A method for determining the detection performance of a locomotive brake

Abstract

The present invention discloses a method for determining the detection performance of a locomotive brake, the method comprising: using the on-board equipment on the locomotive to collect the dynamic data generated during the brake test in real time, the dynamic data including the air cylinder pressure, pipeline pressure, brake handle position signal and other operation-related parameters; pre-processing the collected dynamic data, and building a test standard curve set in combination with historical test data, the standard curve set being classified and stored according to the locomotive model and the brake model. The present invention constructs a standard test data set based on the test data of a large number of locomotives and brakes of different models, combines the feature sampling algorithm, the least squares fitting algorithm and the multivariate linear regression model, realizes the feature extraction and fitting analysis of the dynamic data, and by comparing the fitting curve with the standard test curve, the system automatically determines the standardization and integrity of the operation, avoiding subjective errors and omissions in human analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of locomotive brake detection, and in particular to a method for determining the detection performance of a locomotive brake. Background Art

[0002] The performance of the locomotive brake is of vital importance to driving safety. In the daily operation of the locomotive depot, in order to ensure the normal performance of the locomotive brake, the "five-step brake" test is currently generally used as the main method for testing the performance of the locomotive brake. During the quality acceptance process after the locomotive is prepared in the depot or the locomotive is overhauled, the test personnel operate the brake handle, combined with the relevant status and air pressure changes displayed on the driver's station brake screen, air cylinder pressure gauge and LKJ display screen, to synchronously observe and judge the performance of the locomotive brake. At the same time, the test personnel need to manually analyze the abnormal data fed back during the test to determine the fault location and cause. After the test, the test personnel also need to summarize the test results through manual records, and the foreman and quality supervision personnel will review and check the operation quality.

[0003] Although this method has been used for many years as a traditional test method, its manual operation and analysis mode has obvious shortcomings and limitations in practical applications. In order to make up for the supervision loopholes of manual recording methods, some locomotive depots have introduced the method of downloading LKJ record data with IC cards in recent years, and completed the analysis of test results through manual retrieval of data. However, this method is still highly dependent on manual judgment, the analysis efficiency is low and prone to errors, and it is difficult to meet the requirements of modern railway maintenance and preparation operations for data and informationization.

[0004] The existing locomotive brake performance detection method has the following problems and disadvantages:

[0005] 1. The entire testing process is completely dependent on the professional level and self-discipline of the operator. Any slight negligence or mistake may lead to quality problems;

[0006] 2. The existing detection method requires test personnel to make judgments by observing the feedback information displayed by the equipment, and human analysis cannot avoid subjective errors and may even miss key data;

[0007] 3. Manual recording and retrieval of raw data involves a large workload. Test personnel need to complete data collection, analysis, recording and reporting step by step. This process is not only time-consuming, but also repetitive and labor-intensive, which can easily lead to operator fatigue.

[0008] 4. The mode of manual operation and manual analysis makes the verification of work quality complicated and difficult. In order to ensure the quality of the test, it is often necessary to verify it through repeated tests by multiple people, which further increases the complexity and labor cost of the test.

[0009] Based on this, a method for determining the performance of a locomotive brake is needed. Summary of the invention

[0010] To achieve the above object, the present invention provides the following solution: a method for determining the detection performance of a locomotive brake, the method comprising:

[0011] The on-board equipment on the locomotive is used to collect dynamic data generated during the brake test in real time, the dynamic data including air cylinder pressure, pipeline pressure, brake handle position signal and other operation-related parameters;

[0012] Preprocess the collected dynamic data, and build a test standard curve set in combination with historical test data. The standard curve set is classified and stored according to locomotive model and brake machine model;

[0013] Generate a characteristic fitting curve of the collected data by using characteristic sampling technology and multi-segment fitting algorithm; compare and analyze the fitting curve with the standard curve to determine whether the change of the working pressure of the brake meets the operating specifications, and the analysis includes global trend state analysis and local characteristic point change analysis;

[0014] According to the operating specifications of the five-step gate test, the completeness of the operating steps and the execution quality of the key steps are automatically identified, including the judgment of the push-pull speed, push-pull amplitude and operation continuity of the brake handle; if an unfinished or abnormal operation is identified, the corresponding abnormal section is marked;

[0015] In the judgment process, the dynamic characteristics of abnormal data and the historical fault case library are combined to use the decision tree model to deduce possible fault causes and output high-probability fault types and location information;

[0016] In response to the detection requirements of different locomotive brakes, the analysis rules and parameter thresholds in the algorithm model are dynamically adjusted to adapt to different operating environments, test conditions and regional specifications.

[0017] Further preferably, the dynamic data collection includes collecting key parameters of the brake test through an IC card recording device, and dumping, querying and retrieving the data in the IC card through a handheld device; the handheld device can selectively copy, delete, modify or wirelessly transmit the IC card data to a ground analysis main station;

[0018] The combined use of IC card recording devices and handheld devices improves the data collection efficiency and flexibility of key parameters of locomotive brake test. The IC card recording device realizes efficient collection and storage of dynamic data of brake test, ensuring data integrity and traceability; the handheld device provides convenient mobile operation capabilities, supports IC card data dump, query and retrieval, and quickly transmits it to the ground analysis main station via wireless mode, providing support for remote analysis and real-time monitoring;

[0019] The IC card recording device and the handheld device complete data exchange through a dedicated hardware interface or wireless communication. The IC card recording device and the handheld device are adapted to the data collection requirements of various brake machine models, and support parameter settings and storage format adjustments based on the requirements of different railway companies, optimize the data interaction process, and ensure efficient transmission and accurate analysis of dynamic data.

[0020] Further preferably, the analysis of the dynamic data is based on a standard test data set and an algorithm model, wherein the standard test data set is generated by collecting pressure change data of different types of locomotives under normal test conditions; in the dynamic data analysis, the preset algorithm model includes a feature sampling algorithm, a least squares fitting algorithm and a multivariate linear regression algorithm, which are used to perform feature extraction, fitting analysis and determine whether the operation is standardized on the collected data, including the following steps:

[0021] (1) Feature sampling algorithm: Extract key feature points of pressure change from dynamic data, including peak value, inflection point and slope change rate. Use sliding window method to calculate the feature value of feature sampling window by the following formula:

[0022]

[0023] in, is the eigenvalue of the feature sampling window, is the pressure point collected, is the mean value within the window, is the window size, the extracted feature values ​​are sorted according to the time series, and the key feature points are screened out to form a preliminary feature set as the input data for fitting analysis;

[0024] (2) Least squares fitting algorithm: Based on the peak value, inflection point and slope change rate of the pressure change of the collected data, the piecewise least squares method is used to fit the pressure change trend curve, and the sum of squares of the fitting error is It is expressed as:

[0025]

[0026] in, Indicates the pressure value of the collected data point, Indicates the corresponding time point, and Respectively The fitted slope and intercept of the segment, is the number of segments; by minimizing Optimize fitting parameters and ,The fitting analysis generates a continuous trend curve based on the characteristic sampling points, and verifies the credibility of the characteristic points through the fitting error, and optimizes the input data of the regression analysis;

[0027] (3) Multiple linear regression algorithm: combining the slope of the fitting curve parameters ,intercept , fitting error, and the push-pull speed and push-pull amplitude of the handle of the dynamic variables of the characteristic sampling points, and construct a multivariate linear regression model. The formula is:

[0028]

[0029] in, Indicates the predicted value of pressure change trend, , , , are the local characteristic values ​​of the handle push-pull speed, push-pull amplitude and pressure change, respectively. is the bias term, , , , is the regression coefficient, The regression analysis comprehensively fits the global trend information of the curve and the local characteristics of the feature points to determine the standardization of the operation, and combines the regression residual analysis to mark the abnormal sections that do not meet the standards.

[0030] The feature sampling algorithm extracts the key feature points of pressure changes through a sliding window to form the basis for analysis input; the least squares fitting algorithm further generates a pressure change trend curve, and reduces the impact of complex dynamic changes through segmented fitting to improve fitting accuracy; the multivariate linear regression algorithm comprehensively fits the curve parameters and the dynamic characteristics of the operation to make a comprehensive judgment on the standardization of the operation.

[0031] Further preferably, the determination of the operation steps includes the integrity determination of the five-step brake test, and the determination rules of the five-step brake test are formulated according to the operating procedures of different types of locomotives, including the handle movement amplitude, the handle movement speed and the timing relationship between each movement;

[0032] The identification of the five-step gate test operation is based on the combination of global trend analysis of pressure changes and local feature sampling. The segmented fitting algorithm is used to determine whether the pressure change of each action segment is consistent with the standard specification. When there are missing steps or actions that do not meet the operation requirements in the five-step gate test, the system will automatically mark the abnormal area and output the abnormal data of the relevant operation for analysis and processing;

[0033] The pressure change trend curve is generated through a segmented fitting algorithm to fully reflect the pressure response characteristics of each operation step. Combined with local feature sampling, the system extracts key feature points such as peak value, inflection point and change slope in the pressure curve to ensure that subtle abnormalities in the operation process can be captured. By comparing and analyzing the pressure changes in each operation segment with the standard specifications, it can accurately identify problems such as missing steps, non-standard operations or lack of action continuity. For detected abnormal operations, the system automatically marks the abnormal area and outputs detailed abnormal data, including abnormal location, pressure deviation value and time information, for operators to quickly review and correct.

[0034] Further preferably, the analysis of the abnormal data is implemented by a fault case decision tree model, the decision tree model is constructed based on historical test data and a fault case library, and the node weights of the decision tree model are Calculated by the following formula:

[0035]

[0036] in, Indicates the fault type The weight value of Indicates the historical occurrence frequency of the corresponding fault type, Represents the total number of all fault types in the decision tree;

[0037] Match the dynamic features of the detected abnormal data with the fault features in the decision tree nodes and calculate the matching confidence of each fault node , the matching confidence value range is 0~1, indicating the abnormal data and fault type The degree of feature matching;

[0038] Combined node weights and matching confidence , calculate the probability of each failure type , the formula is:

[0039] in Fault Type The probability value of the fault is generated by recursively analyzing all nodes in the decision tree model. , and sort them from high to low according to the probability value, output the fault type with the highest probability as the judgment result, and locate the fault in combination with the characteristic information of the fault type. When multiple fault types have close probability values, mark the fault type and prompt manual confirmation.

[0040] Further preferably, the dynamic data is collected by an onboard device integrated on the locomotive, and the onboard device includes but is not limited to BCU, LKJ, 6A, TCMS and CMD;

[0041] The on-board equipment records the air cylinder pressure, pipeline pressure, brake handle position signal and test time during the brake operation in real time, and synchronously stores the collected data in time series, and exports the data for analysis through an IC card recording device or wireless transmission method; the on-board equipment supports adaptation to different models of locomotives to ensure synchronization and completeness of data collection.

[0042] Further preferably, the method is implemented based on the locomotive five-step brake test operation specifications, analyzes the push-pull speed, movement amplitude and operation interval of the handle position during the brake operation, and determines whether the operation steps of the five-step brake test are complete and in compliance with the specifications; verifies whether the test operation complies with the operation standards by performing segmented fitting analysis on the pressure change trend of each action segment; if missed operation or irregular operation is detected in the five-step brake test, the abnormal operation segment is marked and the specific abnormal cause is output; the analysis is based on the standard operation database, and dynamically adjusts the judgment range of the operation parameters in combination with the operation specifications of different railway companies or regions.

[0043] Further preferably, the method adopts an intelligent analysis system, which is divided into a parameter management module, a five-step gate step identification module and an abnormal point analysis module;

[0044] The parameter management module is used to store and configure the operating specifications, locomotive models and environmental conditions involved in the test process, and supports users to dynamically modify parameters and intervene in the analysis process during the test operation;

[0045] The five-step gate step identification module identifies the completeness and accuracy of each step in the test by sampling and fitting the pressure change trend during the test, and marks the steps that do not meet the operating specifications;

[0046] The abnormal item point analysis module combines the preset algorithm model and adopts the method of combining global trend analysis with local feature point extraction to analyze the detection results and generate abnormal pressure areas and possible fault types.

[0047] Further preferably, the entire process of collecting, analyzing and reporting the test data is automatically recorded by an intelligent analysis system, and the intelligent analysis system includes:

[0048] Automatically generate standardized test reports, including the integrity of the operating steps, marking of abnormal operating areas, and detailed pressure change data;

[0049] Automatically sort and retrieve test records by time, operator, and locomotive model, supporting quick location of specific test data;

[0050] Provides stage identification and graphical conversion functions for test data, converting raw data into graphs, tables or other visual forms to intuitively display pressure change trends and abnormal points;

[0051] It supports the storage and export of test reports, which can be used for subsequent analysis and review. It also supports multi-user access rights management to achieve efficient sharing and management of test data.

[0052] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0053] 1. The present invention utilizes locomotive onboard equipment and external acquisition devices to collect dynamic data such as air cylinder pressure, pipeline pressure, brake handle position signal, etc. generated during the locomotive brake test in real time, and transmits the collected test data to the analysis main station or terminal equipment through wireless transmission or data interface, ensuring that data collection is fully automated and accurate, eliminating dependence on manual records.

[0054] 2. The present invention constructs a standard test data set based on test data of a large number of different types of locomotives and brakes, and combines feature sampling algorithm, least squares fitting algorithm and multivariate linear regression model to realize feature extraction and fitting analysis of dynamic data. By comparing the fitting curve with the standard test curve, the system automatically determines the standardization and completeness of the operation, avoiding subjective errors and omissions in human analysis.

[0055] 3. The present invention automatically records the entire process of test data collection, analysis and reporting through the system, and automatically generates a standardized test report based on the analysis results, including abnormal operation areas and detailed data, reducing the workload and operator fatigue of manual data processing. At the same time, the system supports automatic sorting and retrieval based on time, operator, locomotive model and other conditions, which facilitates rapid problem location.

[0056] 4. The present invention comprehensively covers all the judgment rules of the five-step gate test through intelligent analysis, automatically marks unfinished steps, irregular operations or abnormal data, and generates quantitative scoring results to ensure the accuracy and traceability of the detection process. With the help of digitized results, it replaces the traditional multi-manual repeated test verification process, reducing labor costs and test complexity. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 It is a flow chart of the steps of the present invention;

[0059] Figure 2 The figure is a block diagram of the software and hardware system of the method of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] Example

[0063] like Figure 1-2 As shown, a method for determining the performance of a locomotive brake detection in this embodiment includes the following specific implementation steps:

[0064] Data collection:

[0065] The dynamic data generated during the brake test is collected in real time using the on-board equipment on the locomotive. The on-board equipment includes but is not limited to BCU (brake control unit), LKJ (train operation control recording device), 6A (on-board safety protection system), TCMS (train control and management system) and CMD (remote monitoring and diagnosis system);

[0066] Dynamic data include air cylinder pressure, pipeline pressure, brake handle position signal and test time, and the collected data are stored synchronously in time series; to ensure the synchronization and integrity of data collection, the on-board equipment exports the data through the IC card recording device or wireless transmission. During data collection, the IC card recording device and the handheld device complete data interaction through a dedicated hardware interface or wireless communication. The handheld device can dump, query, and retrieve the data in the IC card, and even selectively copy, delete, modify or wirelessly transmit the data to the ground analysis main station. In order to meet the adaptation requirements of different types of locomotives, the equipment supports parameter settings based on regional specifications and dynamically adjusts the storage format; at the same time, to ensure data timing consistency, the system matches and corrects the timestamps of the collected data to provide high-quality original data.

[0067] Data processing:

[0068] The collected dynamic data is denoised and outliers are removed to improve the accuracy of subsequent analysis. Combined with historical test data, a test standard curve set is constructed. The standard curve set is classified and stored based on different locomotive models and brake models, and the parameter range can be expanded through dynamic learning. The extended parameters include upper and lower pressure thresholds, time tolerance range and handle action amplitude threshold. The standard curve set is generated by collecting pressure change data of locomotives of different models under normal test conditions, forming a curve database adapted to different locomotive test conditions.

[0069] Fitting and analysis:

[0070] After data preprocessing, feature sampling technology and multi-segment fitting algorithm are used to generate the feature fitting curve of the collected data:

[0071] Feature sampling: The sliding window method is used to extract key feature points in dynamic data, including the peak value, inflection point and slope change rate of pressure change. The characteristic value is calculated by the following formula:

[0072]

[0073] Multi-segment fitting: The pressure change trend curve is fitted by the segmented least squares method, and the sum of squares of the fitting error is calculated by the following formula:

[0074]

[0075] Fitting analysis generates a continuous pressure change trend curve, and verifies the credibility of characteristic points through fitting errors to optimize the input data of regression analysis;

[0076] Multiple linear regression analysis: Slopes of combined fitted curve parameters ,intercept , fitting error, and the push-pull speed and push-pull amplitude of the handle of the dynamic variables of the characteristic sampling points, and construct a multivariate linear regression model. The formula is: .

[0077] Five-step gate test operation judgment:

[0078] The five-step brake test is the key to the performance detection of the locomotive brake. By analyzing the multi-step operation of the brake handle and the corresponding pressure changes, it is determined whether the operation of the brake meets the operating specifications and performance standards. This method uses an intelligent analysis system to automatically identify the operating steps of the five-step brake test, and combines feature extraction and curve fitting to comprehensively determine the integrity of the operating steps and the execution quality of key steps.

[0079] In the five-step brake test, the operator pushes and pulls the brake handle according to the test specifications. In order to accurately determine the standardization of the operation, the key characteristics of each step of the operation are analyzed, including:

[0080] Push and pull speed: Record the movement speed of the handle in each step of the operation and compare it with the standard speed range. For example, if the handle is pushed or pulled too fast, it may cause a lag in pressure response. If the speed is too slow, it may cause the hidden danger of incomplete braking.

[0081] Movement amplitude: The movement amplitude is the distance from the starting position to the target position of the handle. The system extracts the amplitude value of each push and pull action to determine whether it reaches the specified range. For example, insufficient amplitude may cause the pressure to fail to reach the set value, affecting the braking effect;

[0082] Operation interval: The time interval between handle operations is a parameter of the five-step gate test. By recording the time point of each step of operation and calculating the time interval between adjacent actions, it is determined whether it meets the specification requirements. If the time interval is too short, the pressure change will not be stable, and if it is too long, it will easily affect the continuity of the operation.

[0083] The system uses the collected dynamic data, combined with the characteristic sampling algorithm and the piecewise least squares fitting algorithm, to generate the pressure change trend curve corresponding to each step of the operation. By comparing the fitting curve with the standard test curve, the system analyzes the standardization of the operation, including:

[0084] The system compares the overall trend of the actual curve with the standard curve to determine whether each operation step is completed in accordance with the established specifications. For example, when the standard curve shows a steady upward trend, if the actual curve shows abnormal fluctuations, it may be caused by unstable handle operation or equipment failure.

[0085] The system automatically marks abnormal areas when it detects the following conditions in a test step:

[0086] If a step in the five-step gate test is not executed or the data is missing, the system automatically records the step as missing and generates a prompt message;

[0087] Actions that do not meet operating requirements: If the push-pull speed, movement amplitude or time interval of the handle are not within the specification range, the system will mark the abnormal action and output detailed abnormal parameters, such as the operation speed is too fast or the amplitude is insufficient;

[0088] Abnormal pressure response: If the pressure response in the fitting curve does not meet the standard requirements, such as the pressure peak is low or the recovery speed is slow, the system will mark the corresponding abnormal pressure section and associate its operation steps.

[0089] In order to adapt to different locomotive models, brake types and regional specifications, the determination range of operating parameters is dynamically adjusted, including:

[0090] Adjustment of upper and lower pressure limits: The pressure response ranges of different locomotives may differ. The system dynamically adjusts the upper and lower pressure limits based on the comparison results of the standard curve and the actual curve. For example, the default 20kPa error range can be expanded or reduced to adapt to the pressure fluctuation characteristics of specific equipment.

[0091] Time interval adjustment: Based on the operation standards of different regions, the system allows users to set different time tolerance ranges. For example, the default 5-second operation interval can be adjusted to 3-7 seconds to meet the regulatory requirements of a specific region.

[0092] Adjustment of handle push-pull speed range: The system supports adjustment of the standard speed range, for example, expanding the handle push-pull speed from 0.5 m / s to 0.4-0.6 m / s according to the locomotive model to adapt to specific operating conditions.

[0093] Fault analysis and deduction:

[0094] In the test judgment process, the dynamic characteristics of abnormal data and the historical fault case library are combined to use the fault case decision tree model to deduce the possible fault causes. Specifically,

[0095] The analysis of abnormal data is realized through the fault case decision tree model. The decision tree model is built based on historical test data and fault case library. The node weights of the decision tree model are Calculated by the following formula:

[0096]

[0097] in, Indicates the fault type The weight value of Indicates the historical occurrence frequency of the corresponding fault type, Represents the total number of all fault types in the decision tree;

[0098] Match the dynamic features of the detected abnormal data with the fault features in the decision tree nodes and calculate the matching confidence of each fault node , the matching confidence value range is 0~1, indicating the abnormal data and fault type The degree of feature matching;

[0099] Combined node weights and matching confidence , calculate the probability of each failure type , the formula is:

[0100] in Fault Type The probability value of the fault is generated by recursively analyzing all nodes in the decision tree model. , and sort them from high to low according to the probability value, output the fault type with the highest probability as the judgment result, and locate the fault in combination with the characteristic information of the fault type. When multiple fault types have close probability values, mark the fault type and prompt manual confirmation.

[0101] Methods An intelligent analysis system was used, which was divided into parameter management module, five-step gate step recognition module and abnormal point analysis module.

[0102] The parameter management module is used to store and configure the operating specifications, locomotive models and environmental conditions involved in the test process, and supports users to dynamically modify parameters and intervene in the analysis process during the test operation;

[0103] The five-step gate step identification module identifies the completeness and accuracy of each step in the test by sampling and fitting the pressure change trend during the test, and marks the steps that do not meet the operating specifications;

[0104] The abnormal item point analysis module combines the preset algorithm model and adopts the method of combining global trend analysis with local feature point extraction to analyze the detection results and generate abnormal pressure areas and possible fault types.

[0105] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for determining the performance of a locomotive brake, characterized in that: The determination method comprises: The on-board equipment on the locomotive is used to collect dynamic data generated during the brake test in real time, the dynamic data including air cylinder pressure, pipeline pressure, brake handle position signal and other operation-related parameters; Preprocess the collected dynamic data, and build a test standard curve set in combination with historical test data. The standard curve set is classified and stored according to locomotive model and brake machine model; Generate a characteristic fitting curve of the collected data by using characteristic sampling technology and multi-segment fitting algorithm; compare and analyze the fitting curve with the standard curve to determine whether the change of the working pressure of the brake meets the operating specifications, and the analysis includes global trend state analysis and local characteristic point change analysis; According to the operating specifications of the five-step gate test, the completeness of the operating steps and the execution quality of the key steps are automatically identified, including the judgment of the push-pull speed, push-pull amplitude and operation continuity of the brake handle; if an unfinished or abnormal operation is identified, the corresponding abnormal section is marked; In the judgment process, the dynamic characteristics of abnormal data and the historical fault case library are combined to use the decision tree model to deduce possible fault causes and output high-probability fault types and location information; Dynamically adjust the analysis rules and parameter thresholds in the algorithm model to meet the different locomotive brake inspection requirements to adapt to different operating environments, test conditions and regional specifications; The dynamic data collection includes collecting key parameters of the brake test through an IC card recording device, and dumping, querying and retrieving the data in the IC card through a handheld device; the handheld device can selectively copy, delete, modify or wirelessly transmit the IC card data to a ground analysis main station; The IC card recording device and the handheld device complete data exchange through a dedicated hardware interface or wireless communication. The IC card recording device and the handheld device are adapted to the data collection requirements of various brake machine models and support parameter settings and storage format adjustments based on the requirements of different railway companies; The analysis of the dynamic data is based on a standard test data set and an algorithm model. The standard test data set is generated by collecting pressure change data of different models of locomotives under normal test conditions. In the dynamic data analysis, the preset algorithm model includes a feature sampling algorithm, a least squares fitting algorithm and a multivariate linear regression algorithm, which are used to extract features, perform fitting analysis on the collected data and determine whether the operation is standardized, including the following steps: (1) Feature sampling algorithm: Extract key feature points of pressure change from dynamic data, including peak value, inflection point and slope change rate. Use sliding window method to calculate the feature value of feature sampling window by the following formula: ,in, is the eigenvalue of the feature sampling window, The pressure points collected are is the mean value within the window, is the window size, the extracted feature values ​​are sorted according to the time series, and the key feature points are screened out to form a preliminary feature set as the input data for fitting analysis; (2) Least squares fitting algorithm: Based on the peak value, inflection point and slope change rate of the pressure change of the collected data, the piecewise least squares method is used to fit the pressure change trend curve, and the sum of squares of the fitting error is It is expressed as: ,in, Indicates the pressure value of the collected data point, Indicates the corresponding time point, and Respectively The fitted slope and intercept of the segment, is the number of segments; by minimizing Optimize fitting parameters and ,The fitting analysis generates a continuous trend curve based on the characteristic sampling points, and verifies the credibility of the characteristic points through the fitting error, and optimizes the input data of the regression analysis; (3) Multiple linear regression algorithm: combining the slope of the fitting curve parameters ,intercept , fitting error, and the push-pull speed and push-pull amplitude of the handle of the dynamic variables of the characteristic sampling points, and construct a multivariate linear regression model. The formula is: ,in, Indicates the predicted value of pressure change trend, , , , are the local characteristic values ​​of the handle push-pull speed, push-pull amplitude and pressure change, respectively. is the bias term, , , , is the regression coefficient, The regression analysis comprehensively fits the global trend information of the curve and the local characteristics of the feature points to determine the standardization of the operation, and combines the regression residual analysis to mark the abnormal sections that do not meet the standards.

2. A method for determining the performance of a locomotive brake according to claim 1, characterized in that: The determination of the operation steps includes the integrity determination of the five-step brake test. The determination rules of the five-step brake test are formulated according to the operating procedures of different types of locomotives, including the handle action amplitude, handle action speed and the timing relationship between each action; The identification of the five-step gate test operation is based on a combination of global trend analysis of pressure changes and local feature sampling. A segmented fitting algorithm is used to determine whether the pressure changes in each action segment are consistent with standard specifications. When there are missing steps or actions that do not meet the operating requirements in the five-step gate test, the system will automatically mark the abnormal area and output the abnormal data of the relevant operations for analysis and processing.

3. A method for determining the performance of a locomotive brake according to claim 1, characterized in that: The analysis of abnormal data is realized by a fault case decision tree model, which is constructed based on historical test data and a fault case library. The node weights of the decision tree model are Calculated by the following formula: ,in, Indicates the fault type The weight value of Indicates the historical occurrence frequency of the corresponding fault type, Represents the total number of all fault types in the decision tree; Match the dynamic features of the detected abnormal data with the fault features in the decision tree nodes and calculate the matching confidence of each fault node , the matching confidence value range is 0~1, indicating the abnormal data and fault type The degree of feature matching; Combined node weights and matching confidence , calculate the probability of each failure type , the formula is: ,in Fault Type The probability value of the fault is generated by recursively analyzing all nodes in the decision tree model. , and sort them from high to low according to the probability value, output the fault type with the highest probability as the judgment result, and locate the fault in combination with the characteristic information of the fault type. When multiple fault types have close probability values, mark the fault type and prompt manual confirmation.

4. A method for determining the performance of a locomotive brake according to claim 1, characterized in that: The dynamic data is collected by on-board equipment integrated on the locomotive, including but not limited to BCU, LKJ, 6A, TCMS and CMD; The on-board equipment records the air cylinder pressure, pipeline pressure, brake handle position signal and test time during the brake operation in real time, and synchronously stores the collected data in time series, and exports the data for analysis through an IC card recording device or wireless transmission method; the on-board equipment supports adaptation to different models of locomotives to ensure synchronization and completeness of data collection.

5. A method for determining the performance of a locomotive brake according to claim 1, characterized in that: The method is implemented based on the locomotive five-step brake test operation specification, analyzes the push-pull speed, action amplitude and operation interval of the handle position during the brake operation process, and judges whether the operation steps of the five-step brake test are complete and meet the specifications; by performing segmented fitting analysis on the pressure change trend of each action segment, it is verified whether the test operation meets the operation standard; if it is detected that there is a missed operation or non-standard operation in the five-step brake test, the abnormal operation segment is marked and the specific abnormal reason is output; the analysis is based on the standard operation database, and the judgment range of the operation parameters is dynamically adjusted in combination with the operation specifications of different railway companies or regions.

6. A method for determining the performance of a locomotive brake according to claim 1, characterized in that: The method adopts an intelligent analysis system, which is divided into a parameter management module, a five-step gate step identification module and an abnormal point analysis module; The parameter management module is used to store and configure the operating specifications, locomotive models and environmental conditions involved in the test process, and supports users to dynamically modify parameters and intervene in the analysis process during the test operation; The five-step gate step identification module identifies the completeness and accuracy of each step in the test by sampling and fitting the pressure change trend during the test, and marks the steps that do not meet the operating specifications; The abnormal item point analysis module combines the preset algorithm model and adopts the method of combining global trend analysis with local feature point extraction to analyze the detection results and generate abnormal pressure areas and possible fault types.

7. A method for determining the performance of a locomotive brake according to claim 6, characterized in that: The method automatically records the entire process of collecting, analyzing and reporting test data through an intelligent analysis system.

Images