unit 23.

[0024] The first data collection unit 21 collects log data relating to the operation or processing of lane devices. The log data includes first log data D22, which is structured data, and second log data, which is unstructured data.

[0025] The first log data D22 is standard data that is output periodically when each lane device is operating normally, and standard data that is output when an event such as an error is triggered, and includes the date and time the log was generated, information for identifying the device (e.g., route number, toll gate number, lane number, device identification ID), status code (code indicating normal or abnormal), device operating time or number of times, sensor value, processing result (e.g., vehicle type determination result, license plate recognition result), etc.

[0026] The second log data D23 includes the date and time the log was generated, information for identifying the device, video files (video) captured by a surveillance camera, image files (still images) captured by a vehicle license plate recognition camera, and audio files recorded by a communication device.

[0027] The first data collection unit 21 further collects external data D24 from the external system 6. The external data D24 includes weather data and traffic volume data. The weather data includes, for example, information identifying the target area (information indicating the prefecture, city, etc.), date and time (time zone), temperature, precipitation, wind speed, air pressure, etc. The traffic volume data includes information identifying the target area, date and time (time zone), number of vehicles, average speed, etc.

[0028] The first data collection unit 21 may further collect maintenance history data (repair history data D31, maintenance and inspection history data D32, and notification history data D33) from the later-described failure analysis device 3. The first data collection unit 21 may further collect anomaly detection results (anomaly detection history data D21) by the later-described monitoring unit 23.

[0029] The anomaly detection model learning unit 22 uses learning data including log data of lane equipment (first log data D22, second log data D23) and master data (toll gate master data D51, lane equipment master data D52) recorded in the data server 5 to learn an anomaly detection model M1 in which the log data is an explanatory variable and the prediction result of the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is an objective variable. The learning data may also include maintenance history data (repair history data D31, maintenance and inspection history data D32, and report history data D33) collected and generated by the failure analysis device 3 described later.

[0030] The monitoring unit 23 inputs log data into the anomaly detection model M1, and when a detection result indicating an abnormality or a sign of an abnormality in lane equipment is output, the monitoring unit 23 notifies the monitor of the detection result. The detection result (anomaly detection result) includes the date and time of detection, information for identifying the equipment that detected the abnormality, details of the detected abnormality (type of abnormality, status such as serious or suspected), etc. The monitoring unit 23 notifies the monitor of the abnormality detection result via an operation console 24 such as a PC used by the monitor. The monitor refers to the abnormality detection result and performs remote operation of lane equipment (such as closing a lane) or reports the abnormality to a maintenance worker.

[0031] (Functional configuration of failure analysis device) 3 is a block diagram showing the functional configuration of the failure analysis device according to the first embodiment. As shown in FIG. 3, the failure analysis device 3 includes a second data collection unit 31, a failure analysis model learning unit 32, and an analysis unit 33.

[0032] The second data collection unit 31 collects maintenance history data including information on abnormalities, repairs, and maintenance inspections that have occurred in lane equipment. The maintenance history data includes, for example, repair history data D31, maintenance and inspection history data D32, and notification history data D33.

[0033] The repair history data D31 is data that records the details of the actions (repair, cleaning, etc.) taken by a maintenance technician to resolve an abnormality in a lane device. The repair history data D31 includes the report date and time, the report history ID, the date and time the action was completed, the reporter, information for identifying the repaired lane device (route number, lane number, device identification ID, component ID), the action classification (e.g., classification such as replacement, repair, cleaning, or no abnormality), the maintenance and inspection ID, and the action details. The report date and time and the report history ID are information for identifying which record in the report history data D33 (described later) the work corresponds to. The maintenance and inspection ID is information for identifying which record in the maintenance and inspection history data D32 (described later) the work corresponds to. The action details are information indicating the work performed by the maintenance technician to resolve the abnormality, and may be free-form report sentences such as "replace xx," "clean the dirt from xx," or "adjust yy of xx."

[0034] The maintenance and inspection history data D32 is data related to the periodic inspection of lane equipment, and includes the inspection date and time, the maintenance and inspection ID, the reporter, information for identifying the inspected lane equipment (route number, lane number, equipment identification ID), the inspection result classification (e.g., classification such as no abnormality, abnormality detected, etc.), etc. Note that the maintenance and inspection history data D32 may also include the report ID if the inspection result classification is abnormality detected and a report has been made.

[0035] The report history data D33 is data relating to the details of abnormalities that have been discovered and reported during maintenance inspections, remote monitoring, etc., and includes the report date and time, report history ID, reporter, information for identifying the reported lane equipment (route number, lane number, equipment identification ID), abnormality details, etc. The abnormality details are information indicating a breakdown or malfunction that has occurred in the lane equipment, and may be free-form report text such as "xx cannot be detected," "xx does not work," or "xx is damaged."

[0036] 3 shows an example in which the maintenance worker who has carried out the response inputs the data D31 to D33 via a mobile terminal 35 such as a tablet, but the present invention is not limited to this. For example, if the urgency is low, such as when no abnormality is found, the repair history data D31 and the maintenance and inspection history data D32 may be input via an operation console 34 such as a PC after the maintenance worker has finished the work and returned to the office. Furthermore, the notification history data D33 may be input via the operation console 34 by the maintenance worker based on the content of notification received from the monitor by telephone or the like.

[0037] The second data collection unit 31 may further collect from the central device 2 the abnormality detection history data D21, the first log data D22, the second log data D23, and the external data D24.

[0038] The failure analysis model learning unit 32 uses learning data including the maintenance history data D31 to D33 and the master data D51 to D52 recorded in the data server 5 to learn a failure analysis model M2 in which the abnormality details of the lane equipment are used as explanatory variables and the analysis results including potential causes or potential countermeasures for the abnormality details are used as objective variables. The learning data may also include various data collected and generated by the central device 2 (abnormality detection history data D21, first log data D22, second log data D23, and external data D24).

[0039] The analysis unit 33 receives an inquiry about an abnormality occurring in the lane equipment, inputs the abnormality into the failure analysis model M2, and notifies the maintenance personnel of the analysis results of the abnormality obtained as an output. The maintenance personnel inputs the abnormality (report history data D33) reported by, for example, a supervisor into the failure analysis device via a terminal device (a console 34 such as a PC or a mobile terminal 35 such as a tablet). The analysis unit 33 also notifies the terminal devices 34, 35 of the analysis results. The maintenance personnel refer to the analysis results on the terminal devices 34, 35, and prepares items such as replacement parts (components), tools, and cleaning supplies according to the possible causes or countermeasures of the abnormality, and then goes to the site where the lane equipment is installed to perform the response work.

[0040] (Functional configuration of the maintenance planning device) Fig. 4 is a block diagram showing the functional configuration of the maintenance planning device according to the first embodiment. As shown in Fig. 4, the maintenance planning device 4 includes a third data collection unit 41, a maintenance planning model learning unit 42, and a planner 43.

[0041] The third data collection unit 41 collects spare parts inventory data D41 for lane equipment and maintenance history data D31 to D33. The maintenance history data D31 to D33 are collected and generated by the failure analysis device 3.

[0042] The spare parts inventory data D41 is data that records the receipt and delivery of spare parts held by the road operator, and includes equipment identification ID, component ID, purchase date, purchase quantity, delivery date, delivery quantity, etc. The spare parts inventory data D41 is input by the maintenance planner operating the operation console 44.

[0043] The maintenance planning model learning unit 42 uses learning data including spare parts inventory data D41, maintenance history data D31 to D33, and master data D51 to D52 recorded in the data server 5 to learn a maintenance planning model M3 in which the toll gate is used as an explanatory variable and a maintenance plan proposal including the inspection timing or lane equipment to be replaced at the toll gate is used as an objective variable.

[0044] The planning unit 43 inputs the toll gate to be planned into the maintenance plan model M3, and creates a maintenance plan for the toll gate based on the maintenance plan proposal obtained as an output. The planning unit 43 may also create a spare parts replenishment plan based on the created maintenance plan and spare parts inventory data D41. The maintenance plan and replenishment plan created by the planning unit 43 are recorded and accumulated as maintenance plan history data D42.

[0045] (Example of central device processing: learning phase) Fig. 5 is a flowchart showing an example of the learning phase of the central device according to the first embodiment. Fig. 6 is a diagram for explaining the learning process of the central device according to the first embodiment. Here, the flow of the process of the learning phase of the central device 2 will be explained with reference to Figs. 5 and 6.

[0046] First, the first data collection unit 21 collects and stores first log data D22 and second log data D23 from all lane devices installed at each toll gate on the route (step S101). The first data collection unit 21 may also collect and store weather data and traffic volume data from the external system 6 and maintenance history data D31 to D33 from the failure analysis device 3. The collected data is used as learning data.

[0047] When the log data D22 and D23 are accumulated, the anomaly detection model learning unit 22 performs machine learning of the anomaly detection model M1 (step S102). For example, the anomaly detection model learning unit 22 performs machine learning of the anomaly detection model M1 every time a certain period of time passes or every time a certain amount of data (a certain number of samples) is accumulated.

[0048] 6 shows an example of machine learning by the anomaly detection model learning unit 22. For example, the anomaly detection model learning unit 22 learns the range of values ​​(normal range) that the log data D22 and D23 can take when normal for each lane device from the relationship between the log data D22 and D23 included in the learning data and other data. The other data includes toll gate master data D51, lane device master data D52, external data D24, and log data D22 and D23 of other lane devices. The other data may also include maintenance history data D31 to D33 collected from the failure analysis device 3.

[0049] For example, the normal ranges of the sensor values, processing results (first log data D22), and images (second log data D23) may differ depending on factors such as the installation environment of the lane device, configuration information (model, version, special specifications), time of day, and weather conditions around the toll gate. First, an example of a vehicle license plate recognition camera and a vehicle license plate recognition device will be described. A vehicle license plate recognition camera installed at a certain toll gate may be susceptible to direct sunlight due to its installation environment (installation position, orientation, presence or absence of surrounding structures), resulting in lower image contrast compared to devices installed at other toll gates, and thus reduced recognition accuracy of the vehicle license plate recognition device (increasing the rate of unknown recognition results). Furthermore, even at the same toll gate, some devices may be more susceptible to the effects of direct sunlight than others, depending on the lane in which they are installed. Therefore, if anomaly detection is performed using a threshold common to all devices (such as contrast or recognition accuracy thresholds), there is a possibility that anomalies or deterioration may be erroneously detected, even when both the vehicle license plate recognition camera and the vehicle license plate recognition device are actually operating normally. Similarly, the normal range of log data may differ depending on the influence of the inclination and width of the lane L, the presence or absence of curves, the presence or absence of vibrations, differences in characteristics due to the model and version, and the presence or absence of special specifications. For this reason, the anomaly detection model learning unit 22 according to this embodiment includes the tollgate master data D51 and the lane equipment master data D52 in the learning data, thereby allowing the anomaly detection model M1 to learn normal ranges or anomaly detection thresholds that differ for each lane equipment depending on the installation environment and the characteristics of each model and version. Furthermore, weather data (external data D24) may be included in the learning data, allowing the anomaly detection model M1 to learn changes in normal ranges and thresholds due to the influence of weather conditions.

[0050] Other examples of lane devices will also be described. Step boards and wireless communication devices installed on lanes L with heavy traffic volumes tend to be operated more frequently (number of times stepped on, frequency of communication) and are therefore more likely to deteriorate than those installed on lanes L with lighter traffic volumes. Furthermore, step boards installed on lanes L with a lot of large vehicles such as trucks tend to deteriorate more quickly than those installed on other lanes L. Therefore, the anomaly detection model learning unit 22 may cause the anomaly detection model M1 to learn normal ranges and thresholds according to the characteristics (tendencies) of the toll gate, such as the traffic volume and breakdown of vehicle types passing through, from the traffic volume data (external data D24) and the operating time and number of operations contained in the log data D22 of each device, and the vehicle type identification results contained in the log data D22 of other devices (vehicle type identification devices).

[0051] The anomaly detection model learning unit 22 may include the maintenance history data D31 to D33 collected by the failure analysis device 3 in the learning data. The maintenance history data D31 to D33 include information such as when and for which equipment anomalies were reported, and whether the anomalies were actually malfunctions. For example, the anomaly detection model learning unit 22 may identify the malfunctioning lane equipment (equipment identification ID) and the date and time of the report based on the repair history data D31 in which the response category is "replacement" or "repair," and perform supervised machine learning by labeling the log data D22 and D23 included in the period from the report date and time to the response completion date and time for this lane equipment as "malfunctioning." Similarly, the anomaly detection model learning unit 22 may identify the malfunctioning lane equipment (equipment identification ID) and the date and time of the report based on the repair history data D31 in which the response category is "no abnormality," and perform supervised machine learning by labeling the log data D22 and D23 included in the period from the report date and time to the response completion date and time for this lane equipment as "no abnormality." In this way, the anomaly detection model learning unit 22 can improve the accuracy of detecting anomalies for each lane device by automatically performing supervised learning, without, for example, having a maintenance worker prepare supervised data that has been labeled with the presence or absence of a malfunction.

[0052] (Example of central device processing: Operation phase) Fig. 7 is a flowchart showing an example of the operation phase of the central device according to the first embodiment. Here, the flow of processing in the operation phase of the central device 2 will be described with reference to Fig. 7. In the operation phase, it is assumed that the anomaly detection model M1 has already been trained.

[0053] First, the first data collection unit 21 acquires log data D22 and D23 from each lane device (step S111). The first data collection unit 21 may also acquire external data D24 including current weather data and traffic volume data from the external system 6.

[0054] Next, the monitoring unit 23 uses the trained anomaly detection model M1 to determine whether or not each lane device has a current anomaly (failure) or a sign of an anomaly (such as aging or malfunction) (step S112). When the log data D22 and D23 of each lane device are input, the anomaly detection model M1 outputs a detection result of whether or not an anomaly exists for each lane device (device identification ID). The anomaly detection model M1 may also receive external data D24 as input.

[0055] If the monitoring unit 23 detects an abnormality in any of the lane equipment (step S113; YES), it outputs the abnormality detection result to the operation console 24 and notifies the monitor (step S114). The abnormality detection result includes the date and time of detection, information for identifying the equipment that detected the abnormality (route number, toll gate number, lane number, equipment identification ID), details of the detected abnormality (type of abnormality, status such as serious or suspected), etc. The abnormality detection result is also recorded and accumulated as abnormality detection history data D21. The monitor refers to the notified abnormality detection result and performs remote operation of the lane equipment (such as closing the lane) or reports the abnormality to a maintenance worker.

[0056] On the other hand, if no abnormality is detected in any of the lane devices (step S113; NO), the monitoring unit 23 does not issue an abnormality detection notification and ends the process.

[0057] The central device 2 repeats the series of processes in FIG. 7 every time it acquires the log data D22, D23 or every fixed time (n seconds) to monitor and detect in real time whether or not there is an abnormality in each lane device.

[0058] (Example of failure analysis equipment processing: learning phase) Fig. 8 is a flowchart showing an example of the learning phase of the failure analysis device according to the first embodiment. Fig. 9 is a diagram for explaining the learning process of the failure analysis device according to the first embodiment. Here, the flow of the process of the learning phase of the failure analysis device 3 will be explained with reference to Figs. 8 and 9.

[0059] First, the second data collection unit 31 collects and stores maintenance history data D31 to D33 input by, for example, a maintenance worker via the mobile terminal 35 (step S201). The second data collection unit 31 may also collect and store abnormality detection history data D21, first log data D22, second log data D23, and external data D24 from the central device 2.

[0060] When the maintenance history data D31 to D33 are accumulated, the failure analysis model learning unit 32 performs machine learning of the failure analysis model M2 (step S202). For example, the failure analysis model learning unit 32 performs machine learning of the failure analysis model M2 every time a certain period of time passes or every time a certain amount of data is accumulated.

[0061] 9 shows an example of machine learning by the failure analysis model learning unit 32. For example, the failure analysis model learning unit 32 learns the correlation between the reported abnormality details (report history data D33) and the presence or absence of an abnormality, the cause of the abnormality, and the countermeasures taken (the response category and response details in the repair history data D31). The failure analysis model learning unit 32 also learns a failure analysis model M2 so that the cause of the abnormality and the countermeasures to be taken can be estimated from the correlation between the maintenance history data D31 to D33 and other data. The other data includes toll gate master data D51 and lane equipment master data D52. The other data may also include data D21 to D24 collected from the central device 2.

[0062] For example, the malfunctioning equipment may not be the direct cause. For example, if a barrier does not operate, this may be due to a malfunction of the barrier itself or a detection error of the sensor (vehicle detector) that triggers the barrier to operate. Furthermore, if a vehicle license plate number recognition (VLP) device malfunctions (unknown recognition result), this may be due to a malfunction of the VLP recognition device itself, poor image capture due to the installation environment (installation location and orientation, lane inclination and width, presence or absence of curves, presence or absence of vibration), time of day, season, or weather conditions, or dirt on the VLP recognition camera. Furthermore, the rate at which deterioration progresses may vary depending on the traffic volume and vehicle type. For this reason, the failure analysis model learning unit 32 according to this embodiment can train the failure analysis model M2 to predict potential causes of an abnormality and potential countermeasures according to the characteristics (tendencies) of the toll gate and the equipment by including the toll gate master data D51, lane equipment master data D52, external data D24, etc. in the learning data.

[0063] (Example of failure analysis equipment processing: Operation phase) Fig. 10 is a flowchart showing an example of the operation phase of the failure analysis device according to the first embodiment. Here, the flow of processing in the operation phase of the failure analysis device 3 will be described with reference to Fig. 10. In the operation phase, it is assumed that the failure analysis model M2 has already been trained.

[0064] First, the maintenance person inputs data on an unaddressed abnormality from the records recorded as, for example, the report history data D33, from the operation console 34, and queries the failure analysis device 3. Then, the second data collection unit 31 acquires the data on the abnormality of the lane equipment input by the maintenance person (step S211). The data on the abnormality includes the equipment identification ID of the lane equipment and the details of the abnormality or malfunction occurring in the lane equipment (e.g., "the crossing gate is not operating").

[0065] Next, the analysis unit 33 analyzes the candidate causes or candidate countermeasures for the abnormality using the trained failure analysis model M2 (step S212). When the device identification ID and the details of the abnormality are input, the failure analysis model M2 outputs the suspected candidate causes or candidate countermeasures.

[0066] The analysis unit 33 outputs the top n candidate causes or countermeasures estimated by the failure analysis model M2 to the console 34 and presents them to the maintenance technician (step S213). n is a preset number or the number of candidates that meet or exceed a preset probability. The maintenance technician prepares replacement parts (components), tools, cleaning tools, and other items according to the output candidate causes or countermeasures, and travels to the site where the lane equipment is installed to perform the necessary work. In conventional maintenance work, for example, when a report is received that a crossing gate is not operating, a maintenance technician travels to the site carrying replacement parts and tools for the crossing gate. However, if the crossing gate operation trigger is not output due to dirt on the vehicle detector rather than a crossing gate malfunction, the maintenance technician must return to retrieve the vehicle detector cleaning tools, which can delay the resolution of the abnormality. In contrast, in this embodiment, the failure analysis device 3 analyzes possible causes or possible countermeasures for the abnormality and presents them to the maintenance personnel, allowing them to prepare and carry in advance replacement parts, cleaning tools, and other items that are likely to be necessary to resolve the abnormality before heading to the site. This reduces the waste of time that would otherwise be incurred by the maintenance personnel having to return to get tools, and shortens the time it takes to recover from the abnormality.

[0067] (Example of maintenance planning device processing: learning phase) Fig. 11 is a flowchart showing an example of the learning phase of the maintenance planning device according to the first embodiment. Fig. 12 is a diagram for explaining the learning process of the maintenance planning device according to the first embodiment. Here, the flow of the process of the learning phase of the maintenance planning device 4 will be explained with reference to Figs. 11 and 12.

[0068] First, the third data collection unit 41 collects the maintenance history data D31 to D33 and also collects spare parts inventory data D41 from the failure analysis device 3 (step S301). The third data collection unit 41 may also collect external data D24 (weather data, traffic volume data) from the central device 2. The collected data is used as learning data.

[0069] When the maintenance history data D31 to D33, the spare parts inventory data D41, etc. are accumulated, the maintenance plan model learning unit 42 performs machine learning of the maintenance plan model M3 (step S302). For example, the maintenance plan model learning unit 42 performs machine learning of the maintenance plan model M3 every time a certain period of time passes or every time a certain amount of data is accumulated.

[0070] 12 shows an example of machine learning by the maintenance plan model learning unit 42. For example, the maintenance plan model learning unit 42 learns the timing of parts (spare parts) replacement and maintenance inspection based on past repair and replacement history, spare parts inventory status, past maintenance plans, the installation environment and configuration information (model, version number, presence or absence of special specifications) of each lane equipment, traffic volume, weather conditions, etc. from the maintenance history data D31 to D33, spare parts inventory data D41, past maintenance plan history data D42, toll booth master data D51, lane equipment master data D52, external data D24, etc. For example, lane equipment at toll booths with heavy traffic deteriorates more quickly than lane equipment at toll booths with light traffic, and the repair history data D31 should also include many replacement and repair histories. Therefore, the maintenance planning model learning unit 42 learns the maintenance planning model M3 so that for toll gates with heavy traffic or toll gates with many records of replacements and repairs in the repair history data D31, the frequency of replacement and maintenance inspection of lane equipment (or parts) is increased compared to other toll gates. In addition, the maintenance planning model learning unit 42 learns the timing of replenishment of spare parts from the past spare part inventory data D41.

[0071] (Example of maintenance planning device processing: Operation phase) Fig. 13 is a flowchart showing an example of the operation phase of the maintenance planning device 4 according to the first embodiment. Here, the processing flow of the operation phase of the maintenance planning device 4 will be described with reference to Fig. 13. In the operation phase, it is assumed that the maintenance planning model M3 has already been trained.

[0072] First, the maintenance planning person inputs, for example, a toll gate that is the target of the maintenance plan from the operation console 44, and instructs the maintenance planning device 4 to create a maintenance plan for this toll gate. Then, the third data collection unit 41 acquires information (for example, a toll gate number) of the toll gate that is the target of the plan input by the maintenance planning person (step S311). Multiple toll gate numbers (for example, all toll gate numbers on a route) may be input simultaneously as the target of the plan.

[0073] Next, the planning unit 43 uses the trained maintenance planning model M3 to create a maintenance plan and a spare parts replenishment plan for the specified toll gate (step S312). When the toll gate number is input, the maintenance planning model M3 outputs (generates) a maintenance plan proposal including the timing of maintenance inspections for this toll gate over a predetermined period (for example, one year) and the lane equipment to be replaced (equipment identification ID), as well as a replenishment plan proposal including the replenishment timing and number of spare parts to be replenished. Note that the predetermined period may be the period specified by the maintenance planner in step S311, and may be provided as an input (explanatory variable) to the maintenance planning model M3.

[0074] Furthermore, the planning unit 43 outputs to the operation console 44 a maintenance plan and a replenishment plan based on the maintenance plan and the replenishment plan generated by the maintenance plan model M3, and presents them to the maintenance planning staff (step S313). The maintenance plan and the replenishment plan display, for example, on a calendar, the scheduled maintenance and inspection dates for each toll gate, the lane equipment to be replaced (equipment identification ID), the replenishment date (order date) of spare parts, and the number of parts to be replenished. The maintenance planning staff may manually revise the maintenance plan and the replenishment plan as necessary, or may change the conditions (toll gate number and planning period) and have the maintenance planning device 4 re-create the maintenance plan and the replenishment plan.

[0075] (Action, effect) As described above, the remote monitoring system 1 according to this embodiment includes a data server 5 that stores master data D51, D52 including installation environment and configuration information for lane equipment; a first data collection unit 21 that collects log data D22, D23 related to the operation or processing of the lane equipment; an anomaly detection model learning unit 22 that uses learning data including the log data D22, D23 and the master data D51, D52 to learn an anomaly detection model M1 in which the log data D22, D23 are used as explanatory variables and the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is used as a target variable; and a monitoring unit 23 that inputs the log data D22, D23 into the anomaly detection model M1, and, when a detection result indicating the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is output, notifies a monitor of the detection result.

[0076] In this way, the remote monitoring system 1 can monitor the log data D22, D23 of each lane device in real time to detect the presence or absence of abnormalities or signs of abnormalities. Also, by including information about the installation environment and configuration of the lane devices in the learning data, normal ranges and thresholds that serve as standards for abnormality detection can be learned individually according to the characteristics (tendencies) of each toll booth and lane device. As a result, the remote monitoring system 1 can accurately detect abnormalities in each lane device.

[0077] Furthermore, the first data collection unit 21 further collects weather data for the area including the toll gate from the external system 6, and the learning data and explanatory variables of the anomaly detection model M1 further include the weather data.

[0078] In this way, the remote monitoring system 1 can make the abnormality detection model M1 learn about changes in normal ranges and thresholds due to the influence of weather conditions, thereby enabling more accurate detection of abnormalities in each lane device.

[0079] The remote monitoring system 1 also includes a second data collection unit 31 that collects maintenance history data D31 to D33 including information on abnormalities, repairs, and maintenance inspections that have occurred in the lane equipment; a failure analysis model learning unit 32 that uses learning data including the maintenance history data D31 to D33 and master data D51 and D52 to learn a failure analysis model M2 that uses the abnormality details of the lane equipment as explanatory variables and analysis results including potential causes or countermeasures for the abnormality details as objective variables; and an analysis unit 33 that receives inquiries about the abnormality details that have occurred in the lane equipment, inputs the abnormality details into the failure analysis model M2, and notifies a maintenance technician of the analysis results of the abnormality details obtained as output.

[0080] In this way, the remote monitoring system 1 can notify the maintenance personnel of possible causes of the abnormality or possible countermeasures, thereby supporting the maintenance personnel to prepare and carry replacement parts, cleaning tools, etc. that are likely to be necessary to resolve the abnormality before heading to the site. This reduces the waste of time that maintenance personnel spend returning to get tools, and shortens the time it takes to recover from the abnormality.

[0081] In addition, the first data collection unit 21 further collects maintenance history data D31 to D33, and the learning data for the anomaly detection model M1 further includes the maintenance history data D31 to D33. The anomaly detection model learning unit 22 generates supervised data in which the log data D22, D23 of the lane equipment is labeled as having an abnormality or not from a combination of the log data D22, D23 and the maintenance history data D31 to D33, and learns the anomaly detection model M1.

[0082] In this way, the remote monitoring system 1 can improve the accuracy of detecting anomalies for each lane device by automatically performing supervised learning, without the need for a maintenance worker to prepare supervised data labeled with the presence or absence of a malfunction, for example.

[0083] The remote monitoring system 1 also includes a third data collection unit 41 that collects spare part inventory data D41 for lane equipment and maintenance history data D31 to D33; a maintenance planning model learning unit 42 that uses learning data including the spare part inventory data D41, the maintenance history data D31 to D33, and master data D51 and D52 to learn a maintenance planning model M3 that uses the toll gate as an explanatory variable and a maintenance planning plan including the inspection timing or replacement target lane equipment of the toll gate as an objective variable; and a planning unit 43 that inputs the toll gate to be planned into the maintenance planning model M3 and creates a maintenance plan for the toll gate based on the maintenance planning plan obtained as an output.

[0084] In this way, the remote monitoring system 1 can analyze the history of past failures and replacements, and appropriately plan maintenance and inspection for each toll gate.

[0085] Furthermore, the planning unit 43 further creates a replenishment plan for spare parts using the maintenance plan model M3.

[0086] In this way, the remote monitoring system 1 can analyze the history of past failures and replacements, accurately predict the number and timing of spare parts required at each toll booth, and propose a replenishment plan that can prevent excess or shortage of spare parts to be replenished.

[0087] <Hardware configuration> FIG. 14 is a diagram showing an example of the hardware configuration of the central unit, the failure analysis unit, and the maintenance planning unit. The computer 900 includes a CPU 901, a main storage device 902, an auxiliary storage device 903, an input / output interface 904, and a communication interface 905. The central unit 2, the failure analysis device 3, and the maintenance planning device 4 described above are each implemented in the computer 900. The functions described above are stored in the auxiliary storage device 903 in the form of a program. The CPU 901 reads the program from the auxiliary storage device 903, loads it into the main storage device 902, and executes the above processing in accordance with the program. The CPU 901 also allocates a storage area in the main storage device 902 in accordance with the program. The CPU 901 also allocates a storage area in the auxiliary storage device 903 for storing data being processed in accordance with the program.

[0088] Alternatively, a program for implementing all or part of the functions of the central unit 2, the failure analysis unit 3, and the maintenance planning unit 4 may be recorded on a computer-readable recording medium, and the program may be loaded into a computer system and executed to perform processing by each functional unit. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. Furthermore, if a WWW system is used, the term "computer system" also includes a homepage provision environment (or display environment). Furthermore, the term "computer-readable recording medium" refers to portable media such as CDs, DVDs, and USBs, as well as storage devices such as hard disks built into the computer system. Furthermore, if the program is distributed to the computer 900 via a communication line, the computer 900 that receives the program may load the program into the main storage device 902 and execute the above-described processing. Furthermore, the program may be for implementing part of the above-described functions, or may be capable of implementing the above-described functions in combination with a program already stored in the computer system.

[0089] <Other embodiments> Although one embodiment has been described in detail above with reference to the drawings, the specific configuration is not limited to the above, and various design modifications are possible. That is, in other embodiments, the order of the above-described processes may be changed as appropriate. Furthermore, some processes may be executed in parallel.

[0090] <Additional Notes> The remote monitoring system and the remote monitoring method described in the above-described embodiment can be understood, for example, as follows.

[0091] (1) According to the first aspect, the remote monitoring system 1 is a remote monitoring system 1 that remotely monitors lane equipment installed in a lane L of a toll gate, and includes: a data server 5 that stores master data D51, D52 including installation environment and configuration information of the lane equipment; a first data collection unit 21 that collects log data D22, D23 related to the operation or processing of the lane equipment; an anomaly detection model learning unit 22 that uses learning data including the log data D22, D23 and the master data D51, D52 to learn an anomaly detection model M1 that uses the log data D22, D23 as explanatory variables and the presence or absence of an abnormality or a sign of an abnormality in the lane equipment as a target variable; and a monitoring unit 23 that inputs the log data D22, D23 into the anomaly detection model M1, and when a detection result indicating the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is output, notifies a monitor of the detection result.

[0092] In this way, the remote monitoring system 1 can monitor the log data D22, D23 of each lane device in real time to detect the presence or absence of abnormalities or signs of abnormalities. Also, by including information about the installation environment and configuration of the lane devices in the learning data, normal ranges and thresholds that serve as standards for abnormality detection can be learned individually according to the characteristics (tendencies) of each toll booth and lane device. As a result, the remote monitoring system 1 can accurately detect abnormalities in each lane device.

[0093] (2) According to the second aspect, in the remote monitoring system 1 relating to the first aspect, the first data collection unit 21 further collects weather data for the area including the toll gate from the external system 6, and the learning data and explanatory variables of the anomaly detection model M1 further include weather data.

[0094] In this way, the remote monitoring system 1 can make the abnormality detection model M1 learn about changes in normal ranges and thresholds due to the influence of weather conditions, thereby enabling more accurate detection of abnormalities in each lane device.

[0095] (3) According to the third aspect, the remote monitoring system 1 according to the first or second aspect further includes a second data collection unit 31 that collects maintenance history data D31 to D33 including information on abnormalities, repairs, and maintenance inspections that have occurred in the lane equipment; a failure analysis model learning unit 32 that uses learning data including the maintenance history data D31 to D33 and master data D51 and D52 to learn a failure analysis model M2 that uses abnormality details of the lane equipment as explanatory variables and analysis results including potential causes or potential countermeasures for the abnormality details as objective variables; and an analysis unit 33 that receives inquiries about abnormality details that have occurred in the lane equipment, inputs the abnormality details into the failure analysis model M2, and notifies a maintenance technician of the analysis results of the abnormality details obtained as an output.

[0096] In this way, the remote monitoring system 1 can notify the maintenance personnel of possible causes of the abnormality or possible countermeasures, thereby supporting the maintenance personnel to prepare and carry replacement parts, cleaning tools, etc. that are likely to be necessary to resolve the abnormality before heading to the site. This reduces the waste of time that maintenance personnel spend returning to get tools, and shortens the time it takes to recover from the abnormality.

[0097] (4) According to the fourth aspect, in the remote monitoring system 1 relating to the third aspect, the first data collection unit 21 further collects maintenance history data D31 to D33, the learning data of the anomaly detection model M1 further includes the maintenance history data D31 to D33, and the anomaly detection model learning unit 22 generates supervised data in which the log data D22, D23 of the lane equipment is labeled as having or not having an abnormality from a combination of the log data D22, D23 and the maintenance history data D31 to D33, and learns the anomaly detection model M1.

[0098] In this way, the remote monitoring system 1 can improve the accuracy of detecting anomalies for each lane device by automatically performing supervised learning, without the need for a maintenance worker to prepare supervised data labeled with the presence or absence of a malfunction, for example.

[0099] (5) According to the fifth aspect, the remote monitoring system 1 according to any one of the first to fourth aspects further includes a third data collection unit 41 that collects spare part inventory data D41 of lane equipment and maintenance history data D31 to D33; a maintenance planning model learning unit 42 that uses learning data including the spare part inventory data D41, the maintenance history data D31 to D33, and master data D51 and D52 to learn a maintenance planning model M3 in which the toll gate is used as an explanatory variable and a maintenance planning plan including the inspection timing of the toll gate and the lane equipment to be replaced is used as an objective variable; and a planning unit 43 that inputs the toll gate to be planned into the maintenance planning model M3 and creates a maintenance plan for the toll gate based on the maintenance planning plan obtained as an output.

[0100] In this way, the remote monitoring system 1 can analyze the history of past failures and replacements, and appropriately plan maintenance and inspection for each toll gate.

[0101] (6) According to the sixth aspect, in the remote monitoring system 1 according to the fifth aspect, the planner 43 further creates a replenishment plan for spare parts using the maintenance plan model M3.

[0102] In this way, the remote monitoring system 1 can analyze the history of past failures and replacements, accurately predict the number and timing of spare parts required at each toll booth, and propose a replenishment plan that can prevent excess or shortage of spare parts to be replenished.

[0103] (7) According to the seventh aspect, a remote monitoring method is a remote monitoring method for remotely monitoring lane equipment installed in lanes of a toll gate, the remote monitoring method comprising the steps of: collecting log data D22, D23 relating to the operation or processing of the lane equipment; using learning data including the log data D22, D23 and master data D51, D52 including installation environment and configuration information of the lane equipment recorded in a data server 5, learning an anomaly detection model M1 in which the log data D22, D23 are used as explanatory variables and the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is used as a target variable; and inputting the log data D22, D23 into the anomaly detection model M1, and when a detection result indicating the presence or absence of an abnormality or a sign of an abnormality in the lane equipment is output, notifying a monitor of the detection result. [Explanation of symbols]

[0104] 1. Remote monitoring system 2 Central unit 21 First Data Collection Unit 22 Anomaly detection model learning unit 23 Monitoring Department 24 Control console 3 Failure analysis device 31 Second Data Collection Unit 32 Failure analysis model learning section 33 Analysis Department 4 Maintenance planning device 41 Data Collection Unit 3 42 Maintenance Planning Model Learning Section 43 Planning Department 5 Data Server 6 External Systems D21 Anomaly detection history data D22 1st log data D23 2nd log data D24 External Data D31 Repair history data (maintenance history data) D32 Maintenance and inspection history data (maintenance history data) D33 Report history data (maintenance history data) D41 Spare Parts Inventory Data D42 Maintenance plan history data D51 Tollgate Master Data D52 Lane Equipment Master Data M1 anomaly detection model M2 Failure Analysis Model M3 Maintenance Planning Model

Claims

1. A remote monitoring system for remotely monitoring lane equipment installed in lanes of a toll booth, a data server that stores master data including information on the installation environment and configuration of the lane devices; a first data collection unit that collects log data relating to the operation or processing of the lane device; an anomaly detection model learning unit that uses learning data including the log data and the master data to learn an anomaly detection model having the log data as an explanatory variable and the presence or absence of an abnormality or a sign of an abnormality in the lane device as a target variable; a monitoring unit that inputs the log data into the anomaly detection model, and when a detection result indicating that there is an abnormality or a sign of an abnormality in the lane equipment is output, notifies an observer of the detection result; A remote monitoring system comprising:

2. The first data collection unit further collects weather data for an area including the toll booth from an external system; the learning data and the explanatory variables of the anomaly detection model further include the weather data. The remote monitoring system of claim 1 .

3. a second data collection unit that collects maintenance history data including information on abnormalities, repairs, and maintenance inspections that have occurred in the lane equipment; a failure analysis model learning unit that uses learning data including the maintenance history data and the master data to learn a failure analysis model in which an abnormality content of the lane device is an explanatory variable and an analysis result including a candidate cause or candidate countermeasure for the abnormality content is an objective variable; an analysis unit that receives an inquiry about an abnormality occurring in the lane device, inputs the abnormality into the failure analysis model, and notifies a maintenance worker of an analysis result of the abnormality obtained as an output; The remote monitoring system according to claim 1 or 2, further comprising:

4. the first data collection unit further collects the maintenance history data; the learning data of the anomaly detection model further includes the maintenance history data; the anomaly detection model learning unit generates supervised data in which the log data of the lane device is labeled as having an abnormality or not, from a combination of the log data and the maintenance history data, and learns the anomaly detection model; The remote monitoring system according to claim 3 .

5. a third data collection unit that collects spare part inventory data of the lane equipment and the maintenance history data; a maintenance planning model learning unit that uses learning data including the spare parts inventory data, the maintenance history data, and the master data to learn a maintenance planning model in which a toll gate is used as an explanatory variable and a maintenance planning plan including an inspection time of the toll gate and a lane device to be replaced is used as an objective variable; a planning unit that inputs a tollgate to be planned into the maintenance planning model and creates a maintenance plan for the tollgate based on a maintenance plan proposal obtained as an output; The remote monitoring system of claim 3 further comprising:

6. The planning unit further creates a replenishment plan for spare parts using the maintenance planning model. The remote monitoring system according to claim 5 .

7. A remote monitoring method for remotely monitoring lane equipment installed in lanes of a toll booth, comprising: collecting log data relating to the operation or processing of said lane equipment; using learning data including the log data and master data including installation environment and configuration information of the lane devices recorded in a data server, to learn an anomaly detection model in which the log data is used as an explanatory variable and the presence or absence of an abnormality or a sign of an abnormality in the lane devices is used as a target variable; inputting the log data into the anomaly detection model, and when a detection result indicating that there is an anomaly or a sign of an anomaly in the lane equipment is output, notifying a traffic monitor of the detection result; A remote monitoring method comprising:

Images