CN111853113B

CN111853113B - Wear prediction device, wear prediction method, and computer-readable recording medium

Info

Publication number: CN111853113B
Application number: CN202010146129.9A
Authority: CN
Inventors: 丸地康平; 佐藤诚
Original assignee: Toshiba Corp; Toshiba Infrastructure Systems and Solutions Corp
Current assignee: Toshiba Corp; Toshiba Infrastructure Systems and Solutions Corp
Priority date: 2019-04-26
Filing date: 2020-03-05
Publication date: 2022-08-23
Anticipated expiration: 2040-03-05
Also published as: TW202040444A; ZA202001198B; JP2023129446A; CN111853113A; JP2020183767A; JP7454088B2

Abstract

Provided are a wear prediction device, a wear prediction method, and a computer-readable recording medium, which predict wear information of a worn member using data measured during operation control of a vehicle. The wear prediction device according to the present embodiment includes: a model construction unit that constructs an estimation model of the wear state based on a history of a measurement value of a pressing force applied to a wear member provided in a vehicle by a mechanism that presses the wear member against an object to act on the vehicle and a history of an inspection value indicating a wear state of the wear member; and a prediction unit that predicts a wear state of a wear member provided in the target vehicle, using the estimation model.

Description

Wear prediction device, wear prediction method, and computer-readable recording medium

The present application is based on Japanese patent application 2019 and 086732 (application date: 04/26/2019), according to which priority is enjoyed. This application incorporates by reference the entirety of this application.

Technical Field

Embodiments of the present invention relate to a wear prediction device, a wear prediction method, and a computer-readable recording medium.

Background

In order to continue safe and stable operation of the railway system, it is essential to check the integrity of the railway system and to perform repairs as needed. In particular, it is important to inspect parts, such as brake shoes (shoes) or sliding plates of pantographs, which are worn away each time they are used. When wear exceeding a predetermined level is confirmed during inspection, the worn member is replaced, thereby ensuring the safety of the railway system.

Conventionally, as a method of estimating the amount of wear of a brake shoe or a slide plate, there is a method of estimating the amount of wear of a brake pad (brake pad) based on a travel distance. In addition, there is also a method of estimating the amount of wear based on the integrated travel distance and the temperature at the time of braking. In addition, there is a method of detecting the wear amount of the pantograph using a video camera.

However, when these methods are applied to a vehicle not equipped with a sensor necessary for estimation (for example, a temperature sensor at a specific portion, a video camera, or the like), the sensor needs to be newly added.

Disclosure of Invention

In an embodiment of the present invention, the wear state of the wearing part is predicted using data measured in the operation control of the vehicle.

Means for solving the problems

The wear prediction device according to the present embodiment includes: a model construction unit that constructs an estimation model of the wear state based on a history of a measurement value of a pressing force applied to a wear member provided in a vehicle by a mechanism that presses the wear member against an object to act on the vehicle and a history of an inspection value indicating a wear state of the wear member; and a prediction unit that predicts a wear state of a wear member provided in the target vehicle, using the estimation model.

Drawings

Fig. 1 is a block diagram of a wear prediction device according to embodiment 1.

Fig. 2 is a diagram showing an example of the inspection data.

Fig. 3 is a diagram showing a 1 st example of the vehicle data.

Fig. 4 is an explanatory diagram of the brake shoe, the brake cylinder pressure, and the air spring pressure.

Fig. 5 is an explanatory view of a slide plate of the pantograph.

Fig. 6 is a view showing a 2 nd example of the vehicle data.

Fig. 7 is a diagram showing examples of learning data and teacher data.

Fig. 8 is a diagram showing another example of learning data and teacher data.

Fig. 9 is a diagram showing an example of operation data (an actual operation result and an operation plan).

Fig. 10 is a diagram showing an example of data display based on the prediction result.

Fig. 11 is a diagram showing another example of data display based on the prediction result.

Fig. 12 is a diagram showing still another example of data display based on the prediction result.

Fig. 13 is a flowchart of the process of the learning phase.

Fig. 14 is a flowchart of the process of the prediction phase.

Fig. 15 is a block diagram of the wear prediction device according to embodiment 2.

Fig. 16 is a diagram showing a hardware configuration of the wear prediction device according to any one of embodiments 1 to 3.

Description of the reference symbols

11: inspection data storage unit

12: vehicle data storage unit

13: model construction unit

14: model storage unit

15: operation data storage unit

16: wear prediction unit (prediction unit)

17: prediction result storage unit

18: prediction result output unit

19: environmental data storage unit

Detailed Description

Hereinafter, embodiments of the present invention will be described with reference to the drawings.

Fig. 1 is a block diagram of a wear prediction device according to an embodiment of the present invention. The wear prediction device of fig. 1 includes an inspection data storage unit 11, a vehicle data storage unit 12, a model construction unit 13, a model storage unit 14, an operation data storage unit 15, a wear prediction unit (hereinafter, referred to as a prediction unit) 16, a prediction result storage unit 17, and a prediction result output unit 18. The wear prediction device of fig. 1 may further include an input device for inputting instructions or data by a user who is an operator or manager of the device. In this case, the input device is, for example, a keyboard, a mouse, a touch panel, a smartphone, or the like.

The wear prediction device of fig. 1 constructs a model for estimating a wear state (wear amount, presence or absence of replacement, or the like) of a wear member used in a vehicle by collecting and analyzing data relating to the vehicle such as a railway vehicle. Then, the future wear state of the wear member of the target vehicle is predicted using the model. This makes it possible to perform inventory management of the wear member and/or optimization of the inspection timing. In the present embodiment, a railway vehicle is used as the vehicle, but the same embodiment can be realized even for other types of vehicles such as an automobile. The railway vehicle may be a single vehicle or a vehicle formation (formation) in which a plurality of vehicles are connected.

[ learning phase ]

The inspection data storage unit 11 stores a history of inspection values (inspection data) of the worn parts to be monitored among the worn parts of the vehicles in each vehicle formation. The worn member is a member that may be a replacement target due to wear. The inspection data includes a measurement history of the worn part or a replacement history of the worn part. As an example of the wearing part, a brake shoe (brake pad) for braking a wheel of a vehicle is included. Further, a trolley of a pantograph which collects current from a trolley (trolley) line (overhead wire) is also included. Other components are also possible.

The wear member is pressed against the object by a mechanism provided in the vehicle during operation of the vehicle to provide an action on the vehicle. When the wearing member is a brake shoe, the object is a wheel of a vehicle, and the mechanism is a cylinder (BC: brake cylinder) for a brake, the brake shoe is pressed against the wheel by the brake cylinder, and the vehicle is decelerated (braked). When the wear member is a slide plate of a pantograph, the target object is an overhead wire, and the mechanism is a pantograph, the slide plate is pressed against the overhead wire by the pantograph, and the current is collected from the overhead wire to the vehicle.

Each wear member is given a wear member ID. For example, the following are set: the number of vehicles in a certain vehicle formation is 10, and brake shoes are provided as wear members in 4 places for each vehicle. When all the brake shoes are to be monitored, inspection data for a total of 40 worn parts is stored in the vehicle formation. The inspection data is similarly stored for other vehicle formations. The inspection data storage unit 11 or another storage unit not shown may store information on which position of which vehicle in which vehicle formation each wear member is provided. Further, it is assumed that: when a wear member provided at a certain position is replaced with a new wear member, the same ID as that of the wear member before replacement is used as the wear member ID of the wear member after replacement. However, the ID may be changed before and after the replacement.

Fig. 2 (a) shows an example of a measurement history of the wear amount of each wear member (brake shoe in this case) as inspection data. Fig. 2 (B) shows an example of a replacement history of a worn component (here, a brake shoe) of each worn component as inspection data. In fig. 2 (a) or 2 (B), a measurement history or a replacement history of the worn part of ID001 is displayed.

The measurement history in fig. 2 a includes the time when the inspection was performed (inspection time) and the measured value of the wear amount. In this example, the examination (measurement) is performed approximately every 3 months. The amount of remaining thickness may also be determined instead of the amount of wear.

In the case of the replacement history shown in fig. 2 (B), the information includes the time when the inspection was performed (inspection time) and the presence or absence of replacement. In this example, information indicating whether replacement is not performed is stored for each inspection time. However, the form of the replacement history is not limited to this. For example, only the time when the worn component is replaced may be stored.

The vehicle data storage unit 12 stores, as vehicle data, a history of measurement values of sensors (on-vehicle sensors) mounted on vehicles in the formation of vehicles and a history of driving command values in the formation of vehicles.

Fig. 3 shows an example (example 1) of the vehicle data stored in the vehicle data storage unit 12. In this example, data such AS a traction command, a Brake command, an Air Spring (AS) pressure and a Brake Cylinder (BC) pressure of each vehicle are stored in a time series in a certain vehicle formation. In this example, the data is time series data every 1 second. The same vehicle data is also stored for other vehicle formations. The vehicle formation ID may be assigned to each vehicle formation, and the vehicle data of the vehicle formations may be stored in the same database. The brake command value is a control command value of the brake corresponding to a plurality of brake notches indicated by a brake lever provided in the cab. The larger the value of the brake command, the larger the braking force. The value of the traction command is a control command value for performing acceleration and deceleration corresponding to a plurality of traction notches indicated by a traction bar provided in the cab. A larger value of the traction command means a larger acceleration to speed. The brake lever and the drawbar are examples of means for providing a braking command and a towing command, and these commands may be provided by other means, for example, a steering wheel.

Here, the brake shoe, BC pressure, and AS pressure will be described with reference to fig. 4.

Fig. 4 schematically shows a wheel of a vehicle running on a rail 21 and its peripheral structure. Wheels 22 are carried on the rails 21. A tread brake (tread brake)23 is provided as one type of air brake in the vehicle. The wheel 22 is braked by the tread brake 23. Here, only one wheel 22 is shown, but actually, a plurality of sets of a pair of right and left wheels are provided in one vehicle. As an example, four air brakes are provided for each vehicle. However, there may be a vehicle in which no air brake is provided. Further, a plurality of air brakes may be provided for one vehicle.

The tread brake 23 uses the air cylinder 24 as power. The brake shoe 25 is pressed against the tread surface (surface in contact with the rail 21) of the wheel 22 by increasing the brake cylinder pressure (BC pressure) which is the pressure in the cylinder 24. BC pressure is a force pressing the brake shoe 25 against the tread surface of the wheel, and the force pressing the brake shoe 25 against the wheel 22 is proportional to BC pressure. The frictional force between the wheel 22 and the brake shoe 25 becomes the braking force of the tread brake 23.

The brake shoes 25 are subject to constant wear from use due to frictional forces between the wheel 22 and the brake shoes 25. When the brake shoe 25 is worn, there is a possibility that the braking force is reduced, and therefore, the brake shoe 25 needs to be replaced according to the amount of wear. Therefore, the amount of wear of the brake shoe 25 is measured during the inspection, and if the amount of wear is equal to or greater than a predetermined value, the brake shoe is replaced with a new one. The measured wear amount is stored in the inspection data storage unit 11 as an inspection value together with the inspection date and time.

In fig. 4, the cylinder (cylinder) for applying pressure to the brake shoe 25 is a cylinder in which the working fluid is air, but may be a cylinder using a working fluid other than air, such as a hydraulic cylinder.

In addition to the wear of the brake shoe 25, the braking force of the tread brake 23 also varies according to the load applied to the vehicle. As shown in fig. 4, a load compensating device 26 is mounted on the vehicle. The load compensation device 26 includes an air spring 27, and can measure the load applied to the vehicle by detecting the air spring pressure (AS pressure) of the air spring 27. The as (air spring) pressure depends on the number of passengers of the vehicle. The larger the number of passengers, the larger the AS pressure. The braking force of the brake is adjusted on the basis of the AS pressure detected by the load compensation device 26. For example, the greater the AS pressure, the more the braking force of the brake is enhanced. The greater the braking force of the brake, the greater the amount of wear of the brake shoe 25, and therefore the AS pressure affects the amount of wear of the brake shoe 25.

As described above, the wear member may be another member other than the brake shoe, such as a slide plate of the pantograph.

Fig. 5 is a perspective view of the pantograph. The pantograph 31 is a device provided on a roof of a vehicle and receiving power from a trolley wire (overhead wire) 32. The pantograph slider 33 is a friction member provided on the upper portion of the pantograph 31. The sled 33 receives electric power while moving in contact with the trolley wire (while rubbing against the trolley wire). Therefore, the sled 33 is continuously worn by friction with the trolley wire. When the amount of wear becomes a predetermined value or more, an operation such as replacement with a new product is performed.

In the case where the wear member is a slide plate of a pantograph, the vehicle data may include at least one of a pressure for pressing the slide plate, a displacement in the vertical direction of the pantograph, a speed, and a current, a voltage, and the like flowing through the overhead wire. In the case of a structure in which a slide plate is pressed against a power line by a spring as an example, the pressing force of the pressing slide plate may be spring pressure. All or a part of the items shown in fig. 3 may be included in the vehicle data.

The vehicle data may be data obtained by integrating the measurement values of the onboard sensors and the operation commands for each operation of the vehicle formation, instead of the time-series data as shown in fig. 3. An example of the vehicle data in this case will be described with reference to fig. 6.

Fig. 6 shows another example of the vehicle data (example 2). For a certain vehicle formation, an operation ID is set for each operation. The departure time, the end time, the integrated time of each value of the traction command, the integrated time of each value of the brake command, and the AS pressure integrated value and BC pressure integrated value of each vehicle of the vehicle are stored for each operation. The same data is stored for other vehicle formations. Here, the operation means that the vehicle formation travels from the departure point to the end point according to a predetermined operation schedule (operation plan). Here, the accumulated value is stored for each operation, but the accumulated value may be stored in other units such as each section of the station.

The model construction unit 13 constructs an estimated model of the wear state (the amount of wear, the necessity of replacement, or the like) using the vehicle data (the history of the measured values and the history of the operation commands) as learning data and the inspection data (the measurement history of the amount of wear or the replacement history of the worn component) as teacher data.

Examples of constructing the estimation model are shown in the case where the measurement history of the worn component (see fig. 2 a) and the replacement history (see fig. 2B) are used as the teacher data.

[ case of using measurement history of worn parts as teacher data ]

Learning data corresponding to a plurality of explanatory variables and teacher data corresponding to a target variable are created for each measurement period (inspection timing) of inspection data based on the time immediately after replacement (time at which the wear amount is 0). Learning data and teacher data are created for each wear part.

Fig. 7 shows an example of the created learning data and teacher data. In this example, the measurement history of the wear amount shown in fig. 2 (a) is used as the teacher data.

For example, from check period 2018/3/1010 in fig. 2 (a): 00 to next check period 2018/6/817: the amount of wear during the period up to 00 (inspection interval) increased by 0.1. Corresponding to 2018/6/817: in the inspection timing of 00, the wear amount 0.1 is set as the teacher data in the first row of fig. 7.

The values of the explanatory variables in the above-described period are calculated from the vehicle data, and the calculated values of the explanatory variables are cumulatively added to be set as learning data. As an explanatory variable, data that influences the wear of the wearing part is preferably used. In the example of the figure, the accumulated value of the traction command value 1, the accumulated value of the traction command value 2, the accumulated value of the traction command value 3, … …, the accumulated value of the traction command value 6, the accumulated value of the AS pressure, the accumulated value of the BC pressure, and the like are set AS learning data. The cumulative value of the traction command value 0 may be set.

The cumulative BC pressure value in fig. 7 is a cumulative BC pressure value measured for a vehicle provided with a worn component. However, the average of the cumulative BC pressure values measured for a plurality of vehicles (for example, front and rear 1 vehicle) including the vehicle may be used, or the average of the cumulative BC pressure values measured for all the vehicles may be used. The contents described herein can be applied to the AS pressure in the same manner.

The items of the explanatory variables are not limited to the example of fig. 7. For example, the number of times of sudden braking (sudden braking) may be accumulated, or the accumulated value of the velocity (the velocity is a value associated with the rotation velocity of the wheel) may be used. The speed may also be corrected based on the wheel diameter. Note that different explanatory variables may be used depending on whether the vehicle is in a stopped state or an operating state. The value of the explanatory variable may be calculated based on only the data during braking.

In addition, any of the vehicle data in fig. 3 and 6 may be used regardless of the form of the vehicle data used for creating the learning data.

Teacher data and learning data are similarly created for other examination times in the examination data.

As described above, after the learning data and the teacher data as shown in fig. 7 are prepared, an estimation model for calculating a target variable (in the example of fig. 7, a wear amount) from explanatory variables is constructed by machine learning. The algorithm used in learning may be any algorithm such as linear regression, logistic regression, decision tree, random forest, neural network, and the like. A plurality of types of models may be constructed, and a model with the smallest model error (described later) may be selected. Alternatively, a model may be used which combines the results of a plurality of models to output one solution, such as ensemble learning (ensemble learning).

Hereinafter, a method of constructing the estimation model will be described. Consider the case where a linear regression model is used. Equation (1) represents an example of a function (model function) of the estimation model.

y＝b ₁ x ₁ +b ₂ x ₂ +b ₃ x ₃ +…+b _n x _n +b ₀ …(1)

y is a target variable, x ₁ ～x _n Is an explanatory variable, b ₀ ～b _n Are regression coefficients. In order to absorb the difference in the measurement units of the respective explanatory variables, the target variable and all the explanatory variables may be normalized so that the average value is 0 and the variance is 1.

The calculation of the explanatory variables and the regression coefficients (parameter calculation) is performed by, for example, minimizing an objective function defining a difference between the output value of the function and the teacher data (i.e., a model error). For example, the explanatory variable and the regression coefficient are calculated by solving a problem of minimizing a square error described below. i represents the number of learning data (number of teacher data).

min{∑ _i |y-(b ₀ +b ₁ x ₁ +b ₂ x ₂ +b ₃ x ₃ +…+b _n x _n )| ² }…(2)

[ case where replacement history of worn parts is used as teacher data ]

When the replacement history of the worn component is used as the inspection data, a target variable of 2 values, which is whether replacement is performed or not, is set as teacher data (value of the target variable) for each inspection time with reference to the time immediately after replacement. The values of the explanatory variables in the period from the previous inspection timing to the current inspection timing are calculated from the vehicle data, and are set as learning data. The learning data is created in the same manner as the above-described method.

Fig. 8 shows an example of the created learning data and teacher data. The same as fig. 7 is true except that the value (teacher data) of the target variable is 2-value information indicating the presence or absence of replacement. Regarding the value of the target variable, 1 means replacement, and 0 means no replacement.

In this case, the algorithm used in learning may use any algorithm such as SVM, logistic regression, decision tree, random forest, neural network, or the like. As an example, if the teacher data indicates the presence or absence of replacement, the target variable indicates the presence or absence of replacement as an example.

For example, in the case of logistic regression, a logistic function (sigmoid function) may be used as the model function. Equation (3) represents an example of a logic function. The definitions of the symbols in the formula are the same as those in formula (1). For the regression coefficient b ₀ ～b _n The calculation of (b) can be performed by a general method by optimizing (minimizing or maximizing) a predetermined objective function. The output value of the logic function is in a range greater than 0 and smaller than 1. The output value of the logic function corresponds to the probability that the worn part has to be replaced.

A plurality of types of models may be constructed, and a model having a minimum statistical value or a threshold value or less, such as an average value, a minimum value, and a maximum value of model errors, may be selected. The model error is a difference between the output of the model (the value of the target variable) and the value of the test data. When teacher data used in learning is used as test data for testing, explanatory variables calculated from vehicle data used in learning are used as inputs to the model. When the inspection data not used in the learning is used as the inspection data for the test, the explanatory variable calculated from the vehicle data not used in the learning is used as the input of the model.

Further, a presumption model may be constructed using life time analysis. In this case, one explanatory variable is selected, and the lifetime analysis is performed with the value of the selected explanatory variable as time, whereby an estimation model can be constructed. The algorithm may be any algorithm such as Kaplan-Meier method, Weibull method, or the like. When the BC pressure is used as the explanatory variable as an example, a graph (graph) in which the horizontal axis is an accumulated value of the BC pressure and the vertical axis is a survival probability is created as the estimation model based on the vehicle data and the inspection data. The teacher data uses 2-value information of whether or not replacement has been performed. Based on the vehicle data for testing, an accumulated value of the BC voltage is calculated, and a survival probability corresponding to the calculated value is determined according to the coordinate graph. If the determined probability of survival is less than the threshold, it is determined that there is a need to replace the worn component.

The model storage unit 14 stores the estimation model constructed by the model construction unit 13. Since the estimation model is generated for each worn component that is the object to be monitored, the estimation model is stored in the model storage unit 14 for each worn component that is the object to be monitored. Further, the type of model constructed for each type of formation, material and/or type of slide plate and brake shoe may be changed.

[ prediction stage ]

The operation data storage unit 15 stores an actual operation result and an operation plan (operation schedule data) of each vehicle formation as operation data.

Fig. 9 (a) and 9 (B) show an example of the operation data. Fig. 9 (a) shows operation actual result data including a departure time, a final arrival time, a departure station, a terminal station, and the like, with respect to the past operation of each vehicle convoy. Fig. 9 (B) shows operation plan data including a departure time, a final arrival time, a departure station, a terminal station, and the like, with respect to operations planned by each vehicle formation. A run ID is associated with each run. The operation data storage unit 15 or a storage unit not shown may store data in which the operation ID and the ID of the vehicle formation are associated with each other. Information of other items may be further added to the operation actual result data of fig. 9 (a) and the operation plan data of fig. 9 (B). For example, information on a station that stops or passes on the way may be added.

The prediction unit 16 predicts the wear state of the wear member used in the target vehicle, using the estimation model stored in the model storage unit 14, the operation plan data of the target vehicle, the operation actual result data of at least 1 vehicle (the target vehicle may be included or may not be included), and the vehicle data (the history of the measurement values and the history of the driving commands) in the vehicle data storage unit 12. Specifically, the prediction unit 16 predicts the values of the explanatory variables when the operation plan indicated by the operation plan data of the target vehicle is executed, assigns the predicted values of the explanatory variables to the explanatory variables of the estimation model (used as input variables of the estimation model), and calculates the output values (target variables) of the estimation model. The output value of the estimation model is a predicted value of the wear state of the worn component.

In order to calculate the predicted value of the explanatory variable, an instance (target operation actual result data) similar to the operation plan of the target vehicle is extracted from the past operation actual result data. For example, the following are set: regarding a certain vehicle formation, as an operation plan 1 day after the current day, there is a plan (operation plan of target vehicle) from a good quality station to a Δ station for 2 hours of operation. In this case, the case most similar to the operation plan is searched for in the past operation actual result. For example, the cases of operation in the same section from the good quality station to the Δ station are determined, and the case is selected from the determined cases based on the difference from the operation time (2 hours) of the target vehicle. For example, the case with the smallest difference in the running times is searched as the similar case. The value of the explanatory variable is calculated from the vehicle data corresponding to the found case. The calculated value of the explanatory variable is used as a predicted value of the explanatory variable of the operation plan for the target vehicle. Here, one case is selected as a similar case, but a plurality of cases may be selected. In this case, for example, the values of the vehicle data corresponding to the selected case may be averaged, and the value of the explanatory variable may be calculated based on the averaged vehicle data.

In searching for similar cases in the past, for example, the similarity of the vehicle or devices (wear parts, other parts, or the like) used in the vehicle may be considered. For example, the search may be performed by narrowing down the data of the vehicle formation of the same type of wear member (for example, the same type of brake shoe used in each vehicle). This can improve the accuracy of calculating the predicted value of the explanatory variable.

After the predicted value of the explanatory variable is calculated, the prediction unit 16 calculates an output value of the estimation model (value of the target variable) using the calculated predicted value as an input of the estimation model. The calculated value is a predicted value of the wear state. When the teacher data used in the model construction is a measured value of the wear amount, the predicted value is the wear amount. When the teacher data used in the model building process indicates the presence or absence of replacement, the predicted value is the presence or absence of replacement necessity or probability. The predicted value is either the presence or absence of replacement or the probability thereof, depending on the algorithm used in the model construction. For example, in the case of logistic regression, the predicted value is the probability that replacement is necessary, and in the case of decision tree, the predicted value is the necessity of replacement. However, a decision tree may also be used to construct a model that predicts this probability. Instead of the probability of the necessity of replacement, the probability of the necessity of replacement may be used.

The prediction unit 16 can obtain time-series data of a predicted value of the wear state by repeating the same prediction processing for each business day of the target vehicle with respect to the wear component of the target vehicle.

The prediction result storage unit 17 stores the prediction value of the wear state calculated by the prediction unit 16. By repeating the prediction processing in the prediction unit 16, the time-series data of the predicted value of the wear information is stored in the prediction result storage unit 17. By performing the prediction processing similarly for the other wearing components of the target vehicle, the prediction result storage unit 17 stores the prediction values of the wearing states of the other wearing components or the time series data thereof.

The prediction result output unit 18 includes a display device that displays data or information on a screen. The prediction result output unit 18 generates data for display based on the prediction value stored in the prediction result storage unit 17, and displays the generated data on a screen. For example, time-series data of the predicted values are displayed on the screen based on the predicted values stored in the prediction result storage unit 17.

Fig. 10 shows an example of data display of the prediction result output unit 18. The transition of the predicted value of the wear amount is shown. The horizontal axis represents time, and the vertical axis represents the predicted value of the wear amount. The specified value (threshold value) of the wear amount that becomes the standard of replacement and the inspection date are also displayed. The predetermined value is a reference for replacing the wear member, and when the wear amount becomes equal to or greater than the predetermined value, the predetermined value is a reference for replacing the wear member. In the illustrated example, the predicted value of the wear amount is smaller than the predetermined value on the next inspection day, but the predicted value of the wear amount is equal to or larger than the predetermined value on the next inspection day. This makes it clear that the worn component is highly likely to be replaced or replaced on the next inspection day.

Fig. 11 is a table that stores IDs of worn parts whose predicted values of wear amounts of the worn parts to be monitored are equal to or greater than a predetermined value. The data is sorted in the order of the day that becomes the predetermined value or more from near to far. The worn component having ID122 is predicted to have a wear amount equal to or greater than a predetermined value before the next scheduled inspection day. Therefore, the background of the data of the ID122 is painted with a color for attracting attention.

Fig. 12 is a graph showing the predicted number of replacement of worn parts in time series. The horizontal axis represents time and the vertical axis represents the cumulative predicted number of replacements of the worn component. The cumulative replacement predicted number is a total of the number of wear components that are subject to all vehicle formations (or specific formations) that use the same type of wear component (e.g., wear components of the same model), and for which the predicted value of the wear amount is equal to or greater than a predetermined value. The cumulative number of replacements until the previous month was plotted on day 1 of each month (No. 1). In the examples of the figures, brake shoes and a slide plate of a pantograph are shown as examples of wearing parts.

By displaying a graph such as that of fig. 12, it is possible to easily grasp when the number of wear parts is necessary. The database may also be used to manage the inventory of worn parts. In this case, a time period when the stock becomes insufficient may be determined, and the attention mark may be displayed. In fig. 12, the cumulative predicted number of replacement brake shoes in 5 months becomes 12, exceeding 10 of the stock, and therefore, the attention mark M is marked.

In fig. 10 to 12, the output value of the estimation model is the estimated value of the wear amount, but the data may be displayed similarly even when there is a probability that the worn component needs to be replaced or whether the worn component needs to be replaced. In the case where the wear member has a probability of being replaced, if the probability is equal to or higher than a threshold value, it is determined that replacement is necessary, and if the probability is lower than the threshold value, it is determined that replacement is not necessary.

Fig. 13 is a flowchart of the learning phase according to the present embodiment. The present process starts according to a start trigger of the learning phase (S100). The start trigger may be, for example, an instruction from the user of the apparatus received from the input apparatus, a predetermined time, or the establishment of any other event.

When the learning phase starts, the model building unit 13 reads out the inspection data from the inspection data storage unit 11 and the vehicle data from the vehicle data storage unit 12 (S101).

Teacher data corresponding to the target variable is generated based on the read inspection data, and learning data corresponding to one or more explanatory variables is generated based on the vehicle data (S102).

Based on the generated teacher data and learning data, an estimation model as a model for estimating a target variable (e.g., wear amount) from one or more explanatory variables is constructed by machine learning using an arbitrary regression algorithm (linear regression, decision tree, random forest, neural network, or the like) (S103).

The model storage unit 14 stores the estimation model constructed by the model construction unit 13 in the interior in association with the model ID (S104). The model ID differs for each estimated model.

Such an estimation model is constructed for each wear component to be monitored (S105). As an example, the wear member to be monitored is all brake shoes of each vehicle provided in each vehicle formation. However, one brake shoe may be selected for each of a plurality of vehicles as a representative group, and the selected brake shoe may be used as a monitoring target. In this case, it may be determined that: when the brake shoe to be monitored is replaced, the other brake shoes belonging to the same group are also replaced at the same time. The wear member to be monitored may be determined by a method other than the method described here.

Fig. 14 is a flowchart of the prediction stage according to the present embodiment. The present process starts according to the start trigger of the prediction phase (S200). The start trigger may be, for example, an instruction received from the user of the own apparatus, may be a predetermined time, or may be the establishment of any other event. This process is performed for each wear member to be monitored with respect to the target vehicle.

The prediction unit 16 reads the operation plan data of the target vehicle from the operation data storage unit 15 (S201). As an example, operation plan data of the subject vehicle on a business day next to a day on which a worn member serving as a monitoring target in the subject vehicle is replaced is read.

An instance similar to the operation plan data of the subject vehicle is specified from the operation actual result data of 1 or more vehicles in the operation data storage unit 15 (S202).

The vehicle data corresponding to the specified case is read from the vehicle data storage unit 12, and the predicted value of the explanatory variable is calculated with respect to the wear member to be monitored (the wear member mounted on the subject vehicle) (S203).

The wear state (e.g., the amount of wear) is predicted based on the calculated predicted value of the explanatory variable and the estimation model corresponding to the worn component (S204).

The prediction result storage unit 17 stores the prediction value of the wear state in association with the worn component ID (S205).

The prediction result output unit 18 generates data for display (for example, a transition graph of the prediction values) based on the prediction values stored in the prediction result storage unit 17, and displays the generated data (S206).

For example, the above processing is performed while advancing one operation day (business day) at a time. When the value of the explanatory variable (for example, the cumulative value of the BC pressure) is calculated for each day (prediction target day) after day 2, the value of the explanatory variable on the prediction target day may be calculated from the vehicle data for the case specified for each day up to the prediction target. It is needless to say that the present process may be performed not one day after the worn part is replaced but a plurality of days after the worn part is replaced. In this case, similarly, similar past cases may be specified based on the operation schedule for each day from immediately after the replacement of the worn component to the prediction target day, and the value of the explanatory variable for the prediction target day may be calculated based on the vehicle data corresponding to the specified cases.

As described above, according to the present embodiment, the wear state (the amount of wear, the necessity of replacement, or the like) of the wear member at a future time can be predicted with high accuracy. The related art method merely estimates the current wear amount, and cannot estimate the wear amount at a future time point. In contrast, in the present embodiment, the wear state at each time point in the future can be accurately predicted at the current time point. This makes it possible to optimize the inspection timing of the wear member and also to optimize inventory control of the wear member. Further, according to the present embodiment, since learning is performed using data (BC pressure, AS pressure, traction command, brake command, and the like) measured in the operation control of the vehicle, it is not necessary to newly add a sensor or the like to the vehicle.

(modification example)

In embodiment 1 described above, the prediction unit 16 predicts the future wear state based on the operation plan data of the target vehicle, but may estimate the current or past wear state based on the operation actual result data of the target vehicle. For example, the wear state on an arbitrary day after the previous inspection period may be estimated from the actual operation result data.

(embodiment 2)

Fig. 15 is a block diagram of a wear prediction device according to embodiment 2 of the present invention. An environment data storage unit 19 is added to the wear prediction device of fig. 1. The environment data storage unit 19 stores data (environment data) relating to the operating environment in which the vehicle formation is performed.

As an example of the environment data, route information such as a route slope and a route curve is included. The route information is associated with location information on the map. In the present embodiment, the position information is also associated with the vehicle data (see fig. 3). Specific examples of the position information include mileage (distance from a departure point) and position information of a GPS (Global Positioning System).

Other examples of the environmental data include weather information such as air temperature and precipitation amount. The weather information is associated with location information and time information. As weather information, public data of a weather hall or the like may be used. When the weather information is used, the weather information of the observation position closest to the position information of the vehicle data is used.

In the present embodiment, the model construction unit 13 constructs an estimation model using a history of environmental data. For example, when the model building unit 13 creates the learning data, explanatory variables corresponding to the environmental conditions are added as the BC pressure integrated value in the slope section, the BC pressure integrated value in the non-slope section, the BC pressure integrated value in the curve section, the BC pressure integrated value in the non-curve section, the BC pressure integrated value in the rainy weather, and the like. Here, the slope section is a section in which the slope value is equal to or greater than a certain value, the non-slope section is a section in which the slope value is less than a certain value, the curve section is a section in which the radius of curvature is equal to or less than a certain value, and the non-curve section is a section in which the radius of curvature is equal to or greater than a certain value. Here, the slope is divided into two sections, and the curve is divided into two sections, but the slope may be divided into 3 or more sections, and explanatory variables corresponding to the respective sections may be added. Since the wear amount may vary depending on the degree of a slope or a curve, or weather such as rain or snow, the estimation accuracy can be improved by increasing the explanatory variable according to the environmental condition.

When estimating based on the estimation model, the operating environment in the operating period of the operation plan data of the target vehicle may be calculated, and the value of the additional explanatory variable may be calculated based on the environment data indicating the calculated operating environment.

In the present embodiment, the similarity of the environmental data may be used when retrieving a case similar to the operation plan data of the target vehicle from the operation actual result data. For example, the operating environment (here, season) during the operating period of the operation plan data of the target vehicle is calculated, and the search range in the actual operation result data is narrowed down to the same season as the calculated season. In another example, values such as the precipitation amount are determined for each case based on past weather information, and a case having a value close to the weather forecast value is selected. This can improve the accuracy of calculating the predicted value of the explanatory variable. The selection of the case based on the method may be combined with the generation of the learning data of the present embodiment described above, or may not be combined.

(embodiment 3)

In embodiment 1 or embodiment 2 described above, a correction term (explanatory variable) corresponding to the configuration of the vehicle or the device mounted on the vehicle may be added to the model function. Examples thereof include a correction term corresponding to whether the brake shoe is a brake shoe of a model car, a correction term corresponding to whether the brake shoe is an M car (a vehicle having power such as a motor) or a T car (a vehicle without power such as a motor), a correction term corresponding to whether the brake shoe is a front-wheel brake shoe or a rear-wheel brake shoe, a correction term corresponding to which of the left-hand wheel and the right-hand wheel the brake shoe is disposed in the traveling direction, and a correction term corresponding to the type of the brake shoe. Further, a correction term according to the type of vehicle formation, or a correction term according to the type of brake structure may be cited. When the wear member is a pantograph, the wear member includes a correction term according to the type of the pantograph structure, a correction term according to the type of the slide plate, and the like.

For example, in the case of a correction term based on a vehicle number, variables are prepared for each vehicle number, and a variable G1 indicating a vehicle number 1 is given a value G1(G1 is an arbitrary real number other than 0) when it is located in a brake shoe of the vehicle number 1, and is given a value 0 in the case of other vehicle numbers. Similarly, g2(g2 is any real number other than 0) is given to the brake shoe located in vehicle No. 2, and 0 is given to the other vehicle. The correction term is defined similarly for car No. 3 and later. The following equation (4) represents an example of the model function in this case. In this example, the vehicles are number 1 to 10 vehicles. The other types of correction terms described above can be similarly added.

y＝b ₁ x ₁ +b ₂ x ₂ +b ₃ x ₃ +…+b _n x _n +b ₀ +G1+G2+…+G10…(4)

The model function may be learned according to each condition without adding a correction term, and different estimation models may be configured. For example, different models may be constructed as the model for vehicle # 1 and the model for vehicle # 2.

The model construction unit 13 may analyze the vehicle data and select an explanatory variable used in the model function by learning. For example, the correlation with the target variable may be calculated for all the variables of the vehicle data, and a predetermined number of variables may be selected in descending order of correlation. For example, statistics (maximum value, minimum value, cumulative value, or the like) of all variables of the vehicle data may be calculated, the correlation with the objective function of each variable may be calculated for the statistics, and a predetermined number of variables may be selected in descending order of the correlation. In this case, the present process may be performed under the above-described conditions, and a model may be constructed under each condition.

In the case where the wear member is a pantograph slider, the explanatory variables may be defined by whether the vehicle is in a stopped state or in a moving state, or only one of the explanatory variables may be used. Similarly, the explanatory variables may be defined by whether or not the pantograph is being raised, or only one of the explanatory variables may be used. Further, a correction term (for example, a value a in the case of the air section, and a value b in the case of the outside) of whether or not the vehicle formation is in the air section (air section) section, or a correction term (for example, a value c in the case of the distance being equal to or less than a certain value, and a value d in the case of the distance being equal to or more than a certain value) of whether or not the distance from the substation of the vehicle formation is close may be added to the model function.

(embodiment 4)

Fig. 16 shows a hardware configuration of the wear prediction device according to the present embodiment. As the wear prediction device according to the present embodiment, the wear prediction device according to any one of embodiments 1 to 3 can be used. The wear prediction device according to the present embodiment is constituted by a computer device 100. The computer device 100 includes a CPU101, an input interface 102, a display device 103, a communication device 104, a main storage device 105, and an external storage device 106, which are connected to each other via a bus 107.

A CPU (central processing unit) 101 executes a wear prediction program as a computer program on a main storage device 105. The wear prediction program is a program that realizes the above-described functional configurations of the wear prediction device. The CPU101 executes the wear prediction program to realize each functional configuration.

The input interface 102 is a circuit for inputting an operation signal from an input device such as a keyboard, a mouse, and a touch panel to the wear prediction device.

The display device 103 displays data or information output from the wear prediction device. The display device 103 is, for example, an LCD (liquid crystal display), a CRT (cathode ray tube), and a PDP (plasma display), but is not limited thereto. The data or information generated by the prediction result output unit 18 can be displayed on the display device 103.

The communication device 104 is a circuit for the wear prediction device to communicate with an external device by wireless or wired communication. The inspection data, the vehicle data, the operation data, and the environment data can be input from an external device via the communication device 104. The inspection data, the vehicle data, the operation data, and the environment data inputted from the external device can be stored in the inspection data storage unit 11, the vehicle data storage unit 12, the operation data storage unit 15, and the environment data storage unit 19. As an example, the communication device 104 may acquire vehicle data by communicating with a communication device mounted on the vehicle, and store the acquired vehicle data in the vehicle data storage unit 12.

The main storage 105 stores a wear prediction program, data necessary for execution of the wear prediction program, data generated by execution of the wear prediction program, and the like. The wear prediction program is deployed onto main storage 105 and executed. The main memory 105 is, for example, a RAM, a DRAM, or an SRAM, but is not limited thereto. The inspection data storage unit 11, the vehicle data storage unit 12, the operation data storage unit 15, and the environment data storage unit 19 may be built in the main storage device 105.

The external storage device 106 stores a wear prediction program, data necessary for execution of the wear prediction program, data generated by execution of the wear prediction program, and the like. These programs and/or data are read out to main memory 105 upon execution of the wear prediction program. The external storage device 106 is, for example, a hard disk, an optical disk, a flash memory, and a magnetic tape, but is not limited thereto. The inspection data storage unit 11, the vehicle data storage unit 12, the operation data storage unit 15, and the environment data storage unit 19 may be constructed on the external storage device 106.

The wear prediction program may be installed in the computer device 100 in advance, or may be stored in a storage medium such as a CD-ROM. In addition, the wear prediction program may also be uploaded onto the internet.

The wear prediction device may be configured by a single computer device 100, or may be configured as a system including a plurality of computer devices 100 connected to each other via a network. The wear prediction device may be disposed on the cloud and receive an operation input from a user via the internet.

[ solution 1]

A wear prediction device is provided with:

a model construction unit that constructs an estimation model of the wear state based on a history of a measurement value of a pressing force applied to a wear member provided in a vehicle by a mechanism that presses the wear member against an object to act on the vehicle and a history of an inspection value indicating a wear state of the wear member; and

and a prediction unit that predicts a wear state of a wear member provided in the target vehicle, using the estimation model.

[ solution 2]

According to the wear prediction device described in claim 1,

the prediction unit obtains a predicted value of a pressing force applied to the wear member by the mechanism of the target vehicle based on the operation plan data of the target vehicle, and predicts the future wear state of the wear member based on the predicted value of the pressing force.

[ solution 3]

According to the wear prediction device described in claim 2,

the prediction unit selects at least one target operation actual result data from a plurality of operation actual result data of at least 1 vehicle based on the operation plan data of the target vehicle, and uses the measured value of the pressing force during the operation period indicated by the at least one target operation actual result data as the predicted value of the pressing force.

[ solution 4]

According to the wear prediction device described in claim 3,

the plurality of operation actual result data include environment data representing an operation environment of the vehicle,

the prediction unit calculates an operation environment of the target vehicle based on the operation plan data, and specifies the target operation actual result data based on the calculated operation environment.

[ solution 5]

The wear prediction device according to claim 3 or 4,

the prediction unit may use, as the target operation actual result data, operation actual result data having the same departure point and the same final arrival point as the operation plan data.

[ solution 6]

According to the wear prediction device of claim 5,

the prediction unit may use, as the target operation actual result data, operation actual result data in which a difference from the operation time indicated by the operation plan data is minimum or equal to or less than a threshold value.

[ solution 7]

The wear prediction device according to any one of claims 2 to 6,

the model construction unit constructs the estimation model based on a history of environmental data representing an operating environment of the vehicle,

the prediction unit calculates an operation environment of the target vehicle based on the operation plan data, and predicts the wear state based on the calculated operation environment.

[ solution 8]

The wear prediction device according to any one of claims 2 to 7,

the estimation model cumulatively adds the predicted values of the pressing force during the operation period indicated by the operation plan data of the target vehicle, and calculates the output value of the estimation model using the cumulative value as the input of the estimation model.

[ solution 9]

The wear prediction device according to any one of claims 1 to 8,

the wear part is a brake shoe plate,

the mechanism is a cylinder for a brake,

the object is a wheel of the vehicle,

the effect is a deceleration of the vehicle,

the check value indicates an amount of wear or a presence or absence of replacement of the brake shoe.

[ solution 10]

According to the wear prediction device of claim 9,

the pressing force is a pressure of the cylinder.

[ solution 11]

The wear prediction device according to any one of claims 1 to 8,

the mechanism is a pantograph which is arranged at the front end of the pantograph,

the wear part is a sliding plate of the pantograph,

the object is a cable to be erected,

the action is the collection of current to the vehicle,

the inspection value indicates an amount of wear or presence or absence of replacement of the pantograph.

[ solution 12]

A wear prediction method comprising:

a model construction step of constructing an estimation model of the wear state based on a history of a measurement value of a pressing force applied to a wear member provided in a vehicle by a mechanism that presses the wear member against an object to act on the vehicle and a history of an inspection value indicating a wear state of the wear member; and

and a prediction step of predicting a wear state of a wear member provided in the target vehicle, using the estimation model.

[ solution 13]

A recording medium storing a computer program for causing a computer to execute:

a model construction step of constructing an estimation model of the wear state based on a history of a measurement value of a pressing force applied to a wearing member provided in a vehicle by a mechanism that presses the wearing member against an object to act on the vehicle and a history of an inspection value indicating a wear state of the wearing member; and

The present invention is not limited to the above embodiments, and constituent elements may be modified and embodied in the implementation stage without departing from the spirit thereof. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments. For example, a configuration may be considered in which some of the components are deleted from all the components shown in the embodiments. Further, the constituent elements described in the different embodiments may be appropriately combined.

Claims

1. A wear prediction device is provided with:

a model construction unit that constructs an estimation model of a wear state of a wear member provided in a vehicle based on a history of a measurement value of a pressing force applied to the wear member by a mechanism that presses the wear member against an object to act on the vehicle and a history of an inspection value indicating the wear state of the wear member; and

a prediction unit that predicts a wear state of a wear member provided in the target vehicle using the estimation model,

the prediction unit obtains a predicted value of a pressing force applied to the wear member by the mechanism of the subject vehicle based on operation schedule data of the subject vehicle, and predicts the future wear state of the wear member based on the predicted value of the pressing force.

2. The wear-prediction device of claim 1,

3. The wear-out prediction device of claim 2,

4. The wear prediction device of claim 2 or 3,

the prediction unit uses, as the target operation actual result data, operation actual result data having the same departure point and the same final arrival point as the operation plan data.

5. The wear-out prediction device according to claim 4,

6. The wear prediction device according to any one of claims 1 to 3 and 5,

the prediction unit calculates a running environment of the target vehicle based on the running plan data, and predicts the wear state based on the calculated running environment.

7. The wear prediction device according to any one of claims 1 to 3 and 5,

8. The wear prediction device according to any one of claims 1 to 3 and 5,

the wear part is a brake shoe plate,

the mechanism is a cylinder for a brake,

the object is a wheel of the vehicle,

the effect is a deceleration of the vehicle,

9. The wear-out prediction device of claim 8,

the pressing force is a pressure of the cylinder.

10. The wear prediction device according to any one of claims 1 to 3 and 5,

the wear part is a sliding plate of the pantograph,

the object is a cable to be erected,

the function is to collect current to the vehicle,

11. A wear prediction method comprising:

a prediction step of acquiring a predicted value of a pressing force applied to the wearing member by the mechanism of the subject vehicle based on operation schedule data of the subject vehicle, and predicting a future wear state of the wearing member provided in the subject vehicle based on the predicted value of the pressing force and the estimation model.

12. A recording medium storing a computer program for causing a computer to execute:

a prediction step of acquiring a predicted value of a pressing force applied to the wear member by the mechanism of the target vehicle based on operation plan data of the target vehicle, and predicting a future wear state of the wear member provided in the target vehicle based on the predicted value of the pressing force and the estimation model.