Method, device, and storage medium for evaluating performance of train braking system

Abstract

The application provides a train braking system performance evaluation method, device and storage medium, the method comprising: obtaining pressure data of a brake cylinder; the pressure data is periodically collected from the time when a braking instruction is issued; calculating evaluation values of each evaluation feature according to the pressure data; forming an evaluation data set according to the evaluation values; evaluating the performance of the train braking system according to the evaluation data set and a pre-trained evaluation model; the evaluation model maps the evaluation data set to a high-dimensional feature space, and evaluates according to the relationship between the mapped data set and the hyperplane in the high-dimensional feature space. The method provided by the application evaluates the performance of the train braking system based on the pressure data of the brake cylinder, realizes real-time monitoring of the train braking system, and improves the reliability and safety of the rail transit system.

Description

Technical Field

[0001] This application relates to the field of rail transit technology, and in particular to a method, device, and storage medium for evaluating the performance of a train braking system. Background Technology

[0002] With the continuous development of cities, rail transit systems are playing an increasingly important role in modern society as an efficient and environmentally friendly mode of public transportation.

[0003] The train braking system plays a central role in rail transit systems, ensuring the safe transport of passengers. Its proper functioning not only affects passenger safety but also directly impacts the efficiency and economic benefits of the rail transit system.

[0004] However, factors such as long-term operation, unpredictable changes in operating conditions, and wear and tear of components can cause the performance of the train braking system to gradually degrade, potentially leading to a decrease in braking performance or even failure, thus posing a potential threat to the safety and operation of the rail transit system. Summary of the Invention

[0005] To address one of the aforementioned technical deficiencies, this application provides a method, device, and storage medium for evaluating the performance of a train braking system.

[0006] The first aspect of this application provides a method for evaluating the performance of a train braking system, the method comprising:

[0007] Acquire pressure data from the brake cylinder; the pressure data is collected periodically starting from the moment the braking command is issued.

[0008] Based on the stress data, calculate the assessment value for each assessment feature;

[0009] An evaluation dataset is formed based on the evaluation values;

[0010] The performance of the train braking system is evaluated based on the evaluation dataset and the pre-trained evaluation model.

[0011] The evaluation model evaluates the dataset by mapping it to a high-dimensional feature space and then evaluating it based on the relationship between the mapped dataset and the hyperplane in the high-dimensional feature space.

[0012] Optionally, the evaluation characteristics include: the time it takes for the pressure to start rising, the time it takes for the pressure to stabilize, the mean of the stable pressure, the root mean square of the stable pressure, the maximum value of the stable pressure, the minimum value of the stable pressure, and the time it takes for the pressure to release.

[0013] Based on the stress data, calculate the assessment values ​​for each assessment characteristic, including:

[0014] The estimated value for the duration of pressure rise is the time between the acquisition time of the first data and the time when the braking command is issued, where the first data is the data in the pressure data where the pressure begins to rise;

[0015] The evaluation value for the time it takes for the pressure to reach stability is the time between the earliest acquisition time of all second data and the acquisition time of the first data. The second data refers to the data in the pressure data that is in the pressure stability phase.

[0016] The mean of the steady-state pressure is calculated as the mean of all second data.

[0017] The evaluation value for calculating the mean square error of steady-state pressure is the mean square error of all second data.

[0018] The evaluation value for calculating the maximum steady-state pressure is the maximum value of all second data.

[0019] The evaluation value for calculating the minimum steady-state pressure is the minimum of all second data.

[0020] The estimated value for the pressure release time is the time between the acquisition time of the third data and the time when the braking command is withdrawn. The third data is the pressure data that reaches the target value after the braking command is withdrawn.

[0021] Optionally, an evaluation dataset is formed based on the evaluation values, including:

[0022] Determine the standard value r of the evaluation feature that has a standard value. i , where i is the identifier of the evaluation feature that has a standard value;

[0023] Calculate the evaluation value d after difference processing i =|k i -r i |or d i =|k i -r i | 2 , where k i The evaluation value of evaluation feature i, which has a standard value;

[0024] The evaluation dataset is formed by combining the evaluation values ​​after difference processing with the evaluation values ​​of evaluation features that do not have standard values.

[0025] Optionally, the evaluation dataset is formed by combining the differenced evaluation values ​​and the evaluation values ​​of evaluation features that do not have standard values, including:

[0026] Determine the mean and standard deviation of the final evaluation values ​​for each evaluation feature. The final evaluation value of the evaluation feature with a standard value is the evaluation value after difference processing, and the final evaluation value of the evaluation feature without a standard value is the evaluation value of the evaluation feature without a standard value.

[0027] Calculate the normalized value of each evaluation feature. Where j is the evaluation feature identifier, z j To evaluate the final evaluation value of feature j, μ j To evaluate the mean of feature j, σ j To evaluate the mean square error of feature j;

[0028] The normalized values ​​of each evaluation feature are used to form the evaluation dataset.

[0029] Optionally, before evaluating the performance of the train braking system based on the evaluation dataset and a pre-trained evaluation model, the following steps are also included:

[0030] Obtain the normal sample dataset x;

[0031] Through mapping function Map x to the high-order feature space Π;

[0032] Determine the hyperplane in Π Where w is the normal vector of the hyperplane, and r is the bias term of the hyperplane;

[0033] Decision function that forms the evaluation model and the distance function from x to the hyperplane

[0034] Optionally, the performance of the train braking system is evaluated based on the evaluation dataset and a pre-trained evaluation model, including:

[0035] The value of the decision function f(y) corresponding to the evaluation dataset y is determined by using a pre-trained evaluation model;

[0036] If the value of f(y) is greater than 0, then the qualitative assessment result of the train braking system performance is determined to be normal.

[0037] If the value of f(y) is not greater than 0, then the qualitative evaluation result of the train braking system performance is determined to be abnormal.

[0038] Optionally, after determining that the qualitative assessment result of the train braking system performance is abnormal, the following steps are also included:

[0039] The value of the distance function d(y) corresponding to y is determined by a pre-trained evaluation model.

[0040] Optionally, after determining the value of the distance function d(y) corresponding to y, the process also includes:

[0041] If the absolute value of d(y) is less than the minimum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be mild degradation.

[0042] If the absolute value of d(y) is not less than the minimum quantitative threshold but less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be severe degradation.

[0043] If the absolute value of d(y) is not less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be a braking system failure.

[0044] A second aspect of this application provides an electronic device, comprising:

[0045] Memory;

[0046] Processor; and

[0047] Computer programs;

[0048] The computer program is stored in the memory and configured to be executed by the processor to implement the method described in the first aspect above.

[0049] A third aspect of this application provides a computer-readable storage medium having a computer program stored thereon; the computer program is executed by a processor to implement the method described in the first aspect above.

[0050] This application provides a method, device, and storage medium for evaluating the performance of a train braking system. The method includes: acquiring pressure data from the brake cylinder; the pressure data is collected periodically starting from the moment the braking command is issued; calculating evaluation values ​​for each evaluation feature based on the pressure data; forming an evaluation dataset based on the evaluation values; evaluating the performance of the train braking system based on the evaluation dataset and a pre-trained evaluation model; the evaluation model performs the evaluation by mapping the evaluation dataset to a high-level feature space and based on the relationship between the mapped dataset and the hyperplane in the high-level feature space. The method provided in this application evaluates the performance of the train braking system based on the pressure data from the brake cylinder, enabling real-time monitoring of the train braking system and improving the reliability and safety of the rail transit system. Attached Figure Description

[0051] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0052] Figure 1 A flowchart illustrating a method for evaluating the performance of a train braking system provided in this application embodiment;

[0053] Figure 2 A schematic diagram of brake cylinder pressure data curve provided in an embodiment of this application;

[0054] Figure 3A schematic diagram of a hyperplane provided for an embodiment of this application;

[0055] Figure 4 This is a schematic diagram of anomaly scores provided in an embodiment of this application. Detailed Implementation

[0056] To make the technical solutions and advantages of the embodiments of this application clearer, the exemplary embodiments of this application will be described in further detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not an exhaustive list of all embodiments. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other.

[0057] In the process of developing this application, the inventors discovered that factors such as long-term operation, unpredictable changes in operating conditions, and wear and tear of parts can cause the performance of the train braking system to gradually degrade, which may lead to a decrease in braking performance or even failure, thereby posing a potential threat to the safety and operation of the rail transit system.

[0058] To address the aforementioned issues, this application provides a method, device, and storage medium for evaluating the performance of a train braking system. The method includes: acquiring pressure data from the brake cylinder; the pressure data is collected periodically starting from the moment the braking command is issued; calculating evaluation values ​​for each evaluation feature based on the pressure data; forming an evaluation dataset based on the evaluation values; evaluating the performance of the train braking system based on the evaluation dataset and a pre-trained evaluation model; the evaluation model maps the evaluation dataset to a high-level feature space and performs evaluation based on the relationship between the mapped dataset and the hyperplane in the high-level feature space. The method provided in this application evaluates the performance of the train braking system based on brake cylinder pressure data, enabling real-time monitoring of the train braking system and improving the reliability and safety of the rail transit system.

[0059] See Figure 1 The implementation process of the train braking system performance evaluation method provided in this embodiment is as follows:

[0060] 101, Obtain the pressure data of the brake cylinder.

[0061] The pressure data is collected periodically starting from the moment the braking command is issued.

[0062] In practical implementation, a high-precision pressure sensor can be placed at a key location on the train brake cylinder. This high-precision pressure sensor periodically collects pressure data from the brake cylinder starting from the moment the braking command is issued. The pressure data of the brake cylinder collected by this high-precision pressure sensor is then obtained in step 101.

[0063] In addition, the data acquisition period is preset, such as 0.1 seconds. That is, starting from the moment the braking command is issued, the high-precision pressure sensor records the brake cylinder pressure data once every 0.1 seconds.

[0064] The moment the braking command is issued marks the start of the braking process. From this point onward, high-precision pressure sensors continuously record changes in brake cylinder pressure over time, ensuring complete data recording of the entire braking process, including the gradual increase in pressure until the moment the braking command is released. This ensures detailed data is obtained at different stages of the braking system's operation, from the initial application of the brakes to their final release.

[0065] After obtaining the pressure data, a brake cylinder pressure data curve can be generated based on the acquisition time, such as... Figure 2 As shown, the horizontal axis represents time, and the vertical axis represents pressure value.

[0066] 102. Based on the stress data, calculate the evaluation value of each evaluation feature.

[0067] The evaluation features include, but are not limited to, one or more of the following: the time it takes for the pressure to start rising, the time it takes for the pressure to stabilize, the mean of the stable pressure, the root mean square of the stable pressure, the maximum value of the stable pressure, the minimum value of the stable pressure, and the time it takes for the pressure to release.

[0068] Based on the above evaluation characteristics, the implementation process of step 102 is as follows:

[0069] The estimated value for the duration of pressure rise is the time between the acquisition of the first data and the issuance of the braking command, where the first data is the data in the pressure data where pressure begins to rise.

[0070] The evaluation value for the time it takes for the pressure to stabilize is the time between the earliest acquisition time of all the second data and the acquisition time of the first data. The second data refers to the data in the pressure data that is in the pressure stabilization phase.

[0071] The mean of the steady-state pressure is calculated as the mean of all the second data.

[0072] The estimated value for calculating the mean square error of steady-state pressure is the mean square error of all second data.

[0073] The evaluation value for calculating the maximum steady-state pressure is the maximum value of all second data.

[0074] The estimated value of the steady-state pressure minimum is the minimum value of all second data.

[0075] The estimated value for the pressure release time is the time between the acquisition time of the third data and the time when the braking command is withdrawn. The third data is the pressure data that reaches the target value after the braking command is withdrawn.

[0076] In practical implementation, the pressure data obtained in step 101 can be used to form a brake cylinder pressure data curve. The brake cylinder pressure data curve can then be divided into a pressure rise stage, a pressure stabilization stage, and a pressure release stage. An evaluation value can be calculated for the pressure data of each stage.

[0077] For example, for pressure data during the pressure rise phase, calculate the time it takes for the pressure to start rising and the time it takes for the pressure to stabilize.

[0078] During the rising phase of the brake cylinder pressure data curve, the pressure data at which the brake cylinder pressure begins to rise is determined; this pressure data is the first data point. Simultaneously, the acquisition time of the first data point (i.e., the horizontal axis value of the first data point) is determined from the brake cylinder pressure data curve. The duration between the acquisition time of the first data point and the time the braking command is issued is calculated; this duration is the assessed value of the pressure start-rising duration. In other words, the pressure start-rising duration is the time required for the brake cylinder pressure to begin rising from the time the command is issued.

[0079] During the pressure rise phase of the brake cylinder pressure data curve, the pressure data at which the brake cylinder pressure reaches a stable state is determined. This data is also the first value of the pressure stabilization phase. If all pressure data in the stabilization phase are considered as the second data, then the pressure data at which the brake cylinder pressure reaches a stable state is the earliest second data acquired. The duration between the earliest acquisition time of all second data and the acquisition time of the first data is calculated; this duration is the pressure stabilization time. In other words, the pressure stabilization time is the time required for the brake cylinder pressure to rise from its initial rise to a stable state.

[0080] For pressure data during the stable pressure phase, calculate the mean stable pressure, root mean square of stable pressure, maximum stable pressure, and minimum stable pressure.

[0081] If all pressure data during the stable pressure phase are taken as the second data, then the mean of all the second data is the evaluation value of the stable pressure mean, the root mean square of all the second data is the evaluation value of the stable pressure root mean square, the maximum of all the second data is the evaluation value of the stable pressure maximum, and the minimum of all the second data is the evaluation value of the stable pressure minimum.

[0082] In other words, the mean stable pressure is the average value of the brake cylinder pressure during the steady phase, the root mean square of the stable pressure is the root mean square of the brake cylinder pressure during the steady phase, and the maximum and minimum stable pressures are the maximum and minimum values ​​of the brake cylinder pressure during the steady phase.

[0083] Calculate the pressure release duration based on the pressure data during the pressure release phase.

[0084] During the pressure release phase of the brake cylinder pressure data curve, the pressure data at which the brake cylinder pressure reaches the preset target after the brake command is cancelled is determined; this pressure data is the third data. Simultaneously, the acquisition time of the third data (i.e., the horizontal axis value of the third data) is determined through the brake cylinder pressure data curve, and the duration between the acquisition time of the third data and the time of brake command cancellation is calculated; this duration is the evaluated value of the pressure release time. In other words, the pressure release time is the time required from the cancellation of the brake command to the release of the brake cylinder pressure to the target value.

[0085] 103. An evaluation dataset is formed based on the evaluation values.

[0086] In specific implementation, step 103 can either form an evaluation dataset by combining all evaluation values, or process the evaluation values ​​based on whether the evaluation features have standard values ​​before forming the evaluation dataset.

[0087] The process of processing the evaluation values ​​based on whether the evaluation features have standard values, and then forming the evaluation dataset, is as follows:

[0088] 201, Determine the standard value r of the evaluation characteristic that has a standard value. i .

[0089] Where i is the identifier of the evaluation feature that has a standard value.

[0090] The standard values ​​for evaluation features that have standard values ​​are recorded in the relevant document specifications. Here, we can directly obtain the standard values ​​recorded in the specifications.

[0091] 202, calculate the difference d i =|k i -r i |or d i =|k i -r i | 2 , will d i As the evaluation value after difference processing.

[0092] Where, k i The evaluation value is the evaluation feature i for which a standard value exists.

[0093] For evaluating feature i, the difference d i The evaluation value k represents the evaluation feature i. i Its standard value r i The degree of deviation between them.

[0094] 203. The evaluation dataset is formed by combining the evaluation values ​​after difference processing and the evaluation values ​​of evaluation features that do not have standard values.

[0095] Step 203 can directly combine the differenced evaluation values ​​and the evaluation values ​​of evaluation features without standard values ​​to form the evaluation dataset y. Alternatively, the differenced evaluation values ​​and the evaluation values ​​of evaluation features without standard values ​​can be normalized to form the evaluation dataset y.

[0096] The following describes the specific implementation details of forming the evaluation dataset y by normalizing the evaluation values ​​after difference processing and the evaluation values ​​of evaluation features that do not have standard values:

[0097] 1. Determine the mean and standard deviation of the final evaluation values ​​for each evaluation feature.

[0098] Among them, the final evaluation value of the evaluation feature with a standard value is the evaluation value after difference processing, and the final evaluation value of the evaluation feature without a standard value is the evaluation value of the evaluation feature without a standard value.

[0099] 2. Calculate the normalized value of each evaluation feature. Where j is the evaluation feature identifier, z j To evaluate the final evaluation value of feature j, μ j To evaluate the mean of feature j, σ j To evaluate the mean squared error of feature j.

[0100] 3. The normalized values ​​of each evaluation feature are used to form the evaluation dataset y.

[0101] 104. Evaluate the performance of the train braking system based on the evaluation dataset and the pre-trained evaluation model.

[0102] The evaluation model maps the evaluation dataset to a high-order feature space and evaluates the dataset based on the relationship between the mapped dataset and the hyperplane in the high-order feature space. Furthermore, the evaluation model is trained using OCSVM (One-Class Support Vector Machine).

[0103] OCSVM can detect anomalous data in an unsupervised manner. By training OCSVM with only normal samples, the resulting evaluation model can find potential anomalous data (including degradation data and fault data) in the pressure data of the brake cylinder.

[0104] In OCSVM, all data points are considered normal samples, and the origin is considered the only outlier. Its main goal is to separate normal samples from outliers by constructing a hyperplane.

[0105] Therefore, the evaluation model will be trained before performing step 104. The training process is as follows:

[0106] 301, Get the normal sample dataset x.

[0107] For example, obtaining a normal sample dataset Where u is the data identifier in the normal sample dataset, N is the total number of data in the normal sample dataset, and x u ∈R n R n To evaluate the feature space, the evaluation features are the same as those in step 102, that is, the evaluation features include, but are not limited to, one or more of the following: the time it takes for the pressure to start rising, the time it takes for the pressure to stabilize, the mean of the stable pressure, the root mean square of the stable pressure, the maximum value of the stable pressure, the minimum value of the stable pressure, and the time it takes for the pressure to release.

[0108] 302, via mapping function Map x to the high-order feature space Π.

[0109] 303, Determine the hyperplane π in Π:

[0110] Where w is the normal vector of the hyperplane, and r is the bias term of the hyperplane.

[0111] This hyperplane π maximizes the distance between the origin (the mapping in the high-dimensional feature space) and normal samples, such as... Figure 3 As shown.

[0112] 304, the decision function that forms the evaluation model and the distance function from x to the hyperplane

[0113] The optimization problem of OCSVM can be constructed as follows:

[0114]

[0115]

[0116] Here, v is a pre-set regularization parameter that controls the maximization of the distance between the hyperplane and the origin and the number of data points allowed to cross the hyperplane.

[0117] Since the mapping function is usually implicit, the above optimization problem is typically solved using its dual form:

[0118]

[0119]

[0120] Where, α u As the dual variable, K(x) u x a ) is the kernel function, such as K(x) u xa ) is the Gaussian kernel, i.e. σ is the width of the Gaussian kernel.

[0121] By solving for the objective function described above, the decision function can be obtained. The distance function from its data points to the hyperplane

[0122] The OCSVM model parameters (such as kernel function and penalty parameters) are trained using sample data to learn the characteristics of normal braking system performance, thereby constructing an evaluation model.

[0123] After training and obtaining the evaluation model, its performance can be assessed, and the best-performing model can be selected. The performance of the evaluation model can be evaluated using various metrics, such as accuracy, recall, F1 score, and ROC curve plotting. Based on the evaluation results, the most suitable evaluation model is selected to provide optimal performance for real-time monitoring of train braking system performance.

[0124] After obtaining the trained evaluation model, step 104 will perform a qualitative evaluation of the train braking system performance based on the evaluation dataset and the pre-trained evaluation model.

[0125] The qualitative assessment process is as follows:

[0126] A. Determine the value of the decision function f(y) corresponding to the evaluation dataset y using a pre-trained evaluation model.

[0127] B. If the value of f(y) is greater than 0, then the qualitative assessment result of the train braking system performance is determined to be normal.

[0128] If the value of f(y) is not greater than 0, then the qualitative evaluation result of the train braking system performance is determined to be abnormal.

[0129] The purpose of qualitative assessment is to determine whether a pressure data point belongs to normal braking data. During the qualitative assessment process, the pressure data obtained in step 101 is evaluated using a trained assessment model. The results can be categorized into the following two cases:

[0130] Normal: that is

[0131] If the evaluation model determines it to be normal, it means that the performance of the train braking system is consistent with expectations and there are no obvious abnormalities.

[0132] Abnormal: i.e.

[0133] If the evaluation model determines that it is abnormal, it indicates that there is a problem with the performance of the train braking system, and further quantitative analysis is needed to determine the degree of abnormality.

[0134] It should be noted that, regardless of whether it is f(x) or f(y), x and y are independent variables, representing the input pressure data. That is, when training the evaluation model, the input pressure data is x, and the decision function is f(x). When evaluating the performance of the train braking system based on the trained evaluation model, the input pressure data is y, and the decision function is f(y).

[0135] After conducting a qualitative assessment, a quantitative assessment can be further conducted for cases where the qualitative assessment results are abnormal.

[0136] The purpose of quantitative assessment is to determine the degree of anomaly (degradation) in stress data. Anomaly scores can be obtained using the distance function d(y) of a trained assessment model.

[0137] The quantitative evaluation process is as follows: using a pre-trained evaluation model, the value of the distance function d(y) corresponding to y is determined. The absolute value of this distance function d(y) (i.e., |d(y)|) is the anomaly score.

[0138] In practical implementation, for stress data whose qualitative assessment results are abnormal, the value of the distance function d(y) corresponding to the trained assessment model can be used as the abnormality score.

[0139] The higher the anomaly score, the more abnormal the data.

[0140] The formula for calculating the outlier score is: score(y)=|d(y)|.

[0141] like Figure 4 As shown, both data point 1 and data point 2 are on the outlier side of the hyperplane. Since both are outliers, and since data point 1 is farther from the decision hyperplane than data point 2, the outlier score score(data point 1) > score(data point 2), indicating that data point 1 is more outlier than data point 2.

[0142] It should be noted that, regardless of whether it is d(x) or d(y), x and y are independent variables, representing the input pressure data. That is, when training the evaluation model, the input pressure data is x, and the distance function is d(x). When evaluating the performance of the train braking system based on the trained evaluation model, the input pressure data is y, and the distance function is d(y).

[0143] After obtaining outliers, the outlier scores can be converted into quantitative measures, i.e.:

[0144] If the absolute value of d(y) is less than the minimum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be mild degradation, that is, mild abnormality.

[0145] If the absolute value of d(y) is not less than the minimum quantitative threshold but less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be severe degradation, i.e., severe abnormality.

[0146] If the absolute value of d(y) is not less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be a braking system failure, that is, a braking system failure.

[0147] The minimum and maximum quantitative thresholds can be determined based on actual needs and performance evaluation indicators.

[0148] The train braking system performance evaluation method provided in this embodiment is an OCSVM method for evaluating and monitoring the performance of train braking systems. By analyzing braking system sensor data, the performance of train braking systems can be qualitatively and quantitatively evaluated, which helps to monitor train braking systems in real time and improve the reliability and safety of rail transit systems.

[0149] The train braking system performance evaluation method provided in this embodiment offers a reliable solution for rail transit systems, enabling timely detection of early signs of degradation in train braking systems, thereby ensuring the reliability and safety of the braking system.

[0150] This embodiment provides a method for evaluating the performance of a train braking system. The method involves acquiring pressure data from the brake cylinders, which is collected periodically starting from the moment the braking command is issued. Based on the pressure data, evaluation values ​​for various evaluation features are calculated. An evaluation dataset is formed based on these values. The performance of the train braking system is then evaluated using the evaluation dataset and a pre-trained evaluation model. The evaluation model maps the evaluation dataset to a high-level feature space and evaluates the system based on the relationship between the mapped dataset and the hyperplane in the high-level feature space. This method, based on brake cylinder pressure data, evaluates the performance of the train braking system, enabling real-time monitoring and improving the reliability and safety of rail transit systems.

[0151] Based on the same inventive concept as the train braking system performance evaluation method, this embodiment provides an electronic device, which includes: a memory, a processor, and a computer program.

[0152] The computer program is stored in memory and configured to be executed by a processor to implement the above-mentioned train braking system performance evaluation method.

[0153] Specifically,

[0154] Acquire pressure data from the brake cylinders. The pressure data is collected periodically starting from the moment the braking command is issued.

[0155] Based on the stress data, calculate the evaluation value for each evaluation feature.

[0156] An evaluation dataset is formed based on the evaluation values.

[0157] The performance of the train braking system is evaluated based on the evaluation dataset and the pre-trained evaluation model.

[0158] The evaluation model evaluates the dataset by mapping it to a high-dimensional feature space and then evaluating it based on the relationship between the mapped dataset and the hyperplane in the high-dimensional feature space.

[0159] Optionally, the evaluation characteristics include: the time it takes for the pressure to start rising, the time it takes for the pressure to stabilize, the mean of the stable pressure, the root mean square of the stable pressure, the maximum value of the stable pressure, the minimum value of the stable pressure, and the time it takes for the pressure to release.

[0160] Based on the stress data, calculate the assessment values ​​for each assessment characteristic, including:

[0161] The estimated value for the duration of pressure rise is the time between the acquisition of the first data and the issuance of the braking command, where the first data is the data in the pressure data where pressure begins to rise.

[0162] The evaluation value for the time it takes for the pressure to stabilize is the time between the earliest acquisition time of all the second data and the acquisition time of the first data. The second data refers to the data in the pressure data that is in the pressure stabilization phase.

[0163] The mean of the steady-state pressure is calculated as the mean of all the second data.

[0164] The estimated value for calculating the mean square error of steady-state pressure is the mean square error of all second data.

[0165] The evaluation value for calculating the maximum steady-state pressure is the maximum value of all second data.

[0166] The estimated value of the steady-state pressure minimum is the minimum value of all second data.

[0167] The estimated value for the pressure release time is the time between the acquisition time of the third data and the time when the braking command is withdrawn. The third data is the pressure data that reaches the target value after the braking command is withdrawn.

[0168] Optionally, an evaluation dataset is formed based on the evaluation values, including:

[0169] Determine the standard value r of the evaluation feature that has a standard value. i , where i is the identifier of the evaluation feature with a standard value.

[0170] Calculate the evaluation value d after difference processing i =|ki -r i |or d i =|k i -r i | 2 , where k i The evaluation value is the evaluation feature i for which a standard value exists.

[0171] The evaluation dataset is formed by combining the evaluation values ​​after difference processing with the evaluation values ​​of evaluation features that do not have standard values.

[0172] Optionally, the evaluation dataset is formed by combining the differenced evaluation values ​​and the evaluation values ​​of evaluation features that do not have standard values, including:

[0173] Determine the mean and standard deviation of the final evaluation values ​​for each evaluation feature. The final evaluation value of the evaluation feature with a standard value is the evaluation value after difference processing, and the final evaluation value of the evaluation feature without a standard value is the evaluation value of the evaluation feature without a standard value.

[0174] Calculate the normalized value of each evaluation feature. Where j is the evaluation feature identifier, z j To evaluate the final evaluation value of feature j, μ j To evaluate the mean of feature j, σ j To evaluate the mean squared error of feature j.

[0175] The normalized values ​​of each evaluation feature are used to form the evaluation dataset.

[0176] Optionally, before evaluating the performance of the train braking system based on the evaluation dataset and a pre-trained evaluation model, the following steps are also included:

[0177] Obtain the normal sample dataset x.

[0178] Through mapping function Map x to the high-order feature space Π.

[0179] Determine the hyperplane in Π Where w is the normal vector of the hyperplane, and r is the bias term of the hyperplane.

[0180] Decision function that forms the evaluation model and the distance function from x to the hyperplane

[0181] Optionally, the performance of the train braking system is evaluated based on the evaluation dataset and a pre-trained evaluation model, including:

[0182] The value of the decision function f(y) corresponding to the evaluation dataset y is determined by using a pre-trained evaluation model.

[0183] If the value of f(y) is greater than 0, then the qualitative assessment result of the train braking system performance is determined to be normal.

[0184] If the value of f(y) is not greater than 0, then the qualitative evaluation result of the train braking system performance is determined to be abnormal.

[0185] Optionally, after determining that the qualitative assessment result of the train braking system performance is abnormal, the following steps are also included:

[0186] The value of the distance function d(y) corresponding to y is determined by a pre-trained evaluation model.

[0187] Optionally, after determining the value of the distance function d(y) corresponding to y, the process also includes:

[0188] If the absolute value of d(y) is less than the minimum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be mild degradation.

[0189] If the absolute value of d(y) is not less than the minimum quantitative threshold but less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be severe degradation.

[0190] If the absolute value of d(y) is not less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be a braking system failure.

[0191] The electronic device provided in this embodiment has a computer program executed by a processor to evaluate the performance of the train braking system based on the pressure data of the brake cylinder, thereby realizing real-time monitoring of the train braking system and improving the reliability and safety of the rail transit system.

[0192] Based on the same inventive concept as the train braking system performance evaluation method, this embodiment provides a computer-readable storage medium on which a computer program is stored. The computer program is executed by a processor to implement the aforementioned train braking system performance evaluation method.

[0193] Specifically,

[0194] Acquire pressure data from the brake cylinders. The pressure data is collected periodically starting from the moment the braking command is issued.

[0195] Based on the stress data, calculate the evaluation value for each evaluation feature.

[0196] An evaluation dataset is formed based on the evaluation values.

[0197] The performance of the train braking system is evaluated based on the evaluation dataset and the pre-trained evaluation model.

[0198] The evaluation model evaluates the dataset by mapping it to a high-dimensional feature space and then evaluating it based on the relationship between the mapped dataset and the hyperplane in the high-dimensional feature space.

[0199] Optionally, the evaluation characteristics include: the time it takes for the pressure to start rising, the time it takes for the pressure to stabilize, the mean of the stable pressure, the root mean square of the stable pressure, the maximum value of the stable pressure, the minimum value of the stable pressure, and the time it takes for the pressure to release.

[0200] Based on the stress data, calculate the assessment values ​​for each assessment characteristic, including:

[0201] The estimated value for the duration of pressure rise is the time between the acquisition of the first data and the issuance of the braking command, where the first data is the data in the pressure data where pressure begins to rise.

[0202] The evaluation value for the time it takes for the pressure to stabilize is the time between the earliest acquisition time of all the second data and the acquisition time of the first data. The second data refers to the data in the pressure data that is in the pressure stabilization phase.

[0203] The mean of the steady-state pressure is calculated as the mean of all the second data.

[0204] The estimated value for calculating the mean square error of steady-state pressure is the mean square error of all second data.

[0205] The evaluation value for calculating the maximum steady-state pressure is the maximum value of all second data.

[0206] The estimated value of the steady-state pressure minimum is the minimum value of all second data.

[0207] The estimated value for the pressure release time is the time between the acquisition time of the third data and the time when the braking command is withdrawn. The third data is the pressure data that reaches the target value after the braking command is withdrawn.

[0208] Optionally, an evaluation dataset is formed based on the evaluation values, including:

[0209] Determine the standard value r of the evaluation feature that has a standard value. i , where i is the identifier of the evaluation feature with a standard value.

[0210] Calculate the evaluation value d after difference processing i =|k i -r i |or d i =|k i -r i | 2 , where k i The evaluation value is the evaluation feature i for which a standard value exists.

[0211] The evaluation dataset is formed by combining the evaluation values ​​after difference processing with the evaluation values ​​of evaluation features that do not have standard values.

[0212] Optionally, the evaluation dataset is formed by combining the differenced evaluation values ​​and the evaluation values ​​of evaluation features that do not have standard values, including:

[0213] Determine the mean and standard deviation of the final evaluation values ​​for each evaluation feature. The final evaluation value of the evaluation feature with a standard value is the evaluation value after difference processing, and the final evaluation value of the evaluation feature without a standard value is the evaluation value of the evaluation feature without a standard value.

[0214] Calculate the normalized value of each evaluation feature. Where j is the evaluation feature identifier, z j To evaluate the final evaluation value of feature j, μ j To evaluate the mean of feature j, σ j To evaluate the mean squared error of feature j.

[0215] The normalized values ​​of each evaluation feature are used to form the evaluation dataset.

[0216] Optionally, before evaluating the performance of the train braking system based on the evaluation dataset and a pre-trained evaluation model, the following steps are also included:

[0217] Obtain the normal sample dataset x.

[0218] Through mapping function Map x to the high-order feature space Π.

[0219] Determine the hyperplane in Π Where w is the normal vector of the hyperplane, and r is the bias term of the hyperplane.

[0220] Decision function that forms the evaluation model and the distance function from x to the hyperplane

[0221] Optionally, the performance of the train braking system is evaluated based on the evaluation dataset and a pre-trained evaluation model, including:

[0222] The value of the decision function f(y) corresponding to the evaluation dataset y is determined by using a pre-trained evaluation model.

[0223] If the value of f(y) is greater than 0, then the qualitative assessment result of the train braking system performance is determined to be normal.

[0224] If the value of f(y) is not greater than 0, then the qualitative evaluation result of the train braking system performance is determined to be abnormal.

[0225] Optionally, after determining that the qualitative assessment result of the train braking system performance is abnormal, the following steps are also included:

[0226] The value of the distance function d(y) corresponding to y is determined by a pre-trained evaluation model.

[0227] Optionally, after determining the value of the distance function d(y) corresponding to y, the process also includes:

[0228] If the absolute value of d(y) is less than the minimum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be mild degradation.

[0229] If the absolute value of d(y) is not less than the minimum quantitative threshold but less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be severe degradation.

[0230] If the absolute value of d(y) is not less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be a braking system failure.

[0231] The computer-readable storage medium provided in this embodiment has a computer program thereon that is executed by a processor to evaluate the performance of the train braking system based on the pressure data of the brake cylinder, thereby realizing real-time monitoring of the train braking system and improving the reliability and safety of the rail transit system.

[0232] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0233] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0234] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0235] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0236] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0237] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0238] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for evaluating the performance of a train braking system, characterized in that, The method includes: Acquire pressure data from the brake cylinder; the pressure data is collected periodically starting from the moment the braking command is issued; The evaluation features include: duration of pressure rise, duration of pressure stabilization, mean stable pressure, root mean square deviation of stable pressure, maximum stable pressure, minimum stable pressure, and duration of pressure release. Based on the pressure data, the evaluation values ​​for each feature are calculated, including: the evaluation value for the duration of pressure rise is the time between the acquisition time of the first data and the time when the braking command is issued, where the first data is the data where pressure begins to rise; the evaluation value for the duration of pressure stabilization is the time between the earliest acquisition time of all second data and the acquisition time of the first data, where the second data is the data in the pressure stabilization phase; the evaluation value for the mean stable pressure is the mean of all second data; the evaluation value for the root mean square deviation of stable pressure is the root mean square deviation of all second data; the evaluation value for the maximum stable pressure is the maximum value of all second data; the evaluation value for the minimum stable pressure is the minimum value of all second data; and the evaluation value for the duration of pressure release is the time between the acquisition time of the third data and the time when the braking command is withdrawn, where the third data is the data in the pressure data that reaches the target value after the braking command is withdrawn. An evaluation dataset is formed based on the evaluation values, including: standard values ​​for determining the evaluation features that have standard values. ,in, Identify the evaluation characteristics that have a standard value; calculate the evaluation value after difference processing. or ,in, Evaluation characteristics with existing standard values The evaluation values ​​are then used to form an evaluation dataset, which includes the evaluation values ​​after difference processing and the evaluation values ​​of evaluation features that do not have standard values. Obtain normal sample dataset ; Through mapping function Will Mapping to high-bit feature space ; Sure hyperplane in ;in, Let be the normal vector of the hyperplane. For the bias term of the hyperplane; Decision function that forms the evaluation model ,as well as Distance function to the hyperplane ; The performance of the train braking system is evaluated based on the evaluation dataset and the pre-trained evaluation model. The evaluation model performs evaluation by mapping the evaluation dataset to a high-dimensional feature space and then evaluating the dataset based on the relationship between the mapped dataset and the hyperplane in the high-dimensional feature space.

2. The method according to claim 1, characterized in that, The evaluation dataset is formed by combining the evaluation values ​​after difference processing and the evaluation values ​​of evaluation features that do not have standard values, including: Determine the mean and standard deviation of the final evaluation values ​​for each evaluation feature. The final evaluation value of the evaluation feature with a standard value is the evaluation value after difference processing, and the final evaluation value of the evaluation feature without a standard value is the evaluation value of the evaluation feature without a standard value. Calculate the normalized value of each evaluation feature. ,in, To evaluate feature identifiers, To evaluate features The final evaluation value, To evaluate features The mean, To evaluate features The mean squared error; The normalized values ​​of each evaluation feature are used to form the evaluation dataset.

3. The method according to claim 2, characterized in that, The evaluation of train braking system performance based on the evaluation dataset and a pre-trained evaluation model includes: The evaluation dataset is determined using a pre-trained evaluation model. Corresponding decision function The value; like If the value is greater than 0, the qualitative assessment result of the train braking system performance is determined to be normal; like If the value is not greater than 0, then the qualitative assessment result of the train braking system performance is determined to be abnormal.

4. The method according to claim 3, characterized in that, After determining that the qualitative assessment result of the train braking system performance is abnormal, the following steps are also included: Determine using a pre-trained evaluation model Corresponding distance function The value of .

5. The method according to claim 4, characterized in that, The determination Corresponding distance function Following the value, it also includes: like If the absolute value of the value is less than the minimum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be mild degradation. like If the absolute value of the value is not less than the minimum quantitative threshold but less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be severely degraded. like If the absolute value of the value is not less than the maximum quantitative threshold, then the quantitative assessment result of the train braking system performance is determined to be a braking system failure.

6. An electronic device, characterized in that, include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and configured to be executed by the processor to implement the method as described in any one of claims 1-5.

7. A computer-readable storage medium, characterized in that, It stores a computer program thereon; the computer program is executed by a processor to implement the method as described in any one of claims 1-5.

Images