Vehicle operation target estimation system and vehicle operation target estimation method

Abstract

According to the present invention, a vehicle operation target estimation system that estimates an operation target for a train is a computer that has a processing unit that executes prescribed processing and a storage unit that is connected to the processing unit. The vehicle operation target estimation system comprises a line feature quantity generation unit that generates a line feature quantity on the basis of an operation diagram, an operation feature quantity generation unit that generates an operation feature quantity on the basis of operation information, a degradation prediction formula generation unit that estimates a prediction formula for degradation that uses the generated operation feature quantity as an explanatory variable, an operation variation pattern generation unit that generates a plurality of operation patterns for the train, a generation unit that estimates statistical values for remaining life, a failure occurrence range, and failure occurrence in accordance with the generated operation patterns, and a remaining life / operation feature quantity modeling unit that estimates a model that represents the relationship between remaining life and the operation feature quantity from the estimated statistical values for remaining life, failure occurrence range, and failure occurrence.

Description

Vehicle operation target estimation system and vehicle operation target estimation method Incorporation by Reference

[0001] This application claims the priority of Japanese Patent Application No. 2025-38732 filed on March 11, 2025 (Reiwa 7), and the content thereof is incorporated into the present application by reference.

[0002] The present invention relates to a vehicle operation target estimation system that estimates an operation target for a train formation.

[0003] There exists an estimation device that estimates a prediction interval for a future normal behavior range using the variance of actual values of an operation amount.

[0004] As background art in the present technical field, there is the following prior art. Patent Document 1 (Japanese Unexamined Patent Application Publication No. 2020-170304) discloses an estimation device comprising: a prediction interval data acquisition unit that acquires prediction interval data, which is data generated using monitoring data including a monitoring result of a monitoring target and a monitoring time thereof, the prediction interval data indicating an estimation result of a time transition of an n% prediction interval of the monitoring target; and an estimation time output means that generates and outputs output data indicating a time when the reference value is estimated to fall within the n% prediction interval, using the prediction interval data and a reference value set for the monitoring target.

[0005] In the aforementioned prior art, the prediction interval is obtained on the assumption that fluctuation in the actual values of the operation amount is noise following a uniform distribution. However, in the case of railway vehicles, fluctuation in the deterioration of the operation amount is contributed by differences in the operation method and driving operation of the railway vehicle, but this point is not taken into consideration, and there is no model that enables prediction including fluctuation in the deterioration of the operation amount. Furthermore, it is difficult to predict deterioration when the operation method or driving accuracy is changed.

[0006] Furthermore, in railway vehicles, the degree of deterioration is measured at fixed intervals such as monthly inspections and general inspections, making continuous monitoring of the degree of deterioration difficult.

[0007] Accordingly, there is a demand for an estimation system that models the relationship between fluctuation factors and the spread of prediction intervals to estimate the deterioration time resulting from changes in operation.

[0008] A representative example of the invention disclosed in this application is as follows: A vehicle operation target estimation system for estimating the operational target of a train formation, comprising a computer having a processing unit that performs predetermined processing and a storage unit connected to the processing unit, and characterized by comprising: a line feature generation unit that generates line feature quantities based on the operation timetable; an operation feature generation unit that generates operation feature quantities based on operation information; a deterioration prediction formula generation unit that estimates a deterioration prediction formula using the generated operation feature quantities as explanatory variables; an operation change pattern generation unit that generates a plurality of patterns for the operation of a train formation; a generation unit that estimates remaining life, failure occurrence range, and failure occurrence statistics according to the generated operation patterns; and a remaining life / operation feature modeling unit that estimates a model representing the relationship between remaining life and operation feature quantities from the estimated remaining life, failure occurrence range, and failure occurrence statistics.

[0009] According to one aspect of the present invention, the deterioration period of railway vehicles can be accurately estimated. Problems, configurations, and effects other than those mentioned above will be clarified by the following description of the embodiments.

[0010] This is a block diagram showing the configuration of a vehicle operation target estimation system according to an embodiment of the present invention. This is a diagram showing the configuration of a system including the vehicle operation target estimation system of this embodiment. This is a flowchart of the relationship modeling process shown in Figure 2, which is performed by the vehicle operation target estimation system of this embodiment. This is a flowchart of the remaining life estimation of the current operation, which is performed by the vehicle operation target estimation system of this embodiment. This is a flowchart of the operational change plan formulation process shown in Figure 2, which is performed by the vehicle operation target estimation system of this embodiment. This is a diagram showing the prediction results of this embodiment. This is a diagram showing an example of the configuration of the feature quantities for the train lines in this embodiment. This is a diagram showing an example of the configuration of the feature quantities for the driver in this embodiment. This is a diagram showing an example of the configuration of feature quantities for the operation in this embodiment. This is a diagram showing the predicted change in remaining life in the current operation when this embodiment is performed. This is a diagram showing the predicted change in remaining life in the operational change plan when this embodiment is performed. This is a diagram showing an example of feature quantity listing.

[0011] Figure 1 is a block diagram showing the configuration of the vehicle operation target estimation system 100 in this embodiment.

[0012] The vehicle operation target estimation system 100 of this embodiment is composed of a computer having an input unit 110, an output unit 120, a processing unit 130, a communication unit 140, a main memory unit 150, and an auxiliary memory unit 170.

[0013] The input unit 110 is an interface to which an input device is connected and which receives input from a user (for example, a maintenance manager).

[0014] The output unit 120 is an interface to which an output device is connected, and which outputs the program execution results in a format that can be recognized by the user.

[0015] The processing unit 130 is an arithmetic unit that executes programs stored in the main memory unit 150. By executing various programs, the functions of each functional unit of the vehicle operation target estimation system 100 are realized. Note that some of the processing performed by the processing unit 130 by executing programs may be performed by other arithmetic units (for example, hardware such as ASICs or FPGAs).

[0016] The communication unit 140 is a network interface device that controls communication with other devices via the network 230 according to a predetermined protocol.

[0017] The main memory unit 150 is a memory that includes ROM, which is a non-volatile memory element, and RAM, which is a volatile memory element. ROM stores immutable programs (e.g., BIOS). RAM is a high-speed, volatile memory element such as DRAM (Dynamic Random Access Memory), and temporarily stores programs executed by the processing unit 130 and data used during program execution. The main memory unit 150 stores programs that implement the functions of the following units: streak feature generation unit 151, operational feature generation unit 152, degradation prediction formula generation unit 153, operational change pattern generation unit 154, remaining life / failure range estimation unit 155, remaining life / operational feature modeling unit 156, effective feature extraction unit 157, feature selection unit 161, objective-based operation planning unit 162, and remaining life prediction unit 163.

[0018] The suji feature generation unit 151 to the effective feature generation unit 157 are a group of functions used in relational modeling, which will be described later. The suji feature generation unit 151 generates suji features. In this specification, a suji refers to the travel time for a single train in a train schedule, from the starting station to the final station. A service consists of one or more suji, indicating where and when a train moves from. Usually, it often refers to a trip from leaving one train depot to returning to another train depot.

[0019] The operational feature extraction unit 152 generates features for operations. As shown in Figure 7, an operation in this specification consists of one or more duties. Using this operation, for example, a train operation is carried out in which, for train formation A, duty 2 is performed on the first day and duty 3 is performed on the second day.

[0020] The degradation prediction formula generation unit 153 estimates a degradation prediction formula using operational features as explanatory variables. Specifically, it estimates a degradation prediction formula (equation (1)) using operational features as explanatory variables. In equation (1), Y is the degradation degree of each operation, and X is the feature matrix of each operation. β is a weight indicating the degree of influence of the explanatory variables on degradation, and is a vector with the number of explanatory variables in the feature matrix X as its dimension. Δ is the intercept. Equation (1) is a linear multiple regression equation, but may include nonlinear terms.

[0021]

[0022] If we provide an n-dimensional vector of the degradation degree Y for n operations and the feature matrix X for each of the n operations (an n x p matrix if there are n operations and p explanatory variables for each operation) to equation (1), we can calculate the weights β and intercept Δ. The weights β and intercept Δ then become the output of the prediction formula.

[0023] The operation change pattern generation unit 154 has the function of generating a different schedule and duties from the current operation and generating an alternative operation. In this embodiment, the purpose is to determine the changes in remaining lifespan and failure timing when the operation is changed, so transportation demand may not be particularly considered when generating the patterns. Several generation methods are possible, for example, generating an operation with some duties reduced from an existing operation, or adding a new schedule considering the mobility performance of the train formation.

[0024] The remaining life and failure range estimation unit 155 estimates the remaining life, failure range, and failure statistics based on the specified operation. The remaining life is the time from the present until the timing when a failure is predicted to occur, and is the shortest predicted life shown in Figure 4. The failure range is the time range between the timing when a failure occurs first and the timing when a failure occurs latest, and is the predicted range shown in Figure 4. The median failure is the median of the above predicted range. For this median failure, the prediction formula generated by the degradation prediction formula generation unit 153 is used. The remaining life and failure range can be obtained by using the variance obtained when calculating the above prediction formula. Specifically, they can be calculated using equations (2) and (3) shown below.

[0025]

[0026] In equation (2), S is an unbiased estimator. h P is the maximum likelihood estimate of β. Equation (3) is the predicted range for future degradation. low and P high x represents the lower and upper limits of the prediction range at a future point in time, and x represents the value of the feature at that future point in time.

[0027] Then, using equation (1), we estimate the future degradation amount Y and set this as the prediction center. By adding the upper and lower limits shown in equation (3) to the prediction center, we can calculate the colored range in Figure 4. Furthermore, by setting a certain threshold for the degree of degradation, we can calculate the time when the threshold is exceeded relative to the colored range (shortest predicted lifespan, longest predicted lifespan) and the center value of failure occurrence.

[0028] The remaining life / operational feature modeling unit 156 estimates a model representing the relationship between remaining life and operational features, or the relationship between the failure range and operational features, or the relationship between the median failure and operational features, from a value vector whose elements are at least one of the operational features, remaining life, failure range, and median failure occurrence, which are estimated from multiple (for example, m operational patterns) operational features. Several methods can be considered for modeling this relationship, but one method is to obtain it using linear multiple regression. Equation (4) shows how this modeling works. T represents the target variable, and in this process, one of the remaining life, failure range, or median failure occurrence is selected as the target variable. W is an m × p dimension feature matrix whose elements are the p features of each operation in the m operational patterns, and γ is the weight for the features. Also, e is the intercept. The output is γ and e. By referring to the weight represented by γ, for example, the relationship between operational features and remaining life becomes clear, and the features to be considered when extending / reducing the remaining life become clear.

[0029]

[0030] The effective feature extraction unit 157 extracts features that contribute to a specific parameter. Specifically, when the target variable T is an m-dimensional vector whose elements are the remaining lifetimes of m operational patterns, the weights γ = (γ1, γ2, ... γ k ) Increase T in γ i List them. Similarly, reduce γ iThese are also listed as candidates to be reduced. Similarly, the prediction range and the median failure rate are also listed with weights γ. Figure 9 shows an example of this listing. The attribute column is a column that shows one of the following: {remaining life, prediction interval, median failure rate}, and the direction column shows how the selected attribute changes with an increase in the feature. The corresponding feature column shows the result of the list of weights γ. In this example, one list is shown in [], and each element of the list is shown in () in key-value format. The Key corresponds to the item name shown in Figure 5 or Figure 6, and the value shows the contribution to the selected attribute. Note that there may be negative contributions (i.e., inverse effect). If it is negative, it means that the operation should be in the direction of decreasing the value of the item name.

[0031] The feature selection unit 161 and the objective-based operation planning unit 162 are functions that will be implemented in the operation change plan flowchart described later.

[0032] The feature selection unit 161 is a function that selects features that are closest to the user's objective from a list of features. Specifically, it uses the objective of operational change, refers to the list of features generated by the effective feature extraction unit 157, and selects those with the highest absolute contribution values.

[0033] The objective-based operation planning unit 162 is a function that generates operations that are close to the user's objective. There are several ways to generate operations, but for example, if the current operation is used as a provisional solution, and the objective of changing the operation is to extend the remaining lifespan, then features that contribute to extending the remaining lifespan are selected. If the features are to increase the distance traveled and reduce the tunnel ratio, then the operation to be changed is to increase the distance traveled with fewer tunnel sections, so the number of routes and operations should be increased and the work modified accordingly.

[0034] The remaining lifespan prediction unit 163 has the function of estimating the remaining lifespan of the train formation when the specified operation is performed.

[0035] The auxiliary storage unit 170 is a high-capacity, non-volatile storage device such as a flash memory (SSD) or a magnetic storage device (HDD). The auxiliary storage unit 170 also stores data used by the processing unit 130 when executing a program, and the program executed by the processing unit 130. In other words, the program is read from the auxiliary storage unit 170, loaded into the main storage unit 150, and executed by the processing unit 130 to realize each function of the vehicle operation target estimation system 100. The auxiliary storage unit 170 stores the auxiliary storage unit 170 itself, sensing data 171, inspection history 172, operation schedule 173, operation target 174, operation information 175, and environmental information 176.

[0036] Sensing data 171 is the result of the sensor device 300 measuring the operating amount of equipment mounted on the train set 400. Inspection history 172 is the history of inspections of the train set 400. Operation schedule 173 is the operation plan for the train set 400. Operational targets 174 is information that serves as a target for determining the operation of the train set 400. Operational information 175 is information on the operation of the train set 400. Environmental information 176 is information such as the track shape and topography (gradient, curves) of the route on which the train set 400 operates. Estimated model 177 is the output information of the relationship modeling shown in Figure 2, which will be described later.

[0037] The program executed by the processing unit 130 is provided to the vehicle operation target estimation system 100 via removable media (CD-ROM, flash memory, etc.) or a network, and is stored in the non-volatile auxiliary storage unit 170, which is a non-temporary storage medium. For this reason, the vehicle operation target estimation system 100 should have an interface for reading data from the removable media.

[0038] The vehicle operation target estimation system 100 may be configured as a computer system configured on a plurality of logically or physically configured computers, and may operate on a virtual machine constructed on a plurality of physical computer resources. For example, a plurality of programs that implement the functions of the vehicle operation target estimation system 100 may each operate on separate physical or logical computers, or a plurality of programs may be combined to operate on one physical or logical computer. In this case, a terminal connected to the vehicle operation target estimation system 100 may provide the input device and the output device.

[0039] The operation management system 200 is a computer system that monitors the operation status of trains, transmits commands to operating trains, and manages train operation.

[0040] The transportation plan management system 210 is a computer system that creates a train operation diagram 173 based on passenger demand and train operation resources.

[0041] The operation plan management system 220 is a computer system that creates vehicle operation schedules and crew operation schedules for realizing the created train operation diagram 173.

[0042] The sensor device 300 is installed on a route on which a train formation 400 travels, and measures the train formation 400. The sensor device 300 may be mounted on the train formation 400. The sensor device 300 measures the behavior of the train formation 400 and the operation amount of equipment mounted on the train formation 400. For example, the sensor device 300 installed on the route on which the train formation 400 travels is a surveillance camera. Further, the sensor device 300 attached to the train formation 400 is, for example, a current sensor, a voltage sensor, an acceleration sensor, or the like. Measurement results obtained by the sensor device 300 are transmitted to the vehicle operation target estimation system 100 and recorded as sensing data 171.

[0043] Figure 2 is a configuration diagram of a system including the vehicle operation target estimation system 100 of the present embodiment.

[0044] Train set 400 is equipped with a sensor device 300, which measures the operating amounts of the equipment installed on train set 400. The measured operating amounts are the actual performance and response performance of the equipment relative to the required performance, such as the amount of brake operation and deceleration, the opening and closing time of passenger doors, and the time it takes for the air conditioning system to reach the set temperature. The measurement results from the sensor device 300 are recorded as sensing data 171.

[0045] The vehicle operation target estimation system 100, following instructions from the maintenance manager, uses sensing data 171, inspection history 172, operation schedule 173, operation information 175, and environmental information 176 to estimate the relationship between the degree of deterioration and the target operation as a model in relational modeling 181. The estimation results are then stored in estimation model 177. Next, the vehicle operation target estimation system 100 estimates the remaining lifespan, predicted section, and median failure occurrence of the target train set in the current operation remaining lifespan estimation 182. Next, in operation change proposal formulation 183, it estimates the remaining lifespan, predicted section, and median failure occurrence if an operation different from the current operation is performed. Finally, in operation target change 184, it compares the current operation with the proposed operation change and decides which operation to implement (or to keep the current operation). The ultimately selected operation is transmitted to the operation plan management system 220 as operation target 174.

[0046] The operation plan management system 220 uses the operation schedule 173 created by the transportation plan management system 210 and the operation targets 174 input from the vehicle operation target estimation system 100 to formulate an operation plan and create operation information 175. The created operation information 175 is output to the operation management system 200.

[0047] The operation management system 200 manages train operations using the operation schedule 173 created by the transportation planning management system 210 and the operation information 175 created by the operation planning management system 220.

[0048] Figure 3A is a flowchart of the relationship modeling process shown in Figure 2, which is performed by the vehicle operation target estimation system 100 in this embodiment. The following explanation will follow this flowchart.

[0049] First, the feature quantities for the route are generated (301). This process is carried out by the route feature quantity generation unit 151 based on the operation schedule 173 and environmental information 176. As shown in Figure 5, the feature quantities for a route are the characteristics of the route due to linear factors and operational factors. For example, the total length means the distance traveled from the starting station to the ending station of the route, and the elevation gain indicates the cumulative increase in elevation when traveling from the starting station to the ending station. The route feature quantity generation unit 151 obtains feature quantities by referring to the environmental information 176 at regular intervals of travel distance from the starting station. For example, feature quantities at a point 100m from the starting station are obtained from the environmental information 176, then feature quantities at a point 200m, and so on, up to the ending station, to obtain the value of each feature quantity. Then, by summing up each feature quantity item, the feature quantities for that route can be obtained.

[0050] Next, operational feature quantities are generated (302). This process is carried out by the operational feature quantity generation unit 152. Specifically, operational feature quantities are generated based on operational information 175. As shown in Figure 7, the operational feature quantities are generated using the linear factors and operational factors of the route traveled in the train operation included in the train operation. The feature quantities due to operational factors are quantities that change depending on the driver's driving operations.

[0051] Then, a degradation prediction formula is estimated using operational features as explanatory variables for the degradation level (303). This process is performed by the degradation prediction formula generation unit 153.

[0052] Next, multiple operation pattern change patterns are generated (304). This process is achieved by executing the operation change pattern generation unit 154 multiple times, thereby obtaining multiple operations.

[0053] Next, for each operation, estimated values ​​of the remaining lifespan, failure range, and median failure occurrence are calculated from the given threshold (305). This process is performed by the remaining lifespan / failure range estimation unit 155. If the output obtained from one operation is considered as a value vector of (remaining lifespan, failure range, median failure occurrence), then as a result of the processing by the remaining lifespan / failure range estimation unit 155, (remaining lifespan, failure range, median failure occurrence) are obtained for the feature quantities of that operation. If these are again considered as a value vector of (operation feature quantities, failure range, median failure occurrence), then as a result of the processing in (305), multiple value vectors of (operation feature quantities, failure range, median failure occurrence) are obtained.

[0054] Next, a relationship model between remaining lifespan and operational features is estimated from the estimated remaining lifespan, failure range, and median failure rate (306). This process is performed by the remaining lifespan / operational features modeling unit 156.

[0055] Next, the features that increase or decrease the remaining lifespan, the features that increase or decrease the width of the prediction range, and the features that increase or decrease the median failure occurrence are listed (307). This process is performed by the effective feature extraction unit 157.

[0056] Finally, the degradation prediction formula and relational model are saved and the process terminates (308). This not only allows for future predictions of the degradation level, but also enables the identification of features necessary to change the degradation level in a desirable direction by referring to the relational model.

[0057] Figure 3B shows a flowchart related to the estimation of the remaining lifespan under current operation, as shown in Figure 2. The following explanation will follow this flowchart.

[0058] First, a prediction formula for deterioration of the target train set is obtained (311). This prediction formula is stored in the estimation model 177 and can be retrieved. Next, the remaining lifespan is estimated based on current operation (312). This process is performed by the remaining lifespan prediction unit 163. By following this flowchart, it is possible to predict the degree of deterioration from the present moment to the right (future direction) in Figure 4, and to estimate the prediction interval for the degree of deterioration. Furthermore, by finding the intersection points based on separately specified thresholds, the shortest predicted lifespan, longest predicted lifespan, prediction range, etc., can be output.

[0059] Figure 3C shows the flowchart of the proposed operational changes from Figure 2. The following explanation will follow this flowchart.

[0060] First, the objective of the operational change is set (321). This can be entered by the maintenance manager or it can be an objective set in advance. The objectives of the operational change include increasing the remaining lifespan, decreasing the remaining lifespan, widening the prediction range, narrowing the prediction range, accelerating the median failure occurrence, and delaying the median failure occurrence. One of these must be selected.

[0061] Next, we obtain a prediction formula for deterioration for the target train formation (322). Since the deterioration prediction formula is stored in the estimation model 177, we obtain the prediction formula corresponding to the target train formation from it. At the same time, we also obtain a list of features.

[0062] Next, a feature that is closest to the purpose of the operational change is selected from the list of features (323). This process is carried out by the feature selection unit 161.

[0063] Next, operations are generated that closely match the objective of the operational change based on the features (324). This process is carried out by the objective-based operational planning unit 162.

[0064] Next, the remaining lifespan for the generated operations is estimated (325). Here, based on the generated operations, the streak feature generation unit 151 and the operation feature generation unit 152 are used to calculate the features for the generated operations. These features are then substituted into the degradation prediction formula to calculate the future degradation level. For example, if the generated operations consist of streaks that pass through fewer tunnels, the "tunneling rate" in the operation features will also decrease. If the prediction formula acts in such a way that a decrease in the tunneling rate slows down the degradation level, then the output will ultimately show an extended remaining lifespan.

[0065] Next, the effect is output by comparing it with the remaining lifespan under current operation (326). The effect here refers to the change in the remaining lifespan.

[0066] This operational change planning flowchart allows for the output of the remaining lifespan after the operational change, making it possible to verify the effects of the operational change. Although this flowchart uses remaining lifespan as an example, the same flowchart can be used to process changes in the prediction range or the median failure rate. Such operational changes make it possible, for example, to widen the prediction range to use the vehicle until it fails, to extend the remaining lifespan to ensure use until the vehicle replacement time, or to narrow the prediction range to reduce the probability of failure within a specified period.

[0067] Figure 5 shows an example of the configuration of the streak feature quantities in this embodiment.

[0068] The features of a train line are characteristics of factors that put a load on the vehicle's equipment, and it is best to represent them as vector values ​​with the number of items as dimensions. The items of the train line features can be broadly divided into features due to linear factors of the train line and features due to operational factors.

[0069] Linear features are indices that do not change depending on the type of train (express train, local train, freight train, etc.) or the speed of travel. Examples include the total length of the route, the elevation gain (the sum of the elevation gains in uphill sections while the train is traveling along the route), the elevation loss (the sum of the elevation drops in uphill sections while the train is traveling along the route), the average gradient (the difference in elevation between the start and end points of the route divided by the total length), the proportion of left curves in the route, the proportion of right curves in the route, the number of tunnels in the route, the proportion of tunnel sections in the route, and the number of bridges in the route.

[0070] For example, elevation gain correlates highly with load at high notch, while elevation loss correlates highly with brake wear. In the case of railways, since the train's route is determined for each schedule, the proportion of left and right curves in a given schedule can be calculated. Left and right curves result in different loads on the equipment (e.g., wheel wear). The curve-related features could be indicators such as the cumulative value of curvature or the cant, which is the difference in height between the top surfaces of the left and right rail heads, rather than the proportion of left and right curves.

[0071] The operational characteristics are indicators that vary depending on the type of train and the speed at which it runs. Examples include the number of stops on the route, the average speed calculated by dividing the time difference between the start and end points of the route by the total length, the average braking force during braking on the route, and the average number of braking operations on the route.

[0072] Instead of the number of stops, the number of accelerations and brakings within a given timeline may be used as an indicator of the frequency of acceleration and deceleration of the train within that timeline; however, the number of stops may be used as a simpler indicator.

[0073] Figure 6 shows an example of the configuration of the driver's feature quantities in this embodiment.

[0074] Characteristics of a train driver include, for example, the acceleration ratio (the proportion of time spent in powered driving), the average acceleration during acceleration, the deceleration ratio (the proportion of time spent in braking), the average deceleration during deceleration, the average sway (the acceleration applied to the vehicle), and the average buffer time (the average difference between the scheduled time and the actual time). These indicators vary depending on the driver's skill and driving tendencies.

[0075] Figure 7 shows an example of the configuration of feature quantities for operation in this embodiment. Column 701 represents the duty number. Column 702 represents the location and time of the train formation in the duty in question. In this example, the horizontal axis within the column represents the time axis, and the positions where circles or station names are written represent the time. Circles represent train depots, and the locations where station names are written represent the starting or ending station. The line segments represent the train schedule, and the symbols T1, T2, and T3 within the line segments represent the train numbers. In this example, only duty number 1 is described in detail, but the explanation will continue assuming that similar information is attached to the remaining duties.

[0076] The operational features are a vector value obtained by adding the vector values ​​of the feature values ​​of the train schedules included in each of the duties included in a given train operation. For example, as shown in Figure 7, if duty 1 includes the train schedules of train numbers T1, T2, and T3, the operational features are represented by a vector value obtained by adding the vector values ​​of the feature values ​​of these train schedules.

[0077] Figures 8A and 8B show the predicted changes in remaining lifespan when this embodiment is implemented. Figure 8A shows the predicted remaining lifespan if the current operation continues. The horizontal axis represents the time axis, and in this example, it shows the number of months. The vertical axis shows the degree of deterioration, with a deterioration threshold set at 400, and a value of 400 or higher indicates that replacement is necessary. The current time is the 24th month, and 801 shows the predicted degree of deterioration (centerline) for the target train set if the current operation continues. The colored 802 shows the predicted section relative to the deterioration prediction 801. 803 represents the remaining lifespan, calculated from the current time until the predicted section first exceeds the threshold. 804 shows the predicted range, indicating when the train set will fail at the earliest and by the latest. Note that the prediction interval 802 represents the vertical variation (i.e., the range of variation in the degree of deterioration) relative to the predicted degree of deterioration (center line) 801, while the prediction range 804 represents the horizontal variation (i.e., the range of variation in timing) relative to the predicted degree of deterioration (center line) 801 at the point when the threshold is exceeded.

[0078] Figure 8B shows the change in remaining life when the operation is changed to one that narrows the width of the prediction interval from the current point in time. 805 shows the predicted degree of deterioration (center line) when the operation is changed from the current point in time. The colored 806 shows the prediction interval for the predicted degree of deterioration 805. 807 shows the remaining life, calculated from the current point in time until the prediction interval first exceeds the threshold. Furthermore, 808 shows the prediction range. It can be clearly seen that the prediction interval 806 in Figure 8B is narrower than the prediction interval 802 in Figure 8A. As a result, if the operation in Figure 8B is implemented, the range of variation in the replacement timing can be reduced. A smaller range of variation in the replacement timing means that the ordering timing can be determined more accurately, which can lead to a reduction in inventory costs that would otherwise be incurred when ordering in advance.

[0079] It should be noted that the present invention is not limited to the embodiments described above, but includes various modifications and equivalent configurations within the spirit of the attached claims. For example, the embodiments described above are described in detail to make the present invention easier to understand, and the present invention is not necessarily limited to having all of the described configurations. Furthermore, some of the configurations of one embodiment may be replaced with those of another embodiment. Furthermore, some of the configurations of one embodiment may be added to those of another embodiment. Furthermore, some of the configurations of each embodiment may be added, deleted, or replaced with those of other embodiments.

[0080] Furthermore, each of the aforementioned configurations, functions, processing units, and processing means may be implemented in hardware, for example, by designing them as integrated circuits, or they may be implemented in software by having a processor interpret and execute programs that realize each function.

[0081] Information such as programs, tables, and files that implement each function can be stored in memory, hard disks, SSDs (Solid State Drives), or recording media such as IC cards, SD cards, and DVDs.

[0082] Furthermore, the control lines and information lines shown are those deemed necessary for explanation purposes and do not necessarily represent all control lines and information lines required for implementation. In reality, it can be assumed that almost all components are interconnected.

Claims

1. A vehicle operation target estimation system for estimating the operational target of a train formation, comprising a computer having a processing unit that performs predetermined processing and a storage unit connected to the processing unit, the system comprising: a route feature generation unit that generates route feature quantities based on the operation timetable; an operation feature generation unit that generates operation feature quantities based on operation information; a deterioration prediction formula generation unit that estimates a deterioration prediction formula using the generated operation feature quantities as explanatory variables; an operation change pattern generation unit that generates multiple patterns for the operation of a train formation; a remaining lifespan / failure occurrence range estimation unit that estimates remaining lifespan, failure occurrence range, and failure occurrence statistics according to the generated operation patterns; and a remaining lifespan / operation feature modeling unit that estimates a model representing the relationship between remaining lifespan and operation feature quantities from the estimated remaining lifespan, failure occurrence range, and failure occurrence statistics.

2. A vehicle operation target estimation system according to claim 1, comprising: an effective feature extraction unit that extracts feature quantities that contribute to specific parameters; and a remaining life prediction unit that calculates the remaining life by changing the extracted feature quantities.

3. A vehicle operation target estimation system according to claim 1, wherein the remaining lifespan / operational feature modeling unit estimates a model showing the relationship between the remaining lifespan and the operational feature, or the relationship between the median failure occurrence and the operational feature, from at least one of operational feature quantities, remaining lifespan, failure occurrence range, and median failure occurrence, which are estimated from multiple estimated or extracted from history at multiple time points.

4. A vehicle operation target estimation system according to claim 3, wherein the remaining lifespan / operation feature modeling unit generates a predetermined m operation patterns by changing the feature quantities, creates a multiple regression model with an m-dimensional vector whose elements are the remaining lifespans of the m generated operation patterns as the objective variable and an m x p-dimensional matrix whose elements are the operation feature quantities in the m operation patterns as the explanatory variables, and predicts the remaining lifespan using the created multiple regression model.

5. A vehicle operation target estimation system according to claim 1, characterized by comprising a purpose-based operation planning unit that generates an operation pattern by changing the extracted feature quantities.

6. A vehicle operation target estimation system according to claim 1, characterized in that it selects feature quantities that contribute to the purpose of the operation change selected by the user, and generates an operation plan in which the feature quantities change according to the direction of the selected feature quantities.

7. A vehicle operation target estimation method for estimating the operation target of a train formation, wherein the vehicle operation target estimation system is composed of a computer having a processing unit that performs predetermined processing and a storage unit connected to the processing unit, and the vehicle operation target estimation method is characterized in that the processing unit generates a feature quantity of the train schedule based on the train schedule, the processing unit generates a feature quantity of the operation based on the operation information, the processing unit estimates a prediction formula for the degree of deterioration using the generated feature quantity of the operation as an explanatory variable, the processing unit generates a plurality of patterns for the operation of the train formation, the processing unit estimates the remaining lifespan, the failure occurrence range, and the failure occurrence statistics according to the generated operation patterns, and the processing unit estimates a model representing the relationship between the remaining lifespan and the feature quantity of the operation from the estimated remaining lifespan, the failure occurrence range, and the failure occurrence statistics.

Images