Data analysis device, data analysis method, and program

Abstract

To provide a data analysis device capable of obtaining an analysis result even in a newly introduced facility.SOLUTION: A data analysis device 10-1 includes a data acquisition unit 11 acquiring data of an analytic object, a dependence relation calculation unit 16 calculating a dependence relation between a plurality of data items included in the acquired data of the analytic object, and an abnormality probability calculation unit 17 calculating an abnormality probability of the data items based on the dependence relation. The abnormality probability calculation unit 17 calculates an abnormality probability by using different calculation methods for each type of the data items.SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a data analysis apparatus, a data analysis method, and a program.

Background Art

[0002] Techniques for detecting abnormalities in a monitoring target system by analyzing data indicating the state of the monitoring target system are widely used. For example, Patent Document 1 discloses a monitoring system that determines the occurrence of an abnormality in a monitoring target system based on measurement data acquired from the monitoring target system. In this monitoring system, the determination of the occurrence of an abnormality in the monitoring target system is made using a model showing the dependency relationship between data, called a causal structure model of the monitoring target system.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, in the technique disclosed in Patent Document 1, since the presence or absence of the occurrence of an abnormality is output as an analysis result, it is necessary to set in advance an abnormality determination criterion for determining the occurrence of an abnormality, and there is a problem that an analysis result cannot be obtained for newly introduced equipment.

[0005] The present disclosure has been made in view of the above, and an object thereof is to obtain a data analysis apparatus capable of obtaining an analysis result even for newly introduced equipment.

Means for Solving the Problems

[0006] To solve the above-mentioned problems and achieve the objective, the data analysis device according to this disclosure comprises a data acquisition unit that acquires data to be analyzed, a dependency relationship calculation unit that calculates the dependency relationships between multiple data items included in the acquired data to be analyzed, and an anomaly probability calculation unit that calculates the anomaly probability of a data item based on the dependency relationships, wherein the anomaly probability calculation unit calculates the anomaly probability using a different calculation method for each type of data item. The data items include measurement data, which is data related to the target device; environmental data, which indicates the state of the environment around the target device; and operation data, which indicates the operation performed on the target device. The target device is a railway vehicle, and the measurement data includes vehicle operation data acquired from equipment mounted on the vehicle while the vehicle is running, and inspection data acquired when the equipment is inspected. It is characterized by the following: [Effects of the Invention]

[0007] According to this disclosure, it is possible to obtain analysis results even with newly installed equipment. [Brief explanation of the drawing]

[0008] [Figure 1] Diagram illustrating the data used by the data analysis device according to Embodiments 1 to 5. [Figure 2] Diagram showing the functional configuration of the data analysis device according to Embodiment 1. [Figure 3] Diagram illustrating the dependencies between data items. [Figure 4] Figure 2 shows an example of measurement data stored in the measurement data storage unit. [Figure 5] Figure 2 shows an example of environmental data stored in the environmental data storage unit. [Figure 6] Figure 2 shows an example of operating data stored in the operating data storage unit. [Figure 7] Figure 2 shows an example of the organization data stored in the organization data storage unit. [Figure 8] Figure 2 shows an example of the dependency relationships calculated by the dependency calculation unit. [Figure 9] Figure 2 shows an example of a normal distribution calculated by the abnormal probability calculation unit. [Figure 10] Flowchart illustrating the operation of the dependency calculation unit shown in Figure 2. [Figure 11] Figure 2 is a flowchart illustrating the operation of the anomaly probability calculation unit. [Figure 12] Figure showing the functional configuration of the data analysis device according to Embodiment 2 [Figure 13] Figure showing a first example of the inspection plan formulated by the inspection plan formulation unit shown in FIG. 12 [Figure 14] Figure showing a second example of the inspection plan formulated by the inspection plan formulation unit shown in FIG. 12 [Figure 15] Flowchart for explaining the operation of the inspection plan formulation unit shown in FIG. 12 [Figure 16] Figure showing the functional configuration of the data analysis device according to Embodiment 3 [Figure 17] Figure showing an example of the route data stored in the route data storage unit shown in FIG. 16 [Figure 18] Figure showing an example of the operation plan generated by the operation plan formulation unit shown in FIG. 16 [Figure 19] Figure showing the functional configuration of the data analysis device according to Embodiment 4 [Figure 20] Figure showing an example of the operation plan generated by the operation plan formulation unit shown in FIG. 19 [Figure 21] Figure showing the functional configuration of the data analysis device according to Embodiment 5 [Figure 22] Figure showing an example of the driver data stored in the driver data storage unit shown in FIG. 21 [Figure 23] Figure showing an example of the information indicating the driver skill generated by the driver skill prediction unit shown in FIG. 21 [Figure 24] Figure showing an example of the operation plan generated by the operation plan formulation unit shown in FIG. 22 [Figure 25] Figure showing a configuration example of a computer system for realizing the data analysis device according to Embodiments 1 to 5

Mode for Carrying Out the Invention

[0009] Hereinafter, the data analysis device, data analysis method, and program according to the embodiments of the present disclosure will be described in detail based on the drawings.

[0010] Embodiment 1. Figure 1 is an explanatory diagram of the data used by the data analysis device 10 according to Embodiments 1 to 5. The data analysis devices 10 according to Embodiments 1 to 5 described below will be distinguished as data analysis devices 10-1 to 10-5. When there is no need to distinguish between data analysis devices 10-1 to 10-5, they may simply be referred to as data analysis device 10.

[0011] The data analysis device 10 acquires vehicle running data, which is data acquired from the railway train 1 via the acquisition device 2 while the train 1 is running. The vehicle running data includes data acquired by sensors (not shown) attached to the train 1, and train 1 operation data. The data analysis device 10 can also acquire equipment inspection data entered by workers using the inspection terminal 3 when inspecting the equipment of the train 1 at a train depot, and environmental data obtained via a communication network 4 such as the internet. The data analysis device 10 can collect data, analyze the collected data, and output the analysis results.

[0012] Here, the data collected by the data analysis device 10 is referred to as vehicle running data, equipment inspection data, and environmental data. The data analysis device 10 classifies this data into measurement data, which is data related to train 1; environmental data, which shows the environment around train 1; or driving data, which shows the operations entered into train 1 by the driver operating train 1. In other words, vehicle running data may include measurement data, environmental data, and driving data. Also, equipment inspection data may include measurement data and environmental data.

[0013] Furthermore, this section describes a data analysis device 10 that analyzes data related to railway train 1, but the train 1 is just one example of a target device. The target device could be a vehicle other than a train, or a device other than a vehicle. Also, there are no restrictions on the size of the target device; it could be a device smaller than train 1, or a large-scale device generally referred to as equipment.

[0014] Figure 2 shows the functional configuration of the data analysis device 10-1 according to Embodiment 1. The data analysis device 10-1 includes a data acquisition unit 11, a measurement data storage unit 12, an environmental data storage unit 13, an operation data storage unit 14, a train formation data storage unit 15, a dependency relationship calculation unit 16, an abnormality probability calculation unit 17, and an analysis result output unit 18.

[0015] The data analysis device 10-1 calculates the anomaly probability, which indicates the degree of anomaly in the data items of the data to be analyzed. In this case, the data analysis device 10-1 calculates the anomaly probability based on the dependencies between data items.

[0016] Figure 3 is an explanatory diagram of the dependencies between data items. For example, let's consider the pantograph, inverter, motor, and wheels that make up train 1. These components have dependencies, and if there is a malfunction in the operation of a higher-level component, the operation of lower-level components may also appear abnormal. In the example shown in Figure 3, the dependencies are in the order of pantograph, inverter, motor, and wheels, from the highest-level component upwards.

[0017] For example, suppose that only the inverter among these components is actually malfunctioning. In this case, since the pantograph is a higher-level component than the inverter, it is not affected by the inverter malfunction, and the data items related to the pantograph appear normal. However, because the inverter is malfunctioning, the data items related to the inverter will show abnormal values. At this point, the data items related to the motor, which is lower-level than the inverter, will also appear abnormal due to the inverter malfunction. Similarly, the data items related to the wheels, which are lower-level than the motor, will also show abnormal values ​​because they are affected by the inverter malfunction via the motor. Thus, even if only the inverter is actually malfunctioning, the motor and wheels may also appear abnormal. In this case, if we only look at whether the data item values ​​appear abnormal or not, we may end up inspecting the motor or wheels, which are not actually malfunctioning, even if it would be better to prioritize inspecting the inverter.

[0018] The data analysis device 10-1 can provide an anomaly probability that reflects the true cause of an anomaly by calculating the anomaly probability based on the dependencies between data items. The dependencies between data items used in this case include not only hierarchical relationships but also a degree of dependence that indicates how much each item influences the value of the data item. In the example in Figure 3, the degree of dependence between the pantograph and the inverter is 30%, the degree of dependence between the inverter and the motor is 50%, and the degree of dependence between the motor and the wheel is 40%.

[0019] Returning to the explanation of Figure 2, the data acquisition unit 11 acquires the data to be analyzed. The data acquisition unit 11 stores the acquired data in one of the following: the measurement data storage unit 12, the environmental data storage unit 13, the operation data storage unit 14, or the train formation data storage unit 15. Here, the data acquisition unit 11 will classify the acquired data by type of data item and output it.

[0020] The measurement data storage unit 12 stores the measurement data acquired by the data acquisition unit 11. The measurement data is data related to the vehicles of train 1. Figure 4 shows an example of the measurement data stored in the measurement data storage unit 12 shown in Figure 2. Here, for each train set ID (IDentifier) ​​that identifies the train set of train 1, the number of wheel scratches and the motor voltage are stored in association with the acquisition time of each measurement data. Note that the measurement data shown here is just an example, and the measurement data can be any data related to the vehicles of train 1. The measurement data includes not only data acquired from equipment mounted on the vehicle, but also data acquired inside and outside the vehicle while the vehicle is running. For example, the measurement data includes the current value of the motor mounted on the vehicle, the amount of vibration of the vehicle, the amount of vibration of the rails caused by the vehicle running, image data from cameras that photograph the inside and outside of the vehicle, and audio data that captures sounds generated while the vehicle is running.

[0021] The environmental data storage unit 13 stores the environmental data acquired by the data acquisition unit 11. The environmental data is data that indicates the state of the environment around train 1. Figure 5 is a diagram showing an example of the environmental data stored in the environmental data storage unit 13 shown in Figure 2. Here, for each train set of train 1, the route ID, gradient, and precipitation are stored in association with the acquisition time of each environmental data. The route ID is information that identifies the route that train 1 was traveling on at the time of acquisition. The gradient is information that indicates the gradient of the track that train 1 was traveling on at the time of acquisition. The precipitation is information that indicates the amount of rainfall at the location where train 1 was traveling at the time of acquisition.

[0022] The operation data storage unit 14 stores the operation data acquired by the data acquisition unit 11. The operation data is data indicating the operation performed on train 1. Figure 6 shows an example of operation data stored in the operation data storage unit 14 shown in Figure 2. Here, for each driver ID that identifies the driver operating train 1, the assigned train set and notch are stored in association with the acquisition time when each operation data was acquired. The assigned train set is information indicating the train set that the driver identified by the driver ID was in charge of at the time of acquisition. The notch is information indicating the handle position of the master controller at the time of acquisition.

[0023] The train formation data storage unit 15 stores the train formation data acquired by the data acquisition unit 11. The train formation data is information indicating the formation of train 1. Figure 7 shows an example of the train formation data stored by the train formation data storage unit 15 shown in Figure 2. Here, for each formation of train 1, information identifying the cars that make up train 1 is stored in association with the acquisition time when each train formation data was acquired. In the example in Figure 7, the train formation data includes information identifying the first car, second car, and third car.

[0024] The dependency calculation unit 16 calculates the dependencies between multiple data items included in the data to be analyzed. The dependency calculation unit 16 can learn the dependencies between data items using sparse modeling such as LASSO regression, for example, using the history of the data acquired by the data acquisition unit 11. For example, the m-th data item is represented by the following equation (1), where w in equation (1) 1~N This defines the dependency. For simplicity, the m-th data item will be referred to as [item m] below.

[0025] [Item m]=w1×[Item 1]+···+w m-1 ×[item m-1]+w m+1 ×[item m+1]+···w N ×[Item N]...(1)

[0026] Figure 8 shows an example of a dependency relationship calculated by the dependency relationship calculation unit 16 shown in Figure 2. The subject in Figure 8 corresponds to item m on the left side of formula (1), and the object in Figure 8 corresponds to data items other than item m on the right side of formula (1). Each element of the table shown in Figure 8 is a dependency relationship w 1~N This corresponds to the example shown in Figure 8, where the relationship [wheel damage] = 10% × [gradient] holds true.

[0027] The dependency relationship calculation unit 16 outputs the calculated dependency relationship information to the anomaly probability calculation unit 17.

[0028] The anomaly probability calculation unit 17 calculates an anomaly probability indicating the degree of anomaly for each data item based on the dependency relationships calculated by the dependency relationship calculation unit 16. The anomaly probability calculation unit 17 can output the calculated anomaly probabilities to the analysis result output unit 18.

[0029] The anomaly probability calculation unit 17 calculates a predicted value for each of the acquisition history of data item m for each of m=1 to N, using the dependency relationships of data item m and the history of data other than data item m. The anomaly probability calculation unit 17 can obtain the predicted value of data item m using, for example, the following formula (2). Here, the measured value refers to the data acquired by the data acquisition unit 11.

[0030] [Predicted value of item m] = w1 × [Actual value of item 1] + ... + w m-1 ×[Measured value of item m-1]+w m+1 ×[Measured value of item m+1]+···+w N ×[Measured value of item N]···(2)

[0031] The anomaly probability calculation unit 17 calculates the difference between the predicted value and the actual value for each of m=1 to N, and calculates a normal distribution from the calculated difference. Figure 9 shows an example of the normal distribution calculated by the anomaly probability calculation unit 17 shown in Figure 2. The larger the difference between the predicted value and the actual value, the higher the anomaly probability. Here, the normal distribution shown in Figure 9 is defined as the distribution of anomaly probabilities. Based on the difference between the predicted value and the actual value and the distribution of anomaly probabilities, the anomaly probability calculation unit 17 can calculate the apparent anomaly probability for each data item.

[0032] The anomaly probability calculation unit 17 calculates the anomaly probability for each data item based on the calculated apparent anomaly probability and the dependency relationships. At this time, the anomaly probability calculation unit 17 can calculate the anomaly probability using different calculation methods for each type of data item. For example, for measurement data, the anomaly probability calculation unit 17 can calculate the anomaly probability for each data item by distributing the apparent anomaly probability of each data item to higher-level dependency relationships according to their degree of dependence.

[0033] Let's explain this again using the example in Figure 3. In the example shown in Figure 3, the probability of abnormality for each of the wheels, motor, inverter, and pantograph can be calculated using the method described below. Wheel anomaly probability = Apparent anomaly probability Motor malfunction probability = Apparent motor malfunction probability + Apparent wheel malfunction probability × 40% Inverter malfunction probability = Apparent inverter malfunction probability + Motor apparent malfunction probability × 50% + Wheel apparent malfunction probability × 40% × 50% Pantograph malfunction probability = Apparent pantograph malfunction probability + Apparent inverter malfunction probability × 30% + Apparent motor malfunction probability × 50% × 30% + Apparent wheel malfunction probability × 40% × 50% × 30%

[0034] The abnormality probability calculation unit 17 may take the average of the abnormality probabilities for each data item over the driving time and use the average value as the abnormality probability for that data item.

[0035] Furthermore, if the data item is environmental data, the anomaly probability calculation unit 17 can use the average value of the anomaly probabilities calculated by distributing apparent anomaly probabilities according to the degree of dependence as described above for each car of train 1 as the anomaly probability of that data item. Also, if the data item is operational data, the anomaly probability calculation unit 17 can use the average value of the anomaly probabilities calculated by distributing apparent anomaly probabilities according to the degree of dependence as described above for each driver as the anomaly probability of that data item. If the cause of the anomaly lies in the operating environment of train 1, anomalies should be observed in all cars that have operated under the same environment. Therefore, by averaging the environmental data for each car, it becomes possible to determine whether or not the cause of the anomaly lies in the operating environment. Also, if the cause of the anomaly lies with the driver, anomalies should be observed in all cars that have operated with the same driver. Therefore, by averaging the operational data for each driver, it becomes possible to determine whether or not the cause of the anomaly lies with the driver.

[0036] The analysis result output unit 18 outputs the analysis results of the data analysis device 10-1. For example, the analysis result output unit 18 can output the abnormal probability calculated by the abnormal probability calculation unit 17 as the analysis result. The form in which the analysis result output unit 18 outputs the analysis results is not limited. For example, the analysis result output unit 18 may generate a display screen that includes the abnormal probability calculated by the abnormal probability calculation unit 17 and output the generated display screen, or it may output the analysis results as audio, or it may output the analysis results as text data.

[0037] Figure 10 is a flowchart illustrating the operation of the dependency calculation unit 16 shown in Figure 2. The dependency calculation unit 16 repeats the following steps S101 and S102 while varying the value of i from 1 to the number of organization items, and for each organization i, while varying the value of j from 1 to the number of data items in organization i.

[0038] The dependency calculation unit 16 derives the dependency relationships between item j of organization i and each data item other than item j of organization i (step S101). The dependency calculation unit 16 derives the anomalous probability distribution of item j of organization i from the dependency relationships derived in step S101 (step S102).

[0039] Figure 11 is a flowchart illustrating the operation of the anomaly probability calculation unit 17 shown in Figure 2. The anomaly probability calculation unit 17 initializes all apparent anomaly probabilities to 0. The apparent anomaly probability of data item j in group i is denoted as apparent anomaly probability [i][j]. i is a value greater than or equal to 1 and less than or equal to the number of groups, and j is a value greater than or equal to 1 and less than or equal to the number of measured data items in group i (step S201).

[0040] The abnormal probability calculation unit 17 changes the value of i from 1 to the number of formations, and for each formation i, it changes the value of j from 1 to the number of data items in formation i, and for each data item j, it changes the value of time k from 1 to the number of data items in formation i, and for each time k, it repeats the process from step S202 to step S204 shown below.

[0041] The anomaly probability calculation unit 17 derives an estimated value of item j of organization i at time k, based on the dependency relationships of item j of organization i, using the measured values ​​of items other than item j of organization i at time k (step S202).

[0042] The anomaly probability calculation unit 17 derives the apparent anomaly probability of item j in organization i at time k based on the anomaly probability distribution of item j in organization i (step S203). The anomaly probability calculation unit 17 calculates "apparent anomaly probability [i][j] += apparent anomaly probability of item j in organization i at time k" (step S204). After the processing from step S202 to step S204 is completed for all times k, the anomaly probability calculation unit 17 calculates "apparent anomaly probability [i][j] = apparent anomaly probability [i][j] / estimated range of item j in organization i" (step S205). The anomaly probability calculation unit 17 repeats the processing from step S202 to step S205 for all items j, and further repeats the processing from step S202 to step S205 for all organization i.

[0043] As described above, the data analysis device 10-1 according to Embodiment 1 comprises a data acquisition unit 11 that acquires data to be analyzed, a dependency relationship calculation unit 16 that calculates the dependency relationships between multiple data items included in the acquired data to be analyzed, and an abnormality probability calculation unit 17 that calculates the abnormality probability of a data item based on the dependency relationships. The abnormality probability calculation unit 17 is characterized by calculating the abnormality probability using a different calculation method for each type of data item. With this configuration, the data analysis device 10-1 can present an abnormality probability that quantifies the degree of abnormality, rather than simply indicating whether the data to be analyzed is normal or abnormal. Therefore, there is no need to set criteria for abnormality, and analysis results can be obtained even with newly installed equipment. Furthermore, since the abnormality probability is calculated using a different calculation method for each type of data item, it becomes possible to calculate an abnormality probability that allows for the identification of the cause of the abnormality.

[0044] The types of data items can include measurement data, which is data related to the train 1 that is the target device; environmental data, which indicates the state of the environment around the target device; and operation data, which indicates the operation performed on the target device. Furthermore, the abnormality probability calculation unit 17 can calculate the abnormality probability for each target device when the type of data item is environmental data, and can calculate the abnormality probability for each driver operating the target device when the type of data item is operation data.

[0045] The anomaly probability calculation unit 17 can calculate the anomaly probability based on the difference between the predicted value of the data item calculated based on the dependency relationship and the measured value, which is the data to be analyzed.

[0046] Embodiment 2. Figure 12 shows the functional configuration of the data analysis device 10-2 according to Embodiment 2. In addition to the configuration of the data analysis device 10-1, the data analysis device 10-2 has an inspection planning unit 19.Hereafter, detailed explanations of parts that are the same as the data analysis device 10-1 will be omitted, and the parts that differ from the data analysis device 10-1 will be mainly described.

[0047] The inspection planning unit 19 plans an inspection plan for train 1 based on the anomaly probability calculated by the anomaly probability calculation unit 17 and the dependency relationships calculated by the dependency relationship calculation unit 16. Specifically, the inspection planning unit 19 can plan an inspection plan that indicates the inspection items that should be prioritized for inspection by using the anomaly probability that reflects the dependency relationships.

[0048] Figure 13 shows a first example of an inspection plan formulated by the inspection planning unit 19 shown in Figure 12. The inspection plan shown in Figure 13 includes inspection priorities for each car of train 1. The inspection planning unit 19 can determine the priorities based on the probability of anomalies calculated based on dependencies.

[0049] Figure 14 shows a second example of an inspection plan formulated by the inspection planning unit 19 shown in Figure 12. The inspection plan shown in Figure 14 includes the probability of anomalies for each inspection item calculated by the anomaly probability calculation unit 17, as well as the estimated future probability of anomalies based on the anomaly probability calculated by the anomaly probability calculation unit 17. Here, the anomaly probability calculation unit 17 estimates the probability of anomalies if the vehicle continues to travel on the route identified as "RO001," which is currently assigned to this vehicle. In the inspection plan shown in Figure 14, multiple inspection items are also shown in order of priority.

[0050] In Embodiment 2, the analysis result output unit 18 can output the inspection plan generated by the inspection plan planning unit 19 as an analysis result. For example, the analysis result output unit 18 can generate a display screen including the inspection plan as shown in Figure 13 or Figure 14, and output the generated display screen.

[0051] Figure 15 is a flowchart illustrating the operation of the inspection planning unit 19 shown in Figure 12. The inspection planning unit 19 initializes all abnormality probabilities to 0. The abnormality probability of data item j in group i is denoted as abnormality probability[i][j]. i is a value greater than or equal to 1 and less than or equal to the number of groups, and j is a value greater than or equal to 1 and less than or equal to the number of measurement data items in group i (step S301).

[0052] The inspection planning unit 19 repeats the following steps S302 to S305 for each group i, while changing the value of i from 1 to the number of groups. The inspection planning unit 19 declares an apparent list of undistributed anomaly probabilities and inserts all measurement data items for group i (step S302).

[0053] The inspection planning unit 19 repeats the process from steps S303 to S305 described below as long as the apparent list of undistributed anomaly probabilities is not empty.

[0054] The inspection planning unit 19 selects one measurement data item with the maximum depth from the list of items with apparent unallocated anomaly probabilities, and designates the selected data item as data item j (step S303). Here, depth refers to the depth of the dependency relationships between data items. For example, in Figure 3, the depth of the pantograph is 0, the depth of the inverter is 1, the depth of the motor is 2, and the depth of the wheels is 3. If there are multiple measurement data items with the maximum depth within the list of items with apparent unallocated anomaly probabilities, any of the data items can be selected as long as their depth is the maximum within the list of items with apparent unallocated anomaly probabilities.

[0055] Next, the inspection planning unit 19 substitutes the apparent abnormality probabilities [i][j] into the abnormality probabilities [i][j] to determine the abnormality probability of data item j in the configuration i (step S304).

[0056] The inspection planning unit 19 repeatedly executes the process in step S305 for each destination item k, while changing the value of the destination item l for data item j of organization i from 1 to the number of destinations for item j of organization i. The inspection planning unit 19 calculates "apparent anomaly probability [i][k] += apparent anomaly probability [i][j] × dependency of item j on item k" (step S305).

[0057] As described above, the data analysis device 10-2 according to Embodiment 2 further includes an inspection planning unit 19 that formulates an inspection plan indicating inspection items that should be prioritized for inspection based on the probability of anomalies. Therefore, it becomes possible to formulate an inspection plan that takes dependencies into consideration and prioritizes the inspection of inspection items that cause anomalies. Consequently, it becomes possible to identify the true cause of an anomaly and to determine which equipment should be prioritized for inspection when signs of an anomaly are observed.

[0058] Embodiment 3. Figure 16 shows the functional configuration of the data analysis device 10-3 according to Embodiment 3. In addition to the configuration of the data analysis device 10-2, the data analysis device 10-3 includes an operation planning unit 20 and a route data storage unit 21.Hereafter, detailed explanations of parts that are the same as those of Embodiments 1 and 2 will be omitted, and the parts that differ from Embodiments 1 and 2 will be mainly described.

[0059] The data acquisition unit 11 acquires route data in addition to the data described in Embodiment 1 and stores it in the route data storage unit 21. The route data storage unit 21 stores the route data acquired by the data acquisition unit 11. The route data is information that indicates the characteristics of the route on which train 1 travels.

[0060] Figure 17 shows an example of route data stored in the route data storage unit 21 shown in Figure 16. Here, the route data includes the maximum gradient and maximum cant for each route and each section of that route.

[0061] The operational planning unit 20 determines the routes to be assigned to each of the multiple cars that make up train 1, based on the probability of anomalies and route data. Specifically, the operational planning unit 20 can determine the routes to be assigned to each of the multiple cars in such a way that the increase in the probability of anomalies over time is small. For example, for cars with a high probability of anomalies, the operational planning unit 20 can suppress the increase in the probability of anomalies by assigning them relatively flat routes with many sections with small gradients from among the multiple routes to be assigned.

[0062] Figure 18 shows an example of an operation plan generated by the operation plan planning unit 20 shown in Figure 16. In this example, the vehicle with vehicle ID "TR001" is currently assigned to route "RO001". The current abnormality probabilities for this vehicle are 60% for wheel rotation speed, 15% for wheel damage, 12% for motor voltage, and 3% for pantograph wear. Note that "pantograph wear" in the figure refers to wear on the pantograph. If the vehicle continues to operate on the currently assigned route "RO001", the abnormality probabilities will change as follows: wheel rotation speed to 65% and 70%, wheel damage to 15% and 25%, motor voltage to 13% and 17%, and pantograph wear to 3% and 3%. In contrast, when the route "RO002" is assigned to the vehicle with vehicle ID "TR001", the probability of abnormalities changes as follows: wheel rotation speed changes to 61% and 63%, wheel damage changes to 15% and 18%, motor voltage changes to 12% and 13%, and pantograph wear changes to 3% and 3%. Thus, by changing the assigned route, the overall increase in the probability of abnormalities for the vehicle with vehicle ID "TR001" is suppressed.

[0063] The operational planning unit 20 can, for example, assign routes to the vehicles to be allocated in order of the probability of anomalies being highest. In this case, for each vehicle, it can estimate the trend of the probability of anomalies if each of the routes to be allocated is assigned to the vehicle, and assign the route that can suppress the increase in the probability of anomalies over time the most among the routes to be allocated to the vehicle. Note that the example shown here is just one example, and the operational planning unit 20 is not limited in its method as long as it can determine the routes to be assigned to each vehicle in a way that suppresses the overall increase in the probability of anomalies.

[0064] The analysis result output unit 18 can output the operation plan formulated by the operation plan formulation unit 20 as an analysis result. For example, the analysis result output unit 18 can generate a display screen including the operation plan as shown in Figure 18 and output the generated display screen.

[0065] As described above, the data analysis device 10-3 according to Embodiment 3 further includes an operation planning unit 20 that determines the route to be assigned to each of a plurality of vehicles based on the probability of anomalies. The operation planning unit 20 can predict the trend of the probability of anomalies and determine the route to be assigned to the vehicles in such a way that the increase in the probability of anomalies over time is small. Therefore, by changing the route assigned to each vehicle, it becomes possible to suppress the increase in the probability of anomalies over time and to extend the inspection cycle of each piece of equipment on the vehicle.

[0066] The functions of the data analysis device 10-3 according to Embodiment 3 can be applied not only when the target device is a railway train, but also to all vehicles operating along a railway line. For example, the vehicle may be a bus or other type of vehicle.

[0067] Embodiment 4. Figure 19 shows the functional configuration of the data analysis device 10-4 according to Embodiment 4. The data analysis device 10-4 has an operation planning unit 20-1 instead of the operation planning unit 20 of the data analysis device 10-3 according to Embodiment 3, and further has a unit 22 for calculating the probability of anomaly in the formation.Hereafter, detailed explanations of parts that are the same as in Embodiment 3 will be omitted, and the parts that differ from Embodiment 3 will be mainly described.

[0068] The train formation anomaly probability calculation unit 22 calculates the anomaly probability for each train formation of train 1 based on the anomaly probability calculation unit 17. Specifically, the train formation anomaly probability calculation unit 22 calculates the anomaly probability for each car based on the anomaly probability of the data items calculated by the anomaly probability calculation unit 17, and further calculates the anomaly probability for each formation based on the anomaly probability of each of the multiple cars included in each formation of train 1. For example, the train formation anomaly probability calculation unit 22 can calculate the anomaly probability for each car by averaging the anomaly probabilities of the data items for each car. Alternatively, for example, the train formation anomaly probability calculation unit 22 can calculate the anomaly probability for each formation by averaging the anomaly probabilities for each car for each formation. The train formation anomaly probability calculation unit 22 outputs the calculated anomaly probabilities for each formation to the operation plan formulation unit 20-1.

[0069] The operation planning unit 20-1 formulates an operation plan for each train set by assigning a route to each train set based on the abnormality probability for each train set calculated by the train set abnormality probability calculation unit 22 and the route data. The operation planning unit 20-1 formulates the operation plan in such a way that the increase in the abnormality probability of a train set is suppressed.

[0070] Figure 20 shows an example of an operation plan generated by the operation plan planning unit 20-1 shown in Figure 19. In this example, the train set with train set ID "OR001" is currently assigned route "RO001," and by changing the assigned route to "RO002," the increase in each abnormality probability is suppressed. The operation plan planning unit 20-1 can formulate an operation plan using the same process as the operation plan planning unit 20, except that it assigns a route to each train set using the abnormality probability for each train set.

[0071] The probability of an anomaly for each formation may be represented by a single numerical value, or, as shown in Figure 20, it may be represented for each formation and for each item.

[0072] The analysis result output unit 18 can output the operation plan formulated by the operation plan formulation unit 20-1 as an analysis result. For example, the analysis result output unit 18 can generate a display screen including the operation plan as shown in Figure 20 and output the generated display screen.

[0073] As described above, the data analysis device 10-4 according to Embodiment 4 further includes a train set abnormality probability calculation unit 22 that calculates the abnormality probability for each train set 1, which is composed of vehicles, based on the abnormality probability of the data items, and the operation planning unit 20-1 determines the route to be assigned to each of the multiple vehicles for each train set, based on the abnormality probability for each train set. This makes it possible to formulate a vehicle operation plan with high operational efficiency even for train sets with a large number of vehicles, where switching between vehicles in the train set is costly.

[0074] Embodiment 5. Figure 21 shows the functional configuration of the data analysis device 10-5 according to Embodiment 5. The data analysis device 10-5 has an operation planning unit 20-2 instead of the operation planning unit 20-1 of the data analysis device 10-4 according to Embodiment 4, and further has a driver data storage unit 23 and a driver skill prediction unit 24.Hereafter, detailed explanations of parts that are the same as Embodiment 4 will be omitted, and the parts that differ from Embodiment 4 will be mainly described.

[0075] The data acquisition unit 11 further acquires driver data and stores it in the driver data storage unit 23. The driver data storage unit 23 stores the driver data acquired by the data acquisition unit 11. The driver data is data relating to the driver operating train 1.

[0076] Figure 22 shows an example of driver data stored in the driver data storage unit 23 shown in Figure 21. For example, the driver data includes a driver number that identifies each driver, the assigned train set which the driver is responsible for operating, and the period for which the driver is responsible for operating that train set.

[0077] The driver skill prediction unit 24 predicts the skill level of the driver responsible for operating train 1 based on the probability of anomalies. The driver skill prediction unit 24 can predict the driver's skill level based on the probability of anomalies for each train set calculated by the train set anomaly probability calculation unit 22 and the driver data stored in the driver data storage unit 23. The driver skill prediction unit 24 outputs information indicating the predicted driver skill level to the operation plan formulation unit 20-2.

[0078] Figure 23 shows an example of information indicating driver skills generated by the driver skill prediction unit 24 shown in Figure 21. The information indicating driver skills includes, for example, the route number assigned to the driver, associated with the driver identification number, and the driving skill, which is a numerical representation of the driving skills while assigned to that route. Here, the driving skills from previous assignments to that route are evaluated and quantified for each route.

[0079] Returning to the explanation of Figure 22, the operation planning unit 20-2 assigns drivers to operate each train set based on the driver skills predicted by the driver skill prediction unit 24. Specifically, the operation planning unit 20-2 devises a driver operation plan that suppresses an increase in the probability of anomalies for each train set. For example, the operation planning unit 20-2 estimates the probability of anomalies for each train set when each of the multiple drivers to be assigned to that train set is assigned to it, and can assign drivers to each train set that can suppress an increase in the probability of anomalies.

[0080] Figure 24 shows an example of an operation plan generated by the operation plan planning unit 20-2 shown in Figure 22. Here, it can be seen that when the driver currently assigned to the train set with train set ID "OR001" is changed to the driver identified by "DR002", the increase in the abnormal probability for train set number "OR001" is suppressed.

[0081] As described above, the data analysis device 10-5 according to Embodiment 5 further includes a driver skill prediction unit 24 that predicts the driver's driving skills based on the probability of anomalies, and the operation planning unit 20-2 assigns a driver to each train set based on the driving skills. This makes it possible to suppress the increase in the probability of anomalies for each train set by preferentially assigning drivers with high driving skills to train sets with high anomaly probabilities. Furthermore, if driving skills are calculated for each driver for each route, it becomes possible to formulate a driver operation plan that reflects each driver's strengths and weaknesses on each route.

[0082] Next, the hardware configuration of the data analysis device 10 of this embodiment will be described. The data analysis device 10 of this embodiment is implemented by, for example, a computer system. The data analysis device 10 may be implemented by one computer system or by multiple computer systems. For example, the data analysis device 10 may be implemented by a cloud system. In a cloud system, the separation of the computer system hardware and devices such as servers for each function can be arbitrarily configured. For example, one computer system may have the functions of multiple devices, or multiple computer systems may have the functions of one device.

[0083] An example configuration of a computer system that implements the data analysis device 10 will be described. Figure 25 is a diagram showing an example configuration of a computer system that implements the data analysis device 10 according to Embodiments 1 to 5. As shown in Figure 25, this computer system comprises a control unit 101, an input unit 102, a storage unit 103, a display unit 104, a communication unit 105, and an output unit 106, which are connected via a system bus 107.

[0084] In Figure 25, the control unit 101 is, for example, a CPU (Central Processing Unit). The control unit 101 executes a data analysis program that describes each process performed by the data analysis device 10 of this embodiment. The input unit 102 consists of, for example, a keyboard, mouse, etc., and is used by the user of the computer system to input various information. The storage unit 103 includes various types of memory such as RAM (Random Access Memory) and ROM (Read Only Memory), and storage devices such as a hard disk, and stores the program to be executed by the control unit 101, necessary data obtained in the process of processing, etc. The storage unit 103 is also used as a temporary storage area for the program. The display unit 104 consists of an LCD (Liquid Crystal Display) etc., and displays various screens to the user of the computer system. The communication unit 105 is a communication circuit, etc., that performs communication processing. The communication unit 105 may consist of multiple communication circuits corresponding to multiple communication methods. The output unit 106 is an output interface that outputs data to external devices such as a printer or external storage device.

[0085] Note that Figure 25 is an example, and the configuration of the computer system is not limited to the example in Figure 25. For example, the computer system does not have to have an output unit 106. Also, if the data analysis device 10 is implemented by multiple computer systems, not all of these computer systems have to be the computer system shown in Figure 25. For example, some computer systems do not have to have at least one of the display unit 104, output unit 106, and input unit 102 shown in Figure 25.

[0086] Here, we will describe an example of the operation of the computer system until the data analysis program describing the processing of the data analysis device 10 of this embodiment becomes executable. In a computer system with the above configuration, for example, the data analysis program is installed in the storage unit 103 from a CD-ROM or DVD-ROM set in a CD (Compact Disc)-ROM drive or DVD (Digital Versatile Disc)-ROM drive (not shown). When the data analysis program is executed, the data analysis program read from the storage unit 103 is stored in the area that becomes the main memory of the storage unit 103. In this state, the control unit 101 executes the processing of the data analysis device 10 of this embodiment according to the data analysis program stored in the storage unit 103.

[0087] In the above description, a program describing the processing in the data analysis device 10 is provided using a CD-ROM or DVD-ROM as the recording medium. However, the system is not limited to this, and depending on the configuration of the computer system, the capacity of the program to be provided, a program provided via a transmission medium such as the Internet via the communication unit 105 may also be used.

[0088] The data analysis program of this embodiment causes the computer to perform the steps of acquiring data to be analyzed, calculating the dependencies between multiple data items included in the acquired data to be analyzed, and calculating the anomaly probability of the data items based on the dependencies. In the step of calculating the anomaly probability, the program uses different calculation methods for each type of data item to calculate the anomaly probability.

[0089] The data acquisition unit 11 is implemented, for example, using the control unit 101 and the communication unit 105. The measurement data storage unit 12, environmental data storage unit 13, driving data storage unit 14, train formation data storage unit 15, route data storage unit 21, and driver data storage unit 23 are part of the storage unit 103. The dependency relationship calculation unit 16, abnormality probability calculation unit 17, inspection plan formulation unit 19, operation plan formulation units 20, 20-1, 20-2, train formation abnormality probability calculation unit 22, and driver skill prediction unit 24 are implemented by the control unit 101. The analysis result output unit 18 is implemented using the control unit 101 and the display unit 104. The analysis result output unit 18 may be implemented using the control unit 101 and the communication unit 105 if it generates a display screen and outputs the display screen to another computer via a communication channel, or it may be implemented using the output unit 106 if it outputs the analysis results to a printer or to an external storage device.

[0090] The functional divisions in the illustrated data analysis devices 10-1 to 10-5 are merely examples, and the functional divisions of the data analysis device 10 are not limited to the illustrated examples, as long as the operations described above can be performed. For example, the data acquisition unit 11 is provided with the function of classifying data acquired from an external source, but the function of classifying data may be performed by a computer other than the data analysis device 10, or the functions of each functional block may be shared among multiple computers, or some of the functions of each functional block may be shared among multiple computers.

[0091] The configurations shown in the embodiments described above are merely examples of the content of this disclosure, and can be combined with other known technologies, combined with other embodiments, and some parts of the configuration can be omitted or modified without departing from the gist of this disclosure.

[0092] The various aspects of this disclosure are summarized below as an appendix.

[0093] (Note 1) A data acquisition unit that acquires the data to be analyzed, A dependency calculation unit calculates the dependencies between multiple data items included in the acquired data to be analyzed, An abnormality probability calculation unit calculates the abnormality probability of the data item based on the aforementioned dependency relationship, Equipped with, The data analysis device is characterized in that the abnormality probability calculation unit calculates the abnormality probability using a different calculation method for each type of data item. (Note 2) The data analysis device according to Appendix 1, characterized in that the types of data items include measurement data which is data relating to the target device, environmental data which indicates the state of the environment around the target device, and operation data which indicates the content of operations performed on the target device. (Note 3) The data analysis apparatus according to Appendix 2, characterized in that the abnormality probability calculation unit calculates the abnormality probability for each target device when the type of data item is the environmental data. (Note 4) The data analysis device according to Appendix 2 or 3, characterized in that the abnormality probability calculation unit calculates the abnormality probability for each driver operating the target device when the type of data item is the driving data. (Note 5) The data analysis device according to any one of the appendices 1 to 4, characterized in that the anomaly probability calculation unit calculates the anomaly probability based on the difference between the predicted value of the data item calculated based on the dependency relationship and the measured value which is the data to be analyzed. (Note 6) Based on the aforementioned probability of anomalies, the inspection planning unit formulates an inspection plan that indicates the inspection items that should be prioritized for inspection. A data analysis device according to any one of the appendices 1 to 5, further comprising the above. (Note 7) The data to be analyzed is data obtained from the vehicle. Based on the aforementioned abnormal probability, the operational planning unit determines the route to be assigned to each of the multiple vehicles. A data analysis device according to any one of the appendices 1 to 6, further comprising the above. (Note 8) The data analysis device according to Appendix 7, characterized in that the operational planning unit determines the routes to be assigned to each of the multiple vehicles in such a way that the increase in the probability of an anomaly over time becomes small. (Note 9) The aforementioned vehicle is a railway vehicle, A train formation abnormality probability calculation unit calculates the abnormality probability for each train formation composed of the aforementioned vehicles, based on the abnormality probability of the aforementioned data items. Furthermore, The data analysis device according to Appendix 7, characterized in that the operational planning unit determines the route to be assigned to each of the multiple vehicles for each train set based on the probability of anomalies for each train set. (Note 10) A driving skill prediction unit predicts the driver's driving skills based on the aforementioned abnormal probability. Furthermore, The data analysis device according to any one of appendices 7 to 9, characterized in that the operational planning unit assigns the driver to each train set based on the driving skills. (Note 11) The aforementioned device is a railway vehicle, The data analysis device according to Appendix 2, characterized in that the measurement data includes vehicle driving data acquired from equipment mounted on the vehicle while the vehicle is in motion, and inspection data acquired when the equipment is inspected. (Note 12) Analysis result output unit that outputs analysis results including the aforementioned abnormal probability, The data analysis device according to Appendix 1, further comprising the features described above. (Note 13) Analysis result output unit that outputs analysis results including the priority of tests according to the aforementioned probability of abnormality, The data analysis device according to Appendix 6, further comprising the features described above. (Note 14) Analysis result output unit that outputs analysis results including route information assigned to each of the multiple vehicles, The data analysis device according to Appendix 7, further comprising the features described above. (Note 15) The data analysis device according to Appendix 14, characterized in that the analysis results include information on the routes assigned to each train set of vehicles. (Note 16) Analysis result output unit that outputs analysis results including driver information assigned to each train set of the aforementioned vehicles. The data analysis device according to Appendix 10, further comprising the features described above. (Note 17) Steps to obtain the data to be analyzed, The steps include calculating the dependencies between multiple data items included in the acquired data to be analyzed, The steps include: calculating the probability of an anomaly in the data item based on the dependency relationship; Includes, The data analysis method is characterized in that, in the step of calculating the anomaly probability, the anomaly probability is calculated using a different calculation method for each type of data item. (Note 18) On the computer, Steps to obtain the data to be analyzed, The steps include calculating the dependencies between multiple data items included in the acquired data to be analyzed, The steps include: calculating the probability of an anomaly in the data item based on the dependency relationship; Make it run, The program is characterized in that, in the step of calculating the probability of an anomaly, it uses a different calculation method for each type of data item to calculate the probability of an anomaly. [Explanation of symbols]

[0094] 1 Train, 2 Acquisition device, 3 Inspection terminal, 4 Communication network, 10, 10-1 to 10-5 Data analysis device, 11 Data acquisition unit, 12 Measurement data storage unit, 13 Environmental data storage unit, 14 Driving data storage unit, 15 Train formation data storage unit, 16 Dependency calculation unit, 17 Anomaly probability calculation unit, 18 Analysis result output unit, 19 Inspection planning unit, 20, 20-1, 20-2 Operation planning unit, 21 Route data storage unit, 22 Train formation anomaly probability calculation unit, 23 Driver data storage unit, 24 Driver skill prediction unit, 101 Control unit, 102 Input unit, 103 Storage unit, 104 Display unit, 105 Communication unit, 106 Output unit, 107 System bus.

Claims

1. A data acquisition unit that acquires the data to be analyzed, A dependency calculation unit calculates the dependencies between multiple data items included in the acquired data to be analyzed, An abnormality probability calculation unit calculates the abnormality probability of the data item based on the aforementioned dependency relationship, Equipped with, The abnormality probability calculation unit calculates the abnormality probability using a different calculation method for each type of data item. The types of data items include measurement data, which is data relating to the target device; environmental data, which indicates the state of the environment around the target device; and operation data, which indicates the details of operations performed on the target device. The aforementioned device is a railway vehicle, The data analysis device is characterized in that the measurement data includes vehicle driving data acquired from equipment mounted on the vehicle while the vehicle is in motion, and inspection data acquired when the equipment is inspected.

2. The data analysis apparatus according to claim 1, characterized in that the abnormality probability calculation unit calculates the abnormality probability for each target device when the type of data item is the environmental data.

3. The data analysis device according to claim 1, characterized in that the abnormality probability calculation unit calculates the abnormality probability for each driver operating the target device when the type of data item is the driving data.

4. The data analysis apparatus according to claim 1, characterized in that the abnormality probability calculation unit calculates the abnormality probability based on the difference between the predicted value of the data item calculated based on the dependency relationship and the measured value which is the data to be analyzed.

5. Based on the aforementioned probability of anomalies, the inspection planning unit formulates an inspection plan that indicates the inspection items that should be prioritized for inspection. The data analysis device according to any one of claims 1 to 4, further comprising:

6. The data to be analyzed is data obtained from the vehicle. Based on the aforementioned abnormal probability, the operational planning unit determines the route to be assigned to each of the multiple vehicles. The data analysis apparatus according to claim 1, further comprising the features described above.

7. The data analysis device according to claim 6, characterized in that the operational planning unit determines the routes to be assigned to each of the multiple vehicles in such a way that the increase in the probability of an anomaly over time becomes small.

8. A train formation abnormality probability calculation unit that calculates the abnormality probability for each train formation composed of the vehicles based on the abnormality probability of the data items, Furthermore, The data analysis device according to claim 6, characterized in that the operational planning unit determines the route to be assigned to each of the multiple vehicles for each train set based on the probability of anomalies for each train set.

9. A driving skill prediction unit predicts the driver's driving skills based on the aforementioned abnormal probability. Furthermore, The data analysis device according to any one of claims 6 to 8, characterized in that the operational planning unit assigns the driver to each train set based on the driving skills.

10. Analysis result output unit that outputs analysis results including the aforementioned abnormal probability, The data analysis apparatus according to claim 1, further comprising the features described above.

11. Analysis result output unit that outputs analysis results including the priority of tests according to the aforementioned probability of abnormality, The data analysis apparatus according to claim 5, further comprising the features described herein.

12. Analysis result output unit that outputs analysis results including route information assigned to each of the multiple vehicles, The data analysis apparatus according to claim 6, further comprising the features described above.

13. The data analysis device according to claim 12, characterized in that the analysis results include information on the routes assigned to each train set of the vehicles.

14. Analysis result output unit that outputs analysis results including driver information assigned to each train set of the aforementioned vehicles. The data analysis apparatus according to claim 9, further comprising the features described above.

15. Steps to obtain the data to be analyzed, The steps include calculating the dependencies between multiple data items included in the acquired data to be analyzed, The steps include: calculating the probability of an anomaly in the data item based on the dependency relationship; Includes, In the step of calculating the anomaly probability, the anomaly probability is calculated using a different calculation method for each type of data item. The types of data items include measurement data, which is data relating to the target device; environmental data, which indicates the state of the environment around the target device; and operation data, which indicates the details of operations performed on the target device. The aforementioned device is a railway vehicle, The data analysis method is characterized in that the measurement data includes vehicle driving data acquired from equipment mounted on the vehicle while the vehicle is in motion, and inspection data acquired when the equipment is inspected.

16. On the computer, Steps to obtain the data to be analyzed, The steps include calculating the dependencies between multiple data items included in the acquired data to be analyzed, The steps include: calculating the probability of an anomaly in the data item based on the dependency relationship; Make it run, In the step of calculating the anomaly probability, the anomaly probability is calculated using a different calculation method for each type of data item. The types of data items include measurement data, which is data relating to the target device; environmental data, which indicates the state of the environment around the target device; and operation data, which indicates the details of operations performed on the target device. The aforementioned device is a railway vehicle, The program is characterized in that the measurement data includes vehicle driving data acquired from equipment mounted on the vehicle while the vehicle is in motion, and inspection data acquired when the equipment is inspected.

Images