Health monitoring method, device, equipment and medium for vehicle running part

Abstract

The application discloses a vehicle running gear health monitoring method, device, equipment and medium, relates to the technical field of rail transit health monitoring, and comprises the following steps: extracting current characteristic values corresponding to each target degradation feature from the collected current running state data of the vehicle running gear; performing nonlinear mapping on the current characteristic values to obtain target single-feature health indexes of each target degradation feature; determining target single-dimension feature contribution degrees of each target degradation feature under each component fault dimension, weighting and summing the target single-feature health indexes according to the target single-dimension feature contribution degrees to obtain a target comprehensive health index used for quantifying the health state of the vehicle running gear; determining a target operation and maintenance level matched with the target comprehensive health index, and performing operation and maintenance on the vehicle running gear according to an operation and maintenance suggestion corresponding to the target operation and maintenance level. The accuracy and reliability of vehicle running gear health monitoring are improved.

Description

Technical Field

[0001] This invention relates to the field of rail transit health monitoring technology, and in particular to methods, devices, equipment and media for health monitoring of vehicle running gear. Background Technology

[0002] Health index is a state value that characterizes the degradation process of components. In the field of rail transit, the calculation of health index of key components of the running gear is crucial for equipment operation and maintenance decisions. However, the current health index calculation in this field suffers from problems such as inconsistent indicator systems, fixed weights, poor data quality, difficulty in modeling coupling relationships, lack of standards, and insufficient model generalization and interpretability. This results in inaccurate indices, difficulty in implementation, and difficulty in guiding operation and maintenance. The core problems are concentrated in the defects of indicators and weights (incomplete indicator coverage, weights are mostly static and constant, which cannot be dynamically adjusted with the equipment status and are prone to misjudgment) and the lack of models and standards (algorithms mostly rely on single signals, have low efficiency and weak robustness in multimodal fusion, poor model interpretability, and are difficult to use in safety-critical scenarios).

[0003] Currently, there are three main approaches in the industry: the health index calculation method based on multi-feature fusion constructs a health index by fusing multiple signal processing features and dynamically weighting them, but it suffers from drawbacks such as subjective weight allocation, feature redundancy and noise, and limited generalization ability; the health index calculation method based on distance measures the distance between healthy and faulty signals as the health index, but it faces problems such as high sensitivity to data distribution, lack of ability to capture nonlinear relationships, and high dependence on benchmark points; and the health index calculation method based on deep learning uses deep learning architecture to automatically learn features and construct a health index, but it has stringent data requirements, poor model interpretability, and high computational cost.

[0004] In summary, improving the accuracy and reliability of vehicle running gear health monitoring is a problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method, apparatus, device, and medium for health monitoring of vehicle running gear, thereby improving the accuracy and reliability of vehicle running gear health monitoring. The specific solution is as follows: In a first aspect, this application discloses a method for health monitoring of the running gear of a vehicle, comprising: Extract the current feature value corresponding to each target degradation feature from the current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; The current feature value is non-linearly mapped to obtain the target single feature health index of each target degradation feature; The dynamic weight of the target single feature is determined based on the contribution of each target degradation feature to the target single-dimensional feature under the fault dimension of each component, and the target single feature health index is weighted and summed based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear. A target maintenance level matching the target comprehensive health index is determined from the preset maintenance levels, and the vehicle running gear is maintained according to the maintenance recommendations corresponding to the target maintenance level.

[0006] Optionally, before extracting the current feature value corresponding to each target degradation feature from the collected current operating status data of the vehicle running gear, the method further includes: Historical operating status data of the vehicle running gear is collected, and the initial degradation characteristics of each component in the vehicle running gear are determined based on the historical operating status data, so as to construct a basic feature library including each of the initial degradation characteristics; Target degradation features are selected from the basic feature library.

[0007] Optionally, the collection of historical operating status data of the vehicle's running gear includes: Collect historical operating status data of bearings, treads, and gears in the vehicle's running gear; wherein, the historical operating status data includes any one or more of the following types of data: sample data, impact dB data, impact SV data, temperature data, vibration data, and mileage data.

[0008] Optionally, determining the initial degradation characteristics of each component in the vehicle running gear based on the historical operating status data includes: Extract derived features of each component in the vehicle running gear from the historical operating status data; Time-domain and frequency-domain analyses are performed on the historical operating status data to obtain the time-domain and frequency-domain characteristics of each component in the vehicle running gear. The derived features, time-domain features, and frequency-domain features corresponding to the bearings, treads, and gears in the vehicle running gear are respectively determined as the initial degradation features of the bearings, the initial degradation features of the treads, and the initial degradation features of the gears.

[0009] Optionally, the step of filtering target degradation features from the basic feature library includes: The degradation characterization capability of each initial degradation feature in the basic feature library for the component is evaluated, and target degradation features are selected from the basic feature library based on the degradation characterization capability.

[0010] Optionally, the step of selecting target degradation features from the basic feature library based on the degradation characterization capability includes: Candidate degradation features that meet preset conditions are selected from the basic feature library; wherein, the preset conditions are that they effectively characterize the continuous degradation trend of the component and have clear physical meaning. The candidate degradation features are evaluated for predictability and / or subjected to chi-square tests. Based on the evaluation and / or test results, the target degradation features are selected from the candidate degradation features.

[0011] Optionally, the predictability of the candidate degradation features is evaluated, including: Determine the first mean of the candidate degradation feature at the initial time, the second mean at the failure time, and the standard deviation at the failure time; The absolute difference between the first mean and the second mean is determined as the degradation magnitude of the current candidate degradation feature, and the evaluation result of predictability evaluation is obtained based on the ratio between the standard deviation and the degradation magnitude.

[0012] Optionally, a chi-square test is performed on the candidate degradation features, including: The frequency of each historical feature value of the current candidate degradation feature in a preset interval is counted to obtain the actual observation frequency of each historical feature value. Obtain the difference between the actual observation frequency and the corresponding theoretical observation frequency; The ratio of the square of the difference to the theoretical observation frequency is obtained, and the ratios of each of the historical feature values ​​are summed to obtain the chi-square test result.

[0013] Optionally, the process of selecting target degradation features from the candidate degradation features based on the obtained evaluation results and / or test results includes: Target degradation features are selected from the candidate degradation features whose evaluation results are greater than a first preset threshold and / or whose test results are greater than a second preset threshold.

[0014] Optionally, the step of performing a nonlinear mapping on the current feature value includes: The current feature value is processed to handle outliers and missing values, so as to obtain the processed current feature value; The processed current feature value is smoothed using a first preset sliding window to obtain a standardized current feature value; A nonlinear mapping is performed on the standardized current feature values.

[0015] Optionally, the nonlinear mapping of the standardized current feature values ​​includes: The standardized current feature values ​​are nonlinearly mapped using a target arctangent model; wherein the target arctangent model is the optimal mapping relationship between each target degradation feature and a single feature health index fitted based on the historical feature values ​​of the target degradation feature.

[0016] Optionally, before performing the nonlinear mapping on the standardized current feature values, the method further includes: Construct an initial arctangent model with a single-feature health index as output and the feature values ​​of the target degradation feature as input; Construct an objective function; wherein the objective function aims to minimize the error between the single-feature health index output by the initial arctangent model and the standard single-feature health index; The objective function is optimized using the least squares method to solve for the amplitude coefficient, slope coefficient, offset coefficient, and baseline coefficient in the initial arctangent model, thereby obtaining the target arctangent model.

[0017] Optionally, the objective function is: ; Where y is the single-feature health index, and x is the current feature value. denoted as a, b as the amplitude coefficient, c as the slope coefficient, and d as the baseline coefficient. For the first There are 1 eigenvalues, where p is the total number of eigenvalues.

[0018] Optionally, determining the dynamic weight of a single target feature based on the contribution of each target degradation feature to the single-dimensional feature of each component failure dimension includes: Linear regression is performed on the feature value sequence of the current target degradation feature within the second preset sliding window to obtain the fitting determination coefficient of the current target degradation feature under the trend stability dimension and the normalized trend slope under the risk rising trend dimension. The fitting determination coefficient and the normalized trend slope are respectively determined as the contribution of the first target single-dimensional feature and the contribution of the second target single-dimensional feature. The ratio of the current feature value of the current target degradation feature to the preset fault threshold is determined as the third target single-dimensional feature contribution of the current target degradation feature under the fault risk level dimension.

[0019] Optionally, determining the dynamic weight of a single target feature based on the contribution of each target degradation feature to the single-dimensional feature of each component failure dimension includes: The overall target contribution of each target degradation feature is obtained based on its contribution to the single-dimensional feature of each component failure dimension; the formula for obtaining the overall target contribution is as follows: ; ; in, As a comprehensive contribution to the target, Contribution to the first objective's single-dimensional feature. Contribution to the single-dimensional feature of the second objective. Contribution to the single-dimensional features of the third objective. Preset importance under the component failure dimension; The overall contribution of each of the target degradation features is normalized to obtain the dynamic weight of each target single feature of the target degradation feature.

[0020] Optionally, normalizing the overall contribution of each of the target degradation features to obtain the dynamic weight of each target single feature of the target degradation feature includes: Obtain the sum of the target comprehensive contribution of each target degradation feature, and determine the ratio between the target comprehensive contribution of the current target degradation feature and the sum as the target single feature dynamic weight of the current target degradation feature.

[0021] Optionally, the preset maintenance levels include a preset normal level, a preset sub-health level, and a preset fault level. The maintenance recommendation corresponding to the preset normal level is to maintain routine monitoring and maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset sub-health level is to conduct key monitoring and preventive maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset fault level is to perform emergency treatment and repair or replacement of the vehicle's running gear.

[0022] Secondly, this application discloses a health monitoring device for the running gear of a vehicle, comprising: The feature acquisition module is used to extract the current feature value corresponding to each target degradation feature from the current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; The index mapping module is used to perform nonlinear mapping on the current feature value to obtain the target single feature health index of each target degradation feature; The health index quantification module is used to determine the dynamic weight of the target single feature based on the contribution of each target degradation feature to the target single-dimensional feature under each component failure dimension, and to perform a weighted summation of the target single feature health index based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear. The operation and maintenance module is used to determine the target operation and maintenance level that matches the target comprehensive health index from the preset operation and maintenance levels, and to perform operation and maintenance on the vehicle running gear according to the operation and maintenance recommendations corresponding to the target operation and maintenance level.

[0023] Thirdly, this application discloses an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the steps of the aforementioned disclosed method for monitoring the health of the running gear of a vehicle.

[0024] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed method for monitoring the health of the running gear of a vehicle.

[0025] The beneficial effects of this application are as follows: This application extracts the current feature values ​​corresponding to each target degradation feature from the collected current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; the current feature values ​​are nonlinearly mapped to obtain the target single feature health index of each target degradation feature; the target single feature dynamic weight is determined according to the target single feature contribution of each target degradation feature under the fault dimension of each component, and the target single feature health index is weighted and summed according to the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear; a target maintenance level matching the target comprehensive health index is determined from the preset maintenance levels, and the vehicle running gear is maintained according to the maintenance recommendations corresponding to the target maintenance level. Therefore, this application, by extracting the current feature value corresponding to the target degradation feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state from the current operating status data of the vehicle running gear, ensures that the extracted features can accurately reflect the complete degradation process of the component from normal to fault, avoiding misjudgment of health status due to insufficient correlation between features and degradation / fault. The target single-feature health index is obtained by nonlinearly mapping the current feature value, which can simulate the actual law of continuous decline in component health status, making the single-feature health index more closely match the actual health changes of the component and improving the accuracy of single-feature-level health assessment. The dynamic weight of the target single feature is determined based on the contribution of the target degradation feature to the target single-dimensional feature under each component fault dimension, enabling dynamic adjustment of the weight according to the feature degradation state, overcoming the limitation of traditional fixed weights that cannot adapt to changes in component status, making weight allocation more objective and highlighting the impact of high-risk features. The target single-feature health index is weighted based on the dynamic weight of the target single feature. The summation yields a target comprehensive health index, which integrates health information from multiple features to comprehensively reflect the overall health status of components. Furthermore, since feature extraction, nonlinear mapping, and dynamic weight determination all revolve around the correlation between component degradation and faults, the accuracy of the comprehensive health index in quantifying component health status is improved. The method matches the target maintenance level corresponding to the target comprehensive health index from preset maintenance levels and provides maintenance recommendations accordingly. This directly transforms the quantified health index into a targeted basis for maintenance decisions, making maintenance work more scientific, precise, and instructive. It effectively solves the problems of inaccurate indexes, difficulty in implementation, and difficulty in guiding maintenance in traditional rail transit health index calculations. Moreover, the method in this application does not rely on a large number of fault samples to achieve dynamic calculation of the health index, possessing good generalization ability and feature scalability. It can also provide effective support for research on health assessment and life prediction of vehicle running gear components, helping to achieve predictive maintenance of equipment, reduce the probability of fault occurrence, and ensure the stable and safe operation of vehicle running gear. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0027] Figure 1 This application discloses a flowchart of a method for health monitoring of a vehicle running gear. Figure 2 This is a schematic diagram of the structure of a health monitoring device for a vehicle running gear disclosed in this application; Figure 3 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0028] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0029] Health index is a state value that characterizes the degradation process of components. In the field of rail transit, the calculation of health index of key components of the running gear is crucial for equipment operation and maintenance decisions. However, the current health index calculation in this field suffers from problems such as inconsistent indicator systems, fixed weights, poor data quality, difficulty in modeling coupling relationships, lack of standards, and insufficient model generalization and interpretability. This results in inaccurate indices, difficulty in implementation, and difficulty in guiding operation and maintenance. The core problems are concentrated in the defects of indicators and weights (incomplete indicator coverage, weights are mostly static and constant, which cannot be dynamically adjusted with the equipment status and are prone to misjudgment) and the lack of models and standards (algorithms mostly rely on single signals, have low efficiency and weak robustness in multimodal fusion, poor model interpretability, and are difficult to use in safety-critical scenarios).

[0030] Currently, there are three main approaches in the industry: the health index calculation method based on multi-feature fusion constructs a health index by fusing multiple signal processing features and dynamically weighting them, but it suffers from drawbacks such as subjective weight allocation, feature redundancy and noise, and limited generalization ability; the health index calculation method based on distance measures the distance between healthy and faulty signals as the health index, but it faces problems such as high sensitivity to data distribution, lack of ability to capture nonlinear relationships, and high dependence on benchmark points; and the health index calculation method based on deep learning uses deep learning architecture to automatically learn features and construct a health index, but it has stringent data requirements, poor model interpretability, and high computational cost.

[0031] Therefore, this application provides a health monitoring solution for the running gear of a vehicle, which improves the accuracy and reliability of health monitoring of the running gear.

[0032] See Figure 1 As shown in the figure, this application discloses a method for health monitoring of a vehicle running gear, including: Step S11: Extract the current feature value corresponding to each target degradation feature from the collected current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state.

[0033] In this embodiment, before extracting the current feature value corresponding to each target degradation feature from the current operating status data of the collected vehicle running gear, the method further includes: collecting historical operating status data of the vehicle running gear, and determining the initial degradation features of each component in the vehicle running gear based on the historical operating status data, so as to construct a basic feature library including each initial degradation feature; and selecting target degradation features from the basic feature library.

[0034] Before extracting the current feature values ​​corresponding to the degradation characteristics of each target in the current operating status data of the vehicle running gear, historical operating status data of key components such as bearings, gears, and treads of the vehicle running gear are first collected. It should be noted that the collected historical operating status data does not include alarm data, because alarm data can only reflect whether the health status has reached the critical point of the fault threshold, and cannot characterize the degree of degradation or the rate of deterioration of the component's health status. It does not have the ability to represent the continuous trend to support the calculation of the health index. Therefore, alarm data is not collected directly. At the same time, relevant data based on fault mechanism, expert analysis experience, and data statistical indicators are included, and these historical operating statuses are combined with dimensions such as numerical magnitude, peer comparison, and trend change rate. A comprehensive analysis of the data was conducted to extract relevant features that characterize the degradation process of each component as initial degradation features. All initial degradation features were integrated to construct a basic feature library for vehicle running gear components. This feature library covers various features related to the degradation of rotating parts throughout their entire life cycle, laying the foundation for subsequent feature selection and health index calculation. After the basic feature library was constructed, the initial degradation features in the basic feature library were analyzed from the perspectives of data type adaptability and feature representation capability. Target degradation features with good predictive and interpretive capabilities were selected to accurately capture the core information related to component deterioration and failure, providing accurate and effective feature basis for subsequent extraction of current feature values ​​and calculation of health index.

[0035] In this embodiment, the collection of historical operating status data of the vehicle running gear includes: collecting historical operating status data of the bearings, treads, and gears of the vehicle running gear; wherein, the historical operating status data includes any one or more types of data such as sample data, impact dB data, impact SV data, temperature data, vibration data, and mileage data.

[0036] Collecting historical operating status data of bearings, treads, and gears in the running gear of rail transit locomotives / vehicles is a comprehensive monitoring data collection operation targeting the core rotating components of the running gear. The collected historical operating status data comprehensively covers various types of onboard online monitoring data, including sample data, impact dB data, impact SV data, temperature data, vibration data, and mileage data. For bearing components, the collected sample data includes fault mechanism-related characteristic data such as impact spectral clarity, impact spectral modulation, and impact spectral multi-order characteristics. Impact dB data includes statistical data such as dB average value, dB frequency, and dB trend change. Impact SV data includes the SV average value. Impact characteristic data includes daily average SV values ​​and higher values ​​compared to the same location; temperature data includes data indicating temperature anomalies such as higher values ​​compared to the same location and the number of times higher values ​​compared to the same location; vibration data includes multi-dimensional vibration characteristic data such as daily average vibration effective values ​​and higher values ​​of the same location vibration effective values; mileage data consists of operational mileage data at different levels. For tread components, the collected sample data includes specific characteristic data such as impact spectrum clarity, impact spectrum modulation, and impact duty cycle. Impact dB data, impact SV data, temperature data, and vibration data are all set with corresponding collection dimensions matched to the fault characterization characteristics of tread components. Mileage data also consists of operational mileage data at different levels. For gear components, the collected sample data includes impact spectral clarity, impact spectral modulation, and whether fault transmission characteristics exist. Impact dB data includes average dB value and dB frequency. Impact SV data includes average SV value and daily average SV value with higher values ​​compared to the same location. Vibration data includes daily average values ​​of the same location vibration effective value with higher values ​​compared to the same location, etc., data adapted to gear characteristics. Additionally, operational mileage data for different levels of gears is also collected. All data originates from the vehicle's onboard online monitoring system for the running gear, capable of comprehensively capturing various status information of bearings, treads, and gears during operation. Furthermore, each data type has specific collection dimensions set around component degradation and fault characterization. It includes discrete alarm data reflecting abnormal equipment operation, continuous monitoring data reflecting component failure impact, vibration, and temperature changes, mileage data reflecting component fatigue wear, and sample feature data extracted based on failure mechanisms. This enables multi-dimensional and comprehensive data collection of the operating status of bearings, treads, and gears throughout their entire life cycle. It provides comprehensive and accurate raw data support for the subsequent construction of a basic feature library based on historical operating status data and the screening of target degradation features. This ensures that the constructed feature library can fully cover the degradation-related features of the core components of the running gear, and that the target degradation features screened later can accurately characterize the deterioration and failure status of the components.

[0037] In this embodiment, determining the initial degradation characteristics of each component in the vehicle running gear based on the historical operating state data includes: extracting derived features of each component in the vehicle running gear from the historical operating state data; performing time-domain analysis and frequency-domain analysis on the historical operating state data to obtain the time-domain features and frequency-domain features of each component in the vehicle running gear; and determining the derived features, time-domain features, and frequency-domain features corresponding to the bearings, treads, and gears in the vehicle running gear as the initial degradation characteristics of the bearings, the treads, and the gears, respectively.

[0038] First, extract derived features from historical operating status data to identify the degradation and failure characteristics of key components such as bearings, treads, and gears.

[0039] The derived characteristics of bearings include sample data characteristics such as impact spectral clarity and impact spectral modulation; dB data characteristics such as dB average and dB frequency; SV data characteristics such as SV average and daily average SV with higher values ​​compared to the same location; temperature data characteristics such as higher values ​​compared to the same location and the number of times higher values ​​compared to the same location; vibration data characteristics such as daily average vibration effective value and higher values ​​of the same location vibration effective value; and mileage data characteristics such as the operational mileage at different levels.

[0040] The derived features of the tread include sample data features such as specific impact duty cycle and whether there is peeling and abrasion characteristics, vibration data features such as the daily average value of tread out-of-roundness runout and the percentage of tread impact dB, as well as corresponding mileage data features.

[0041] The derived features of gears include alarm, sample, dB, SV, vibration and mileage related data features adapted to their characteristics.

[0042] Various derived features are extracted from historical operating status data by combining dimensions such as numerical magnitude, isotopic comparison, and trend change rate, accurately corresponding to the operating status monitoring dimensions of each component. Then, professional time-domain and frequency-domain analyses are carried out on the historical operating status data of each component. Through time-domain analysis, root mean square value, peak-to-peak value, kurtosis index, waveform index, peak factor, impulse factor, margin index, skewness index, and Gini index are obtained and used as time-domain features. Through frequency-domain analysis, frequency-domain statistical indicators such as average frequency, root mean square frequency, standard deviation frequency, frequency concentration, and frequency kurtosis are obtained and used as frequency-domain features. These time-domain and frequency-domain features serve as general features and can effectively characterize the numerical change patterns and distribution characteristics of vibration, shock, and other states during the operation of each component.

[0043] Finally, features were categorized according to component type. Various derived features extracted from historical operating data for bearings, along with time-domain and frequency-domain features obtained through time-domain and frequency-domain analysis, were integrated to determine the initial degradation features of bearings. Similarly, the specific derived features for treads were integrated with general time-domain and frequency-domain features to determine the initial degradation features of treads. Likewise, the adaptation derived features for gears were integrated with general time-domain and frequency-domain features to determine the initial degradation features of gears. The initial degradation features of each component comprehensively cover both monitoring data-related derived features and signal analysis-related time-domain and frequency-domain features, fully encompassing various relevant features that characterize the entire lifecycle degradation process of components. This provides a rich and consistent feature foundation for the subsequent construction of a basic feature library for vehicle running gear components, allowing the constructed feature library to comprehensively reflect the operating status and degradation trends of key components such as bearings, treads, and gears.

[0044] In this embodiment, the step of selecting target degradation features from the basic feature library includes: evaluating the degradation characterization ability of each initial degradation feature in the basic feature library for the component, and selecting target degradation features from the basic feature library based on the degradation characterization ability.

[0045] First, for all initial degradation features corresponding to bearings, treads, and gears in the basic feature library, a first round of degradation characterization capability assessment is conducted from two core dimensions: data type adaptability and actual feature representation capability. Initial degradation features with basic degradation characterization capability are selected to form a feature set. Then, a second round of quantitative degradation characterization capability assessment is conducted on this feature set, introducing monotonicity, trend, and predictive indicators to construct a quantitative assessment system. Finally, the results of the two rounds of degradation characterization capability assessment are combined with expert analysis experience and practical application experience to eliminate initial degradation features with insufficient characterization capability, low correlation with component degradation and failure state, and no continuous trend characterization capability. Initial degradation features with good continuous trend characterization capability, high failure correlation, strong predictability, and the ability to accurately characterize the degradation trend of the entire component life cycle and have a strong correlation with failure state are selected as target degradation features.

[0046] In this embodiment, the step of selecting target degradation features from the basic feature library based on the degradation characterization capability includes: selecting candidate degradation features from the basic feature library whose degradation characterization capability meets preset conditions; wherein, the preset conditions are that the feature effectively characterizes the continuous degradation trend of the component and has clear physical meaning; performing predictability evaluation and / or chi-square test on the candidate degradation features, and selecting target degradation features from the candidate degradation features based on the obtained evaluation results and / or test results.

[0047] First, all initial degradation features of key components in the running gear, such as bearings, treads, and gears, are identified from the basic feature library. Candidate degradation features meeting preset conditions are selected based on their degradation characterization capabilities. These conditions require effective characterization of the continuous degradation trend of the components and clear physical meaning. The data types of each initial degradation feature are then differentiated. Since the health index is a continuous quantity characterizing the continuous degradation process of a component, the initial degradation features of sample data, dB data, SV data, temperature data, vibration data, and mileage data types are evaluated in conjunction with the component failure mechanism to assess their characterization capabilities for the component degradation process. For example, the evaluation of temperature data for the severe deterioration stage of bearings is conducted. Characteristic value, SV data's characterization value for bearing failure impact intensity, mileage data's characterization value for bearing fatigue degradation, vibration data's characterization value for bearing failure late-stage state, dB data's quantitative characterization value for component failure state, etc., are used to screen out initial degradation features with basic degradation characterization capabilities to form a feature set; then, predictability evaluation and chi-square test are performed on the candidate degradation features, and target degradation features are screened from the candidate degradation features based on the evaluation results and test results, that is, the closer the predictability evaluation value is to 1, the more significant the correlation with the failure index is shown in the chi-square test, the more likely the candidate degradation feature is to be screened as the target degradation feature.

[0048] In this embodiment, the predictability evaluation of the candidate degradation feature includes: determining the first mean of the current candidate degradation feature at the initial time, the second mean at the failure time, and the standard deviation at the failure time; determining the absolute difference between the first mean and the second mean as the degradation magnitude of the current candidate degradation feature, and obtaining the evaluation result of the predictability evaluation based on the ratio between the standard deviation and the degradation magnitude.

[0049] First, based on real-world monitoring data of key components such as bearings, treads, and gears in the vehicle's running gear, the first mean of the characteristic values ​​corresponding to the initial moment of the component's normal operation is determined, and the second mean of the characteristic values ​​corresponding to the failure moment of the component in the fault stage is determined. Simultaneously, the standard deviation of all characteristic values ​​of the candidate degradation feature at the failure moment is calculated. This standard deviation reflects the dispersion of the characteristic values ​​of the candidate degradation feature at the failure moment. Then, the absolute difference between the first mean and the second mean is calculated, and this absolute difference is determined as the degradation magnitude of the current candidate degradation feature. This degradation magnitude directly reflects the overall change of the candidate degradation feature from the component's initial normal state to its failure state. The larger the degradation magnitude, the more significantly the candidate degradation feature represents the degradation process of the component throughout its entire life cycle. Next, the ratio between the standard deviation at the failure moment and the degradation magnitude is calculated. Based on this ratio, the predictability evaluation result is obtained. The calculation formula is as follows: ; In the formula, This represents the second mean of the candidate degradation feature y at the time of failure. Let represent the first mean of the candidate degenerate feature y at the initial time step. This represents the standard deviation of the candidate degenerate feature y at the time of failure.

[0050] The predictability evaluation result ranges from [0,1]. The smaller the ratio of standard deviation to degradation amplitude, the greater the degradation amplitude of the candidate degradation feature and the lower the dispersion of the feature value at the time of failure. The closer the corresponding predictability evaluation result value is to 1, the better the predictive performance of the candidate degradation feature, the stronger its ability to separate and represent the normal state and fault state of the component, and the more accurately it can reflect the actual degradation trend of the component. Conversely, the lower the ratio, the worse the predictive performance of the candidate degradation feature and the poorer its representation effect on the degradation process of the component. The predictability evaluation completed in this way can accurately assess the predictive ability of candidate degradation features on the degradation trend of the component in a quantitative way, providing an objective and scientific quantitative basis for the subsequent selection of target degradation features with strong representation ability and good predictability, and ensuring that the selected target degradation features can effectively support the accurate calculation of the health index.

[0051] In this embodiment, the chi-square test is performed on the candidate degradation features, including: counting the frequency of each historical feature value of the current candidate degradation feature in a preset interval to obtain the actual observation frequency of each historical feature value; obtaining the difference between the actual observation frequency and the corresponding theoretical observation frequency; obtaining the ratio of the square of the difference to the theoretical observation frequency, and summing the ratios of each historical feature value to obtain the test result of the chi-square test.

[0052] For the current candidate degradation characteristics of key components such as bearings, treads, and gears in the vehicle's running gear, based on the component failure mechanism and actual operating patterns, the historical characteristic values ​​of these characteristics are divided into corresponding preset intervals. Then, the specific frequency of each historical characteristic value of the candidate degradation feature appearing within each preset interval is statistically analyzed. This frequency is determined as the actual observation frequency of each historical characteristic value within the corresponding preset interval. This actual observation frequency directly reflects the actual distribution of the historical characteristic values ​​of the candidate degradation feature across different intervals. Furthermore, based on the correlation logic between the candidate degradation feature and the component failure state, industry experience, and data statistical patterns, the historical characteristic values ​​within each preset interval are determined. The theoretical expected frequency of a eigenvalue is calculated by taking the absolute difference between the actual observed frequency and the corresponding theoretical expected frequency in each preset interval. This deviation value reflects the degree of deviation between the actual and theoretical distributions of the candidate degradation feature's historical values. Then, the deviation value in each preset interval is squared, and the squared result is divided by the theoretical expected frequency in the corresponding interval to obtain the ratio for each preset interval. Finally, these ratios are summed across all preset intervals, and the sum is used as the final chi-square test result for the current candidate degradation feature. The calculation formula is shown below: ; In the formula, O represents the actual observation frequency, and E represents the theoretical expected frequency.

[0053] The magnitude of the test result is positively correlated with the degree of correlation between the candidate degradation feature and the component failure index. The higher the value of the test result, the greater the deviation between the actual observation frequency and the theoretical expected frequency of the candidate degradation feature, the more significant the correlation between the candidate degradation feature and the component failure state, and the stronger the ability to characterize the component failure. Conversely, the lower the value, the lower the correlation between the candidate degradation feature and the component failure state. This chi-square test can objectively determine the significance of the correlation between the candidate degradation feature and the failure state in a quantitative way, providing a scientific quantitative basis for the subsequent screening of target degradation features that are strongly correlated with the failure state, and ensuring that the screened target degradation features can accurately characterize the failure-related degradation process of the component.

[0054] In this embodiment, the step of selecting target degradation features from the candidate degradation features based on the obtained evaluation results and / or test results includes: selecting target degradation features from the candidate degradation features whose evaluation results are greater than a first preset threshold and / or whose test results are greater than a second preset threshold.

[0055] First, a first preset threshold and a second preset threshold are set for the predictability evaluation results and chi-square test results of candidate degradation features, respectively. These thresholds are determined by combining the failure mechanisms of key components such as bearings, treads, and gears in the vehicle running gear, industry application experience, and statistical patterns of real monitoring data. Then, according to the preset screening logic, features with predictability evaluation results greater than the first preset threshold, or chi-square test results greater than the second preset threshold, or both results greater than the corresponding thresholds are selected from the candidate degradation features as target degradation features. Among them, a predictability evaluation result greater than the first preset threshold indicates that the candidate degradation feature has a strong ability to separate and represent the normal and fault states of the component, has good predictability, and can accurately reflect the degradation trend of the component. A chi-square test result greater than the second preset threshold indicates that the candidate degradation feature has a significant correlation with the component failure index and has a strong ability to represent the degradation process related to component failure. Through this quantitative screening method, the target degradation features finally determined are all core features that can effectively represent the continuous degradation trend of the component throughout its entire life cycle and are strongly correlated with the failure state, providing a reliable feature basis for the accurate calculation of the subsequent health index.

[0056] Step S12: Perform a nonlinear mapping on the current feature value to obtain the target single feature health index of each target degradation feature.

[0057] In this embodiment, the nonlinear mapping of the current feature value includes: processing the current feature value for outliers and missing values ​​to obtain the processed current feature value; smoothing the processed current feature value using a first preset sliding window to obtain a standardized current feature value; and performing a nonlinear mapping on the standardized current feature value.

[0058] Outlier processing is performed on the current feature values ​​of the vehicle running gear's target degradation characteristics. Invalid feature values ​​caused by the equipment not being actually in operation are removed. For example, if the operating mileage is less than 30km, it is determined to be not in operation and the corresponding feature value is removed. Then, the missing current feature values ​​are filled with the most recent historical data. After completing the outlier and missing value processing, the processed current feature values ​​are obtained. Subsequently, the processed current feature values ​​are smoothed using a first preset sliding window of 7-day average to eliminate short-term noise in the data and output standardized current feature values ​​to ensure the stability and validity of the feature data. Finally, a nonlinear mapping is performed on the standardized current feature values, that is, the nonlinear transformation from feature values ​​to health index is realized by relying on the arctangent model to simulate the continuous decline law of component health status.

[0059] In this embodiment, the nonlinear mapping of the standardized current feature value includes: using a target arctangent model to perform a nonlinear mapping of the standardized current feature value; wherein, the target arctangent model is the optimal mapping relationship between each target degradation feature and a single feature health index fitted based on the historical feature values ​​of the target degradation feature.

[0060] A target arctangent model specific to the degradation characteristics of each target is pre-fitted. This model relies on the nonlinear asymptotic saturation characteristic of the arctangent function to accurately match the continuous degradation pattern of the component's health status. The specific model is shown below: ; In the formula, A single-feature health index, For the current eigenvalue, denoted as amplitude coefficient, b as slope coefficient, c as offset coefficient, and d as baseline coefficient.

[0061] The standardized current feature value is then substituted into the target arctangent model of the corresponding target degradation feature to complete the nonlinear mapping from feature value to single feature health index, realizing the accurate conversion of feature value to health index, and the mapped single feature health index can fit the actual degradation state change of the component.

[0062] In this embodiment, before performing nonlinear mapping on the standardized current feature value, the method further includes: constructing an initial arctangent model with a single-feature health index as output and the feature value of the target degradation feature as input; constructing an objective function; wherein the objective function aims to minimize the error between the single-feature health index output by the initial arctangent model and the standard single-feature health index; optimizing the objective function using the least squares method to solve for the amplitude coefficient, slope coefficient, offset coefficient, and baseline coefficient in the initial arctangent model to obtain the target arctangent model.

[0063] Before performing a nonlinear mapping on the standardized current feature values, an initial arctangent model is first constructed, with the single-feature health index as the output and the feature values ​​of the target degradation feature as the input. Then, an objective function is constructed based on this initial arctangent model. This objective function aims to minimize the error between the single-feature health index output by the model and the standard single-feature health index. The objective function is as follows: ; in, A single-feature health index, For the current eigenvalue, denoted as a, b as the amplitude coefficient, c as the slope coefficient, and d as the baseline coefficient. For the first There are 1 eigenvalues, where p is the total number of eigenvalues.

[0064] Based on industry experience and case data analysis, the correspondence between key points of feature values ​​and health indices is determined. In a specific case, the total number of feature values ​​for a single feature is 5. For example, the mapping between the key feature of the bearing outer ring fault feature (daily average dB value) and the health index is tentatively set as follows: when the feature value is 0, the corresponding health index is 1; when the feature value is 49, the corresponding health index is 0.6; when the feature value is 55, the corresponding health index is 0.4; when the feature value is 61, the corresponding health index is 0.1; and when the feature value is 100, the corresponding health index is 0.001. By fitting the bearing feature data, the following is obtained: The optimal health index mapping trajectory relationship of a single feature was obtained. After fitting, the average absolute error of the model was less than 0.01, and the health index value corresponding to the feature was highly consistent with the actual health index value. Subsequently, the objective function was optimized using the least squares method to solve for the optimal values ​​of amplitude coefficient, slope coefficient, offset coefficient, and baseline coefficient in the initial arctangent model. The arctangent model after substituting the optimal parameters was determined as the target arctangent model. This model can achieve the optimal mapping between feature values ​​and single feature health index, and the average absolute error after fitting is less than 0.01, which can accurately match the continuous decline pattern of component health status.

[0065] Step S13: Determine the dynamic weight of the target single feature based on the contribution of each target degradation feature to the target single-dimensional feature under the fault dimension of each component, and perform a weighted summation of the target single feature health index based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear.

[0066] In this embodiment, determining the dynamic weight of a target single feature based on the target single-dimensional feature contribution of each target degradation feature under each component failure dimension includes: determining the feature value sequence of the current target degradation feature within a second preset sliding window and performing linear regression to obtain the fitting determination coefficient of the current target degradation feature under the trend stability dimension and the normalized trend slope under the risk rising trend dimension, and determining the fitting determination coefficient and the normalized trend slope as the first target single-dimensional feature contribution and the second target single-dimensional feature contribution, respectively; and determining the ratio of the current feature value of the current target degradation feature to a preset failure threshold as the third target single-dimensional feature contribution of the current target degradation feature under the failure risk level dimension.

[0067] Linear regression is performed on the feature value sequence of the current target degradation feature within the second preset sliding window to obtain the fitting determination coefficient of the current target degradation feature under the trend stability dimension and the normalized trend slope under the risk rising trend dimension. The fitting determination coefficient and the normalized trend slope are determined as the contribution of the first target single-dimensional feature and the contribution of the second target single-dimensional feature, respectively. The ratio of the current feature value of the current target degradation feature to the preset fault threshold is determined as the contribution of the third target single-dimensional feature of the current target degradation feature under the fault risk level dimension. The calculation logic is shown in Table 1. Table 1. Contribution of Target Single-Dimensional Features

[0068] This process uses a 5-day sliding window as the second preset window, performs linear regression on the feature value sequence of the current target degradation characteristics within the window, and obtains the coefficient of determination of fit. As the primary objective of trend stability, the single-dimensional feature contribution is considered. A value closer to 1 indicates a more stable trend and a higher contribution. Simultaneously, the linear regression slope of this feature value sequence is calculated and standardized to the 0-1 range, yielding the normalized trend slope. As the contribution of the second objective single-dimensional feature in the risk upward trend dimension, the larger the value, the faster the feature risk rises and the higher the contribution. Then, the ratio of the current feature value of the current objective degradation feature to the preset fault threshold of the feature is calculated. The ratio is used as the contribution of the third objective single-dimensional feature in the dimension of fault risk. The closer the value is to 1, the higher the fault risk and the higher the contribution of the feature. In this way, the contribution of the single-dimensional feature corresponding to the target degradation feature is obtained from the three core dimensions, which provides a quantitative basis for subsequent calculation of comprehensive contribution and determination of dynamic weight.

[0069] In this embodiment, determining the dynamic weight of a single target feature based on its contribution to the single-dimensional feature of each target degradation feature under each component failure dimension includes: obtaining the comprehensive target contribution of each target degradation feature based on its contribution to the single-dimensional feature of each target degradation feature under each component failure dimension; the formula for obtaining the comprehensive target contribution is: ; ; in, As a comprehensive contribution to the target, Contribution to the first objective's single-dimensional feature. Contribution to the single-dimensional feature of the second objective. Contribution to the single-dimensional features of the third objective. The preset importance is set under the component failure dimension; the target comprehensive contribution of each target degradation feature is normalized to obtain the target single feature dynamic weight of each target degradation feature.

[0070] The overall target contribution of each target degradation feature is obtained based on its contribution to the single-dimensional feature of each component failure dimension. The formula for obtaining the overall target contribution is as follows: ; The overall contribution of each target degradation feature is normalized to obtain the dynamic weight of each target single feature. In this process, a preset importance is first set for the three dimensions of trend stability, risk upward trend, and fault risk degree, and the sum of the three is 1. Then, the feature contribution of the first, second, and third target single dimensions of each target degradation feature is multiplied by the preset importance of the corresponding dimension and then multiplied together to obtain the overall contribution of each target degradation feature. This value comprehensively reflects the overall influence of the feature on the health status of the component in different dimensions. Subsequently, the overall contribution of all target degradation features is normalized to eliminate the influence of the dimensions of the overall contribution of different features, meet the requirement that the value is between 0 and 1 and all weights are summed to 1, and realize the effect of dynamic adjustment of weights according to the feature degradation status.

[0071] In this embodiment, the normalization of the target comprehensive contribution of each target degradation feature to obtain the target single feature dynamic weight of each target degradation feature includes: obtaining the sum of the target comprehensive contribution of each target degradation feature, and determining the ratio between the target comprehensive contribution of the current target degradation feature and the sum as the target single feature dynamic weight of the current target degradation feature.

[0072] Obtain the overall contribution of each of the aforementioned target degradation features. The sum of the total contribution of the current target degradation feature and the ratio of the sum are used to determine the dynamic weight of the single target feature of the current target degradation feature. The specific formula is shown below: ; Where n is the total number of features of the target degradation feature. ; The process first sums the overall contribution of all target degradation features involved in the calculation of the vehicle running gear health index. Then, it calculates the ratio of the current overall contribution of each target degradation feature to this sum, and directly determines the dynamic weight of the corresponding target single feature. This normalization method eliminates the dimensional differences in the overall contribution of different target degradation features, ensuring that the obtained dynamic weights of the target single features are all within the range of 0-1. Furthermore, the sum of the dynamic weights of all target degradation features is 1, which meets the quantitative requirements of weighted fusion. At the same time, this weight can dynamically change with the overall contribution of each target degradation feature, accurately reflecting the actual impact of different features on the health index under the current degradation state of the component, and providing a scientific weighting basis for the subsequent accurate calculation of the fused health index.

[0073] Next, a weighted summation method is used to multiply each target single-feature health index by its corresponding target single-feature dynamic weight, and then sum all the multiplication results. The summation result is the target comprehensive health index, as shown in the following formula: ; In the formula, This represents the dynamic weight of a single target feature. This represents the target single-feature health index. This indicates the target comprehensive health index.

[0074] This index integrates health status information on multi-dimensional target degradation characteristics, and can objectively, comprehensively and accurately quantify the overall health status of key components of the vehicle running gear. Its value can directly reflect the degree of degradation of the components, providing core quantitative basis for subsequent health status assessment and operation and maintenance decisions.

[0075] Step S14: Determine the target maintenance level that matches the target comprehensive health index from the preset maintenance levels, and perform maintenance on the vehicle running gear according to the maintenance recommendations corresponding to the target maintenance level.

[0076] Using the comprehensive health index, which quantifies the health status of the vehicle's running gear, as the core criterion, and combining the failure evolution patterns, severity of failure impact, and characteristics of on-site maintenance handling of key components of the running gear, the system matches the target maintenance level with the target comprehensive health index value range against a preset maintenance level classification standard. This preset maintenance level classification defines different health status ranges based on the health index values ​​and configures targeted maintenance suggestions for each range. Then, the system directly follows the maintenance suggestions corresponding to the matched target maintenance level to carry out appropriate maintenance work on the vehicle's running gear, ensuring that the maintenance work is highly matched with the actual health status of the vehicle's running gear, and achieving refined and targeted maintenance decision-making and execution.

[0077] In this embodiment, the preset maintenance levels include a preset normal level, a preset sub-health level, and a preset fault level. The maintenance recommendation corresponding to the preset normal level is to maintain routine monitoring and maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset sub-health level is to conduct key monitoring and preventive maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset fault level is to perform emergency treatment and repair or replacement of the vehicle's running gear.

[0078] Based on the failure evolution patterns, severity of failure impact, and characteristics of on-site operation and maintenance handling of key components of the running gear, and in accordance with the preset operation and maintenance level classification standards, corresponding operation and maintenance recommendations are configured for each level, as shown in Table 2: Table 2. Standards for Classification of Operation and Maintenance Levels

[0079] Specifically, the preset maintenance levels include a preset normal level, a preset sub-health level, and a preset fault level. The maintenance recommendation corresponding to the preset normal level is to maintain routine monitoring and maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset sub-health level is to conduct focused monitoring and preventative maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset fault level is to perform emergency handling and repair or replacement of the vehicle's running gear. These preset maintenance levels are determined based on the fault evolution patterns of key components of the vehicle's running gear, the severity of the fault's impact, and the characteristics of on-site maintenance handling. The preset normal level matches a target comprehensive health index range of 80-100. At this level, the performance indicators of the running gear components meet standards and show no obvious signs of degradation. Therefore, routine monitoring and maintenance recommendations are adopted to ensure the normal operating status of the equipment. The sub-health level is preset to match the target comprehensive health index range of 40-80. Under this level, the components have deteriorated to a certain extent, the performance indicators deviate from the standard, and the probability of failure increases. Therefore, the corresponding maintenance recommendation is to focus on monitoring the status of the components and carry out preventive maintenance to curb the deterioration trend in time. The fault level is preset to match the target comprehensive health index range of 0-40. Under this level, the components have shown obvious deterioration or failure and there are safety hazards. Therefore, it is necessary to carry out emergency handling and repair or replace the faulty components to eliminate operational safety risks. Each maintenance level and corresponding recommendation are highly adapted to the actual health status of the running gear, providing clear and targeted execution basis for on-site maintenance work.

[0080] The beneficial effects of this application are as follows: This application extracts the current feature values ​​corresponding to each target degradation feature from the collected current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; the current feature values ​​are nonlinearly mapped to obtain the target single feature health index of each target degradation feature; the target single feature dynamic weight is determined according to the target single feature contribution of each target degradation feature under the fault dimension of each component, and the target single feature health index is weighted and summed according to the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear; a target maintenance level matching the target comprehensive health index is determined from the preset maintenance levels, and the vehicle running gear is maintained according to the maintenance recommendations corresponding to the target maintenance level. Therefore, this application, by extracting the current feature value corresponding to the target degradation feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state from the current operating status data of the vehicle running gear, ensures that the extracted features can accurately reflect the complete degradation process of the component from normal to fault, avoiding misjudgment of health status due to insufficient correlation between features and degradation / fault. The target single-feature health index is obtained by nonlinearly mapping the current feature value, which can simulate the actual law of continuous decline in component health status, making the single-feature health index more closely match the actual health changes of the component and improving the accuracy of single-feature-level health assessment. The dynamic weight of the target single feature is determined based on the contribution of the target degradation feature to the target single-dimensional feature under each component fault dimension, enabling dynamic adjustment of the weight according to the feature degradation state, overcoming the limitation of traditional fixed weights that cannot adapt to changes in component status, making weight allocation more objective and highlighting the impact of high-risk features. The target single-feature health index is weighted based on the dynamic weight of the target single feature. The summation yields a target comprehensive health index, which integrates health information from multiple features to comprehensively reflect the overall health status of components. Furthermore, since feature extraction, nonlinear mapping, and dynamic weight determination all revolve around the correlation between component degradation and faults, the accuracy of the comprehensive health index in quantifying component health status is improved. The method matches the target maintenance level corresponding to the target comprehensive health index from preset maintenance levels and provides maintenance recommendations accordingly. This directly transforms the quantified health index into a targeted basis for maintenance decisions, making maintenance work more scientific, precise, and instructive. It effectively solves the problems of inaccurate indexes, difficulty in implementation, and difficulty in guiding maintenance in traditional rail transit health index calculations. Moreover, the method in this application does not rely on a large number of fault samples to achieve dynamic calculation of the health index, possessing good generalization ability and feature scalability. It can also provide effective support for research on health assessment and life prediction of vehicle running gear components, helping to achieve predictive maintenance of equipment, reduce the probability of fault occurrence, and ensure the stable and safe operation of vehicle running gear.

[0081] See Figure 2As shown in the figure, this application discloses a health monitoring device for a vehicle running gear, comprising: The feature acquisition module 11 is used to extract the current feature value corresponding to each target degradation feature from the current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; The index mapping module 12 is used to perform nonlinear mapping on the current feature value to obtain the target single feature health index of each target degradation feature; The health index quantification module 13 is used to determine the dynamic weight of the target single feature based on the contribution of each target degradation feature to the target single-dimensional feature under each component failure dimension, and to perform a weighted summation of the target single feature health index based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear. The operation and maintenance module 14 is used to determine the target operation and maintenance level that matches the target comprehensive health index from the preset operation and maintenance levels, and to perform operation and maintenance on the vehicle running gear according to the operation and maintenance recommendations corresponding to the target operation and maintenance level.

[0082] Furthermore, embodiments of this application also provide an electronic device. Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0083] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Specifically, it may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the vehicle running gear health monitoring method disclosed in any of the foregoing embodiments.

[0084] In this embodiment, the power supply 23 is used to provide operating voltage for various hardware devices on the electronic device; the communication interface 24 can create a data transmission channel between the electronic device and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0085] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0086] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0087] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. The operating system can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the vehicle running gear health monitoring method executed by the electronic device as disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0088] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned method for monitoring the health of the vehicle running gear. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0089] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0090] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. The software module may be located in random access memory (RAM), memory, read-only memory (ROM), electrically programmable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), register, hard disk, removable disk, CD-ROM (Compact Disc Read-Only Memory), or any other form of storage medium known in the art.

[0091] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0092] The above provides a detailed description of a method, apparatus, device, and medium for monitoring the health of a vehicle's running gear. Specific examples have been used to illustrate the principles and implementation methods of the invention. The descriptions of the above embodiments are only intended to help understand the method and core ideas of the invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the invention.

Claims

1. A method for health monitoring of a vehicle's running gear, characterized in that, include: Extract the current feature value corresponding to each target degradation feature from the current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; The current feature value is non-linearly mapped to obtain the target single feature health index of each target degradation feature; The dynamic weight of the target single feature is determined based on the contribution of each target degradation feature to the target single-dimensional feature under the fault dimension of each component, and the target single feature health index is weighted and summed based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear. A target maintenance level matching the target comprehensive health index is determined from the preset maintenance levels, and the vehicle running gear is maintained according to the maintenance recommendations corresponding to the target maintenance level. The step of performing a nonlinear mapping on the current feature value includes: The current feature value is nonlinearly mapped using a target arctangent model; wherein, the target arctangent model is the optimal mapping relationship between each target degradation feature and a single feature health index fitted based on the historical feature values ​​of the target degradation feature; The step of determining the dynamic weight of a single target feature based on the contribution of each target degradation feature to the single-dimensional feature of each component failure dimension includes: Linear regression is performed on the feature value sequence of the current target degradation feature within a second preset sliding window to obtain the fitting determination coefficient of the current target degradation feature under the trend stability dimension and the normalized trend slope under the risk rising trend dimension. The fitting determination coefficient and the normalized trend slope are respectively determined as the first target single-dimensional feature contribution and the second target single-dimensional feature contribution. The ratio of the current feature value of the current target degradation feature to the preset fault threshold is determined as the third target single-dimensional feature contribution of the current target degradation feature under the fault risk level dimension. Obtain the overall target contribution of each of the aforementioned target degradation features; the formula for obtaining the overall target contribution is: ； ； in, As a comprehensive contribution to the target, Contribution to the first objective's single-dimensional feature. Contribution to the single-dimensional feature of the second objective. Contribution to the single-dimensional features of the third objective. Preset importance under the component failure dimension; The overall contribution of each of the target degradation features is normalized to obtain the dynamic weight of each target single feature of the target degradation feature.

2. The health monitoring method for the running gear of a vehicle according to claim 1, characterized in that, Before extracting the current feature values ​​corresponding to each target degradation feature from the collected current operating status data of the vehicle running gear, the process also includes: Historical operating status data of the vehicle running gear is collected, and the initial degradation characteristics of each component in the vehicle running gear are determined based on the historical operating status data, so as to construct a basic feature library including each of the initial degradation characteristics; Target degradation features are selected from the basic feature library.

3. The method for health monitoring of the vehicle running gear according to claim 2, characterized in that, The historical operating status data of the vehicle's running gear collected includes: Collect historical operating status data of bearings, treads, and gears in the vehicle's running gear; wherein, the historical operating status data includes any one or more of the following types of data: sample data, impact dB data, impact SV data, temperature data, vibration data, and mileage data.

4. The method for health monitoring of the vehicle running gear according to claim 2, characterized in that, The step of determining the initial degradation characteristics of each component in the vehicle running gear based on the historical operating status data includes: Extract derived features of each component in the vehicle running gear from the historical operating status data; Time-domain and frequency-domain analyses are performed on the historical operating status data to obtain the time-domain and frequency-domain characteristics of each component in the vehicle running gear. The derived features, time-domain features, and frequency-domain features corresponding to the bearings, treads, and gears in the vehicle running gear are respectively determined as the initial degradation features of the bearings, the initial degradation features of the treads, and the initial degradation features of the gears.

5. The method for health monitoring of the vehicle running gear according to claim 2, characterized in that, The step of filtering target degradation features from the basic feature library includes: The degradation characterization capability of each initial degradation feature in the basic feature library for the component is evaluated, and target degradation features are selected from the basic feature library based on the degradation characterization capability.

6. The method for health monitoring of a vehicle running gear according to claim 5, characterized in that, The step of selecting target degradation features from the basic feature library based on the degradation characterization capability includes: Candidate degradation features that meet preset conditions are selected from the basic feature library; wherein, the preset conditions are that the candidate degradation features can effectively characterize the continuous degradation trend of the component and have clear physical meaning; The candidate degradation features are evaluated for predictability and / or subjected to chi-square tests. Based on the evaluation and / or test results, the target degradation features are selected from the candidate degradation features.

7. The method for health monitoring of a vehicle running gear according to claim 6, characterized in that, The predictability evaluation of the candidate degradation features includes: Determine the first mean of the candidate degradation feature at the initial time, the second mean at the failure time, and the standard deviation at the failure time; The absolute difference between the first mean and the second mean is determined as the degradation magnitude of the current candidate degradation feature, and the evaluation result of predictability evaluation is obtained based on the ratio between the standard deviation and the degradation magnitude.

8. The method for health monitoring of a vehicle running gear according to claim 6, characterized in that, Perform a chi-square test on the candidate degradation features, including: The frequency of each historical feature value of the current candidate degradation feature in a preset interval is counted to obtain the actual observation frequency of each historical feature value. Obtain the difference between the actual observation frequency and the corresponding theoretical observation frequency; The ratio of the square of the difference to the theoretical observation frequency is obtained, and the ratios of each of the historical feature values ​​are summed to obtain the chi-square test result.

9. The method for health monitoring of a vehicle running gear according to claim 6, characterized in that, The step of selecting target degradation features from the candidate degradation features based on the obtained evaluation results and / or test results includes: Target degradation features are selected from the candidate degradation features whose evaluation results are greater than a first preset threshold and / or whose test results are greater than a second preset threshold.

10. The method for health monitoring of a vehicle running gear according to claim 1, characterized in that, The step of performing a nonlinear mapping on the current feature value includes: The current feature value is processed to handle outliers and missing values, so as to obtain the processed current feature value; The processed current feature value is smoothed using a first preset sliding window to obtain a standardized current feature value; A nonlinear mapping is performed on the standardized current feature values.

11. The method for health monitoring of a vehicle running gear according to claim 1, characterized in that, Before performing the nonlinear mapping on the standardized current feature values, the method further includes: Construct an initial arctangent model with a single-feature health index as output and the feature values ​​of the target degradation feature as input; Construct an objective function; wherein the objective function aims to minimize the error between the single-feature health index output by the initial arctangent model and the standard single-feature health index; The objective function is optimized using the least squares method to solve for the amplitude coefficient, slope coefficient, offset coefficient, and baseline coefficient in the initial arctangent model, thereby obtaining the target arctangent model.

12. The method for health monitoring of a vehicle running gear according to claim 11, characterized in that, The objective function is: ； in, A single-feature health index, For the current eigenvalue, denoted as a, b as the amplitude coefficient, c as the slope coefficient, and d as the baseline coefficient. For the first There are 1 eigenvalues, where p is the total number of eigenvalues.

13. The method for health monitoring of a vehicle running gear according to claim 1, characterized in that, The normalization of the overall contribution of each of the target degradation features to obtain the dynamic weight of each target single feature of the target degradation feature includes: Obtain the sum of the target comprehensive contribution of each target degradation feature, and determine the ratio between the target comprehensive contribution of the current target degradation feature and the sum as the target single feature dynamic weight of the current target degradation feature.

14. The method for health monitoring of a vehicle running gear according to claim 1, characterized in that, The preset maintenance levels include a preset normal level, a preset sub-health level, and a preset fault level. The maintenance recommendation corresponding to the preset normal level is to maintain routine monitoring and maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset sub-health level is to conduct key monitoring and preventive maintenance of the vehicle's running gear. The maintenance recommendation corresponding to the preset fault level is to take emergency measures and repair or replace the vehicle's running gear.

15. A health monitoring device for a vehicle running gear, characterized in that, The steps for implementing the health monitoring method for the running gear of a vehicle as described in any one of claims 1 to 14 include: The feature acquisition module is used to extract the current feature value corresponding to each target degradation feature from the current operating status data of the vehicle running gear; wherein, the target degradation feature is a feature that characterizes the degradation trend of the component throughout its entire life cycle and is strongly correlated with the fault state; The index mapping module is used to perform nonlinear mapping on the current feature value to obtain the target single feature health index of each target degradation feature; The health index quantification module is used to determine the dynamic weight of the target single feature based on the contribution of each target degradation feature to the target single-dimensional feature under each component failure dimension, and to perform a weighted summation of the target single feature health index based on the target single feature dynamic weight to obtain the target comprehensive health index used to quantify the health status of the vehicle running gear. The operation and maintenance module is used to determine the target operation and maintenance level that matches the target comprehensive health index from the preset operation and maintenance levels, and to perform operation and maintenance on the vehicle running gear according to the operation and maintenance recommendations corresponding to the target operation and maintenance level.

16. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the health monitoring method for the running gear of a vehicle as described in any one of claims 1 to 14.

17. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the health monitoring method for the running gear of a vehicle as described in any one of claims 1 to 14.

Images