A multi-bearing synchronous electrochemical corrosion analysis method and system for an electrified train

Abstract

This invention provides a method and system for synchronous electrochemical corrosion analysis of multiple bearings in electrified trains, relating to the field of rail transit technology. The method includes acquiring electrochemical corrosion data of bearings under a preset number of grease types; filtering all the electrochemical corrosion data using Kalman filtering to obtain filtered electrochemical corrosion data; inputting the filtered electrochemical corrosion data into a preset mutation detection model for mutation detection to obtain mutation detection data of the electrochemical corrosion data; and inputting the mutation detection data of the electrochemical corrosion data into a finite element model for analysis and evaluation to obtain electrochemical corrosion evaluation information for all bearings. This method can dynamically capture corrosion mutation events, improve detection accuracy, and achieve a realistic reconstruction of complex electrochemical corrosion processes through multi-scale fusion and coupled modeling.

Description

Technical Field

[0001] This invention relates to the field of rail transit technology, and more specifically, to a method and system for analyzing the synchronous electrochemical corrosion of multiple bearings in electrified trains. Background Technology

[0002] With the continuous expansion of the high-speed railway network, the safe operation of high-speed trains has received increasing attention. Traction motor bearings, as one of the core components of high-speed trains, are crucial for the smooth and safe operation of the train due to their proper functioning. In recent years, premature bearing failures due to electrochemical corrosion have become increasingly common, accounting for over 30% of cases. This electrochemical corrosion not only threatens the safe operation of trains but also leads to frequent bearing replacements, increasing operating costs.

[0003] Currently, domestic and international research on bearings mainly focuses on common-mode voltage and rotor grounding current. There is a lack of systematic analysis on the impact of typical transient conditions such as train phase transitions on traction motor bearings. The impact of transient conditions such as frequent overvoltage and overcurrent impacts that may occur during train phase transitions on bearings still needs further in-depth research in order to more comprehensively understand and solve the problem of bearing electro-corrosion.

[0004] Therefore, there is an urgent need for a multi-bearing synchronous electrochemical corrosion analysis method and system for electrified trains to solve the above-mentioned technical problems. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for synchronous electrochemical corrosion analysis of multiple bearings in electrified trains, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0006] In a first aspect, this application provides a method for analyzing the synchronous electrochemical corrosion of multiple bearings in electrified trains, including:

[0007] Obtain electrochemical corrosion data of bearings for a preset number of grease types;

[0008] All the electrochemical corrosion data are filtered using Kalman filtering to obtain filtered electrochemical corrosion data.

[0009] The filtered electrochemical corrosion data is input into a preset mutation detection model for mutation detection to obtain mutation detection data of electrochemical corrosion data.

[0010] The abrupt change detection data of the electrochemical corrosion data is input into the finite element model for analysis and evaluation to obtain the electrochemical corrosion evaluation information of all bearings.

[0011] Secondly, this application also provides a multi-bearing synchronous electrochemical corrosion analysis system for electrified trains, comprising:

[0012] The acquisition unit is used to acquire electrochemical corrosion data of bearings under a preset number of grease types;

[0013] The filtering unit is used to filter all the electrochemical corrosion data based on Kalman filtering to obtain filtered electrochemical corrosion data.

[0014] The detection unit is used to input the filtered electrochemical corrosion data into a preset mutation detection model for mutation detection, and obtain mutation detection data of the electrochemical corrosion data;

[0015] The evaluation unit is used to input the abrupt change detection data of the electrochemical corrosion data into the finite element model for analysis and evaluation, so as to obtain the electrochemical corrosion evaluation information of all bearings.

[0016] The beneficial effects of this invention are as follows:

[0017] This invention achieves precise identification and feature extraction of abrupt changes by acquiring multi-source data on the electrochemical corrosion of bearings under various grease types and combining Kalman filtering, Bayesian mutation detection, and multi-algorithm fusion fitting. Furthermore, it constructs a high-gradient corrosion distribution map based on two-dimensional equivalent damage modeling, spectral clustering graph embedding, and adaptive finite element mesh reconstruction, enabling coupled modeling and finite element evaluation of corrosion evolution in multiple bearings. Compared to existing technologies, this invention not only dynamically captures corrosion mutation events and improves detection accuracy but also achieves a realistic reconstruction of complex electrochemical corrosion processes through multi-scale fusion and coupled modeling, thus providing more reliable corrosion assessment results under the operating conditions of electrified trains.

[0018] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1This is a schematic diagram of the multi-bearing synchronous electrochemical corrosion analysis method for electrified trains described in this embodiment of the invention.

[0021] Figure 2 This is a schematic diagram of the multi-bearing synchronous electrochemical corrosion analysis system for electrified trains as described in this embodiment of the invention.

[0022] In the diagram: 701, acquisition unit; 702, filtering unit; 703, detection unit; 704, evaluation unit. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] Example 1:

[0026] This embodiment provides a method for analyzing the synchronous electrochemical corrosion of multiple bearings in electrified trains.

[0027] See Figure 1 The figure shows that the method includes steps S1, S2, S3 and S4.

[0028] Step S1: Obtain electrochemical corrosion data of bearings for a preset number of different types of grease;

[0029] Understandably, this step aims to achieve high-precision modeling of the electrochemical corrosion behavior of multi-bearing systems in electrified trains under different grease conditions. The first step involves acquiring electrochemical corrosion data under various grease conditions. This process ensures that bearings with different greases operate under the same mechanical load and electrical environment by controlling the train's servo motor to maintain a stable speed, thus eliminating unwanted variable interference introduced by speed fluctuations. Simultaneously, by setting an initial overvoltage and applying an external electric field in increments of a set step size, the gradually increasing electric field disturbance effect during actual train operation is simulated, thereby stimulating the realistic evolution of electrochemical corrosion behavior in the bearings. This step also automates the uploading of data to a central storage platform, enabling data grouping, classification, and structured storage under different grease conditions. This provides a complete, accurate, and traceable raw data source for subsequent filtering and mutation detection. This method significantly differs from traditional methods that only monitor a single lubrication condition or a single physical quantity, offering greater engineering adaptability and data representativeness. The technical benefits of this step are reflected in: providing a unified experimental platform and consistent boundary conditions for the comparative analysis of corrosion behavior of multiple bearings, constructing a highly reliable and multi-factor controllable original dataset, and fundamentally improving the modeling accuracy and evaluation reliability of subsequent algorithms for corrosion mechanisms.

[0030] In this step, step S1 includes steps S11 and S12.

[0031] Step S11: Control the servo motor of the electrified train to keep its speed constant, apply different types of grease to a preset number of bearings, set an initial overvoltage, increase the overvoltage value by setting a step size, and monitor the current through the bearing and the bearing temperature parameters in real time.

[0032] Understandably, this step first controls the speed of the electrified train's servo motor to remain constant. This operation ensures consistency in the mechanical contact state, avoiding changes in frictional heat or contact resistance caused by speed fluctuations, thus ensuring that the current and temperature response are only affected by the type of grease and voltage disturbances. Next, different types of grease are injected into multiple test bearings. These greases differ significantly in chemical composition, conductivity, polar additives, and thermal stability, which are key variables affecting electrochemical corrosion behavior.

[0033] Subsequently, by applying an initial overvoltage and gradually increasing the applied voltage value in a set step increment, this operation simulates complex electrical disturbance scenarios that may occur during the operation of an electrified train, such as stray current surges or grounding failures. This voltage increment mechanism can not only effectively induce the initial corrosion state of the bearing system in a micro-electrochemical environment, but also dynamically capture the critical point and acceleration stage of its corrosion development process. Based on this, a high-frequency response current sensor is used to monitor the current changes passing through the bearing in real time, while an infrared thermometer and embedded thermocouples are used to simultaneously collect temperature information of the bearing housing and rolling elements. These current and temperature parameters together constitute multi-dimensional time-series response data, reflecting both the protective performance of the grease in an electrochemical environment and revealing its stability and degradation trend under thermo-electric coupling.

[0034] Step S12: Upload and store the current and temperature parameters of the bearing as electrochemical corrosion data of the bearing under a preset number of grease types.

[0035] Understandably, this step involves synchronous time calibration of the current and temperature signals acquired by the high-precision acquisition module for multiple bearings using different greases. This ensures that the temperature change and current response at each point in time correspond to a specific time series. This step also performs preliminary standardization and normalization on the signals to remove background noise caused by non-target factors such as environmental fluctuations and equipment vibration, thereby improving the purity and discriminability of the data.

[0036] Subsequently, this structured current-temperature joint data will be uniformly encoded into electrochemical corrosion response vector sets and tagged according to the type of lubricating grease, facilitating comparative analysis under multiple bearings and various lubrication conditions. Simultaneously, this tagged electrochemical corrosion data will be uploaded to a central database or industrial cloud platform for unified storage via edge computing modules or industrial data acquisition terminals (such as PLCs or industrial PCs).

[0037] Step S2: Filter all the electrochemical corrosion data based on Kalman filtering to obtain filtered electrochemical corrosion data;

[0038] Understandably, because different greases correspond to different chemical properties and current response modes, Kalman filtering can effectively compare and eliminate the periodic fluctuations and local jumps introduced by different electrochemical reaction rates under various greases, preserving the overall corrosion trend, thereby enhancing data consistency and analytical versatility. In this step, step S2 includes steps S21, S22, and S23.

[0039] Step S21: Set the initial state and covariance matrix of the Kalman filter, and predict the state and covariance of the next preset time step based on the initial state and covariance matrix;

[0040] It is understandable that this step constructs the basic prediction model for the entire electrochemical corrosion data filtering process by setting the initial state and covariance matrix of the Kalman filter. This process plays a decisive role in the accuracy and stability of the final filtering result. The initial state is typically estimated based on the statistical characteristics of the current and temperature parameters collected in the first few time steps. Combined with the physicochemical properties of different lubricants, the potential initial corrosion response trend is derived, using parameters such as steady-state current and temperature rise initiation rate as reference values ​​for the initial state vector. Meanwhile, the setting of the covariance matrix reflects the uncertainty of the system state and the influence of observation errors on the estimation. It requires comprehensive consideration of practical factors such as sensor measurement accuracy, fluctuations in the operating conditions of electrified trains, and differences in mechanical assembly between bearings. The formula for the covariance matrix in this step is shown below:

[0041] ;

[0042] ;

[0043] in, Yes The predicted value of the state at time step. yes The state covariance matrix at time t, It is the state transition matrix. Yes The predicted value of the state at time step. It is a prediction The state covariance matrix at time t, It is a control input matrix. It is a control input. It is the process noise covariance matrix. It is the transpose of the state transition matrix.

[0044] Step S22: Input the electrochemical corrosion data into the preset formula for the covariance matrix of the updated state estimation for processing, and obtain the updated state estimate and covariance for each preset time step;

[0045] It is understandable that in this step, by inputting the actually collected electrochemical corrosion data into the preset state update formula, the state estimate obtained from the previous time step is dynamically corrected based on the observation update stage using Kalman filtering, thereby obtaining the updated state estimate and covariance matrix for each preset time step. The formula for the updated state estimate covariance matrix is ​​shown below:

[0046] ;

[0047] ;

[0048] ;

[0049] in, It is Kalman gain. yes The state covariance matrix at time t, It is a measurement matrix. Transpose of the measurement matrix It measures the noise covariance matrix. yes The optimal state estimate at time t. Yes The predicted value of the state at time step. These are actual measured values. It is the updated state covariance matrix. It is the identity matrix. yes The state covariance matrix at time t.

[0050] Step S23: Use the updated state estimate and covariance of each preset time step as the filtered electrochemical corrosion data.

[0051] Understandably, this step not only outputs numerical results but also provides probabilistic interpretation for subsequent models. For example, covariance information can be used as a prior uncertainty parameter input in subsequent Bayesian mutation detection models, thereby improving the sensitivity and accuracy of mutation identification. Simultaneously, through the overall processing of the time-step sequence, a structurally stable and clearly trending corrosion change trajectory can be constructed in the data stream, providing continuous and stable boundaries and initial conditions for subsequent corrosion mutation identification, regional distribution field construction, and finite element solution.

[0052] Step S3: Input the filtered electrochemical corrosion data into a preset mutation detection model to perform mutation detection, and obtain mutation detection data of electrochemical corrosion data;

[0053] Understandably, this step, through accurate mutation detection, can not only promptly identify electrochemical corrosion mutations that may affect the safe operation of electrified train bearings, but also provide anomaly markers for subsequent analysis. Compared to traditional methods, this mutation detection algorithm has higher sensitivity and a lower false alarm rate, effectively identifying minute but critical mutation phenomena, thereby reducing potential risks caused by delayed or missed detections. Simultaneously, by combining the advantages of different algorithms, the model effectively suppresses noise interference with the detection results, enhancing the reliability and accuracy of the overall analysis system and providing a scientific basis for subsequent corrosion assessment and decision support.

[0054] In this step, step S3 includes steps S31, S32, S33 and S34.

[0055] Step S31: Construct a Bayesian model based on the filtered electrochemical corrosion data, and calculate the posterior probability distribution of the running length for each preset time step based on the filtered electrochemical corrosion data.

[0056] Understandably, this step first involves building a model using a Bayesian network. This model can handle uncertainties in time-series data and dynamically update the estimate of the corrosion process. The Bayesian model, by utilizing prior knowledge and historical data, combined with the current filtered electrochemical corrosion data, calculates the posterior probability distribution for each time step. This process not only considers the regular changes in electrochemical corrosion but also incorporates abrupt changes and abnormal fluctuations into the probability calculation, effectively capturing drastic changes or sudden events that may occur in the system.

[0057] This step first assumes that the current and temperature data follow a Gaussian distribution before and after the abrupt change point, as shown in the following formula:

[0058] ;

[0059] in, Is a given parameter Data The likelihood function, It is a data vector of observed current and temperature. It is the total number of data points. and These are the mean and variance, respectively. This represents the probability density function of a Gaussian distribution. Indicates the first One observation data,

[0060] Secondly, we define a prior distribution, and the formula for the prior probability distribution of the mutation point is as follows:

[0061] ;

[0062] in, It is the prior probability of the mutation point. It is the prior probability of the mutation point occurring.

[0063] Step S32: Calculate the probability of a mutation occurring at each time step based on the posterior probability distribution of the running length of each preset time step, and determine the mutation time based on the probability of a mutation occurring at each time step to obtain the first mutation detection result.

[0064] Understandably, the formula for calculating the probability of a sudden change at each time step in this step is as follows:

[0065] ;

[0066] in, It is data The marginal probability, It is data In parameters The likelihood function under the given conditions, At a certain point in time The posterior probability of a mutation point occurring, given the observed data. , It is the prior probability of the mutation point occurring.

[0067] like If a threshold is set (preset), then a mutation point is considered to exist at that time step.

[0068] Step S33: Perform least squares fitting based on the filtered electrochemical corrosion data, calculate the slope of the fitted line, and obtain the second mutation detection result based on the slope of the fitted line;

[0069] Understandably, this step uses the least squares fitting method to obtain a straight line, from which the slope of the fitted line can be calculated. The slope of the fitted line reflects the rate of change in electrochemical corrosion, that is, the dynamic change in the corrosion rate. If the slope changes significantly at a certain moment, it usually indicates an abrupt change in the corrosion process. A sharp change in the slope may mean that the corrosion process has shifted from slow development to faster accelerated corrosion, or it may indicate an abrupt change in the corrosion mode. In this step, the least squares method is used to calculate the slope of the fitted line, where the slope formula is:

[0070] ;

[0071] in, This represents the slope of the fitted line. This indicates the number of current and temperature data points used to fit the straight line. This represents the filtered current data. This represents the filtered temperature data.

[0072] Step S34: Input the first mutation detection result and the second mutation detection result into the judgment model to obtain the mutation detection data of the final electrochemical corrosion data.

[0073] Understandably, in this step, fusing detection results from different algorithms can reduce the potential for false positives or false negatives that might arise from a single method. For example, Bayesian models can capture more complex mutation signals through their probability distributions, while least squares fitting can more directly reflect the trend changes of mutations. By combining the two, the model can provide more accurate and reliable mutation detection results. In this step, step S34 includes steps S341, S342, S343, and S344.

[0074] Step S341: Perform joint sparse coding processing based on the first mutation detection result and the second mutation detection result to obtain the fusion coefficient matrix representing the two detection results;

[0075] Understandably, this step first uses the first mutation detection result (the posterior probability distribution calculated by the Bayesian model) and the second mutation detection result (the slope change obtained through least-squares fitting) as input data, representing the probability and trend of mutation occurrence, respectively. These results are treated as vector or matrix data, representing the state at each preset time step. Then, based on these two sets of data, a sparse coding model is used to find a set of basis vectors such that each input data (i.e., the two detection results) can be represented as a linear combination of these basis vectors, and the coefficients of the combination have sparse properties, i.e., most coefficients are zero, and only a few basis vectors contribute significantly to the representation. In this way, sparse coding not only compresses the amount of data but also ensures that only the most important features are retained and redundant information is removed. Finally, by simultaneously performing sparse coding on the first and second mutation detection results, a fusion coefficient matrix can be generated. This matrix represents a linear combination of the different results of the two mutation detection methods on the sparse basis, effectively combining their respective advantages. This matrix can capture the correlation between the two detection results and retain the key information of both in a compressed form.

[0076] Step S342: Perform structural similarity measurement processing based on the fusion coefficient matrix, and obtain the coupling similarity weight vector by calculating the relevant structural indices between the mutation time series.

[0077] Understandably, this step allows us to analyze the temporal structure and trends of the results by using the basis vectors (i.e., the fused detection results) contained in the fusion coefficient matrix. For example, two mutation detection results may exhibit consistent mutation patterns at certain times, while differing at other times. Structural similarity metrics can quantify this consistency or difference, thereby establishing relevant structural indices. Structural similarity metrics calculate the local and global similarity between data sequences, including factors such as the temporal location of mutation points and the trend of magnitude changes in the time series. Pearson correlation coefficients are used to establish relevant structural indices for mutation time series. These indices can reveal the temporal consistency or asynchrony of results from different detection methods. Based on the results of the structural similarity metrics, a "coupling similarity weight vector" is calculated, representing the temporal similarity between two mutation detection methods. In this vector, each component represents the similarity weight between the two detection results at the corresponding time step. In this way, the coupling between mutation time series can be accurately captured, providing important basis for subsequent decision-making.

[0078] The coupling similarity weight vector is calculated by taking a weighted average of the similarities at each time step obtained from the Pearson correlation coefficient to obtain the global similarity, and then regularizing the global similarity to obtain the coupling similarity weight vector.

[0079] Step S343: Perform nonlinear feature reconstruction processing based on the coupling similarity weight vector to obtain a multi-scale mutation feature vector group;

[0080] Understandably, this step uses the coupled similarity weight vector as input and processes it through a local linear embedding method to obtain a set of features. These features are then used to reconstruct the deeper characteristics of the electrochemical corrosion data, reflecting multi-scale corrosion patterns. Next, the electrochemical corrosion data is weighted based on the coupled similarity weight vector, strengthening the parts with stronger similarities. This helps to highlight key features of corrosion behavior from the data and suppress the influence of noise. The weighted data is then input into a nonlinear mapping model, where a high-dimensional feature vector is obtained through nonlinear transformation. These vectors will reveal the potential complex patterns and abrupt changes in the electrochemical corrosion data.

[0081] Next, wavelet transform is used to decompose the nonlinear features and extract corrosion features within different frequency ranges. Wavelet transform can analyze data simultaneously in the time and frequency domains, effectively identifying abrupt changes at different scales. Self-similarity features in the data are extracted through fractal analysis. Abrupt changes in the electrochemical corrosion process usually manifest as drastic local variations; fractal theory can help capture the multi-scale features of these local changes, thus obtaining a multi-scale abrupt change feature vector set.

[0082] Step S344: Perform decision discrimination processing based on the multi-scale mutation feature vector group to obtain the mutation detection data of the final electrochemical corrosion data.

[0083] It is understandable that the multi-scale mutation feature vector sets in this step have already been obtained through the previous steps. These vector sets contain the variation characteristics of electrochemical corrosion data at different scales obtained through nonlinear feature reconstruction. Each feature vector represents a mutation mode at a specific scale, and these modes may correspond to different corrosion degrees, environmental factors, or changes in operating conditions.

[0084] The first step in the decision-making process is to perform a weighted fusion of these feature vectors. This weighting typically depends on the contribution of the feature at each scale. For example, features at higher scales may represent more significant abrupt changes, while features at lower scales may reflect minor variations. Weighted fusion assigns a weight value to each feature based on its reliability and importance, ultimately synthesizing a global discriminant value.

[0085] Next, this step uses a support vector machine (SVM) to train and infer these weighted fused feature vectors as input. Based on the training dataset, the model learns how to associate the input feature vectors with abrupt events (such as a sudden increase in corrosion rate). Specifically, the training phase involves optimizing the model's classification performance using labeled electrochemical corrosion data and markers indicating abrupt changes. Through this decision-making process, the final judgment determines whether abrupt changes have occurred in the electrochemical corrosion data at each time step. If abrupt changes occur, the data point is marked as "abrupt" and further analysis or an alarm is triggered; otherwise, the data is considered a normal change.

[0086] Step S4: Input the abrupt change detection data of the electrochemical corrosion data into the finite element model for analysis and evaluation to obtain the electrochemical corrosion evaluation information of all bearings.

[0087] It is understandable that this step, by introducing a physical simulation mechanism and deeply integrating mutation detection data, achieves a closed-loop evaluation from the data layer to the structural performance layer, significantly improving the interpretability and evaluation accuracy of corrosion data, and extending corrosion detection from surface signal identification to quantitative inference of actual structural damage. In this step, step S4 includes steps S41, S42, S43 and S44.

[0088] Step S41: Based on the mutation detection data, construct two-dimensional equivalent damage parameters to obtain the initial distribution field of time-space coupled corrosion;

[0089] Understandably, this step first performs time-series analysis on the mutation detection data, extracting characteristic indicators such as mutation intensity, mutation duration, and the slope of current and temperature changes before and after the mutation at each time point. Then, based on the bearing geometry and the specific installation location of the bearing in the electrified train, spatial mapping is performed to associate the temporal mutation event with the specific contact area or rolling path on the bearing model.

[0090] Next, a two-dimensional equivalent damage parameter construction method is introduced: combining mining damage theory or damage models based on cumulative energy dissipation, the corrosion intensity, duration, and temperature rise at each time point are converted into equivalent damage coefficients reflecting the degree of local material performance degradation. Here, a nonlinear mapping model based on the temperature-current synergistic factor can be used to calculate the weights and iterate cumulatively for each abrupt change event, ultimately constructing an initial corrosion distribution map on the two-dimensional bearing plane, i.e., a two-dimensional equivalent damage field. To enhance the coupling between temporal and spatial distribution, time-location synergistic interpolation based on Gaussian processes is used to fill in missing regions, ensuring that the generated initial damage distribution field reflects both the abrupt location of corrosion occurrence and the continuity of its evolution path and the propagation of intensity.

[0091] Step S42: Perform regional feature decomposition processing based on the initial corrosion distribution field, and extract the active corrosion region by spectral clustering graph embedding method to obtain a high-gradient corrosion aggregation region map.

[0092] This step, as is understandable, first treats each spatial node in the initial corrosion distribution field (such as a discrete element in a bearing mesh) as a vertex in a graph structure, using its corrosion intensity as the node weight. Simultaneously, an edge weight matrix is ​​constructed based on the spatial adjacency relationships between nodes and the differences in corrosion values, forming a weighted undirected graph. The edge weights in the graph characterize the "similarity" or "coupling" of two regions in their corrosion development trends.

[0093] Next, the Laplacian matrix factorization method in spectral clustering is applied to embed the graph structure. By calculating the eigenvalues ​​and eigenvectors of the normalized Laplacian matrix of the graph, the corrosion distribution structure is mapped from the original high-dimensional space to the low-dimensional embedding space, allowing locally similar regions in the corrosion pattern to exhibit clustering in the low-dimensional space. Subsequently, the K-means clustering algorithm partitions the nodes in the embedding space, achieving unsupervised region identification in the corrosion space. To enhance the response to corrosion abrupt gradients, the corrosion intensity gradient is introduced as a penalty factor when constructing edge weights, resulting in edges with significant gradient changes gaining higher clustering sensitivity in spectral decomposition, thus highlighting active corrosion boundaries. The final result is a "corrosion high-gradient clustering region map," where each clustered region represents a local area with highly dynamic characteristics during corrosion evolution.

[0094] Step S43: Based on the corrosion high gradient accumulation region map, perform mesh reconstruction processing to obtain the adaptive corrosion stress field mesh structure;

[0095] Understandably, this step extracts the high-gradient corrosion clustering region map as the key spatial object for mesh reconstruction. This map, through spectral clustering, divides the entire bearing material region into multiple sub-regions with significantly different corrosion behaviors; each sub-region is considered a "hotspot" with localized corrosion characteristics. Subsequently, based on the gradient intensity and spatial scale information of each region in the map, a mesh refinement strategy is set. In regions with high corrosion gradients (i.e., regions with drastic corrosion changes and frequent abrupt changes), unstructured fine meshes are used for modeling to ensure the model can capture detailed features such as micro-stress concentration and crack initiation; while in regions with low corrosion gradients and stable changes, relatively coarse meshes are used, thereby effectively reducing the computational burden of global solution.

[0096] In the specific mesh generation process, an h-adaptive refinement method based on error estimation is used, with corrosion intensity gradient and region boundary as the dominant factors for mesh generation, guiding the dynamic adjustment of mesh element size and shape. Simultaneously, to avoid boundary effects caused by mesh discontinuities, a transition zone design strategy is introduced, automatically generating a transition layer mesh in the boundary region between coarse and fine meshes, ensuring the continuity of physical quantity transfer and the stability of finite element calculations.

[0097] Step S44: Perform multi-field coupled finite element analysis based on the mesh structure to obtain electrochemical corrosion assessment information for all bearings.

[0098] Understandably, the multiphysics coupling model constructed in this step commonly includes, but is not limited to, the following types of fields: ① Electric field – used to describe the path and intensity distribution of current flow in the bearing under overvoltage; ② Temperature field – simulating the heat accumulation caused by current flow and friction and its spatiotemporal evolution; ③ Stress field – characterizing the mechanical response of bearing materials to temperature difference, corrosion damage, and structural loads; ④ Corrosion field – representing the evolution trend of material microstructure during electrochemical corrosion. These physical fields interact and drive each other in the model through a coupling mechanism, forming a dynamic joint simulation of corrosion behavior. For example, a local increase in electric field strength may cause a sudden change in current density, leading to increased heat and accelerated local corrosion rates. This chain reaction needs to be solved uniformly through a coupling model.

[0099] The specific operation involves using finite element analysis software (such as COMSOL Multiphysics, ANSYS, etc.) to map all field variables into an adaptive mesh structure and setting up a coupled solution module. The relationship between the electric field and the corrosion kinetic equations is typically established through the Butler-Wolmer equations, while the interaction between temperature and material properties is represented by a heat conduction-elastoplastic-corrosion degradation model. As the time step progresses, the model tracks the changes in field variables within each element in real time, dynamically reflecting the expansion behavior of the corrosion zone and the stress concentration trend.

[0100] Finally, the finite element solver outputs key electrochemical corrosion indicators for each bearing throughout its entire life cycle, including local corrosion rate, corrosion depth distribution map, and corrosion-induced thermal stress concentration factor. Based on the key indicators and preset thresholds, the corrosion degree of each bearing is classified into severe corrosion, ordinary corrosion, and slight corrosion.

[0101] Example 2:

[0102] like Figure 2 As shown, this embodiment provides a multi-bearing synchronous electrochemical corrosion analysis system for electrified trains. (See also...) Figure 2 The system includes an acquisition unit 701, a filtering unit 702, a detection unit 703, and an evaluation unit 704.

[0103] Acquisition unit 701 is used to acquire electrochemical corrosion data of bearings under a preset number of grease types;

[0104] The filtering unit 702 is used to filter all the electrochemical corrosion data based on Kalman filtering to obtain filtered electrochemical corrosion data.

[0105] The detection unit 703 is used to input the filtered electrochemical corrosion data into a preset mutation detection model for mutation detection, and obtain mutation detection data of the electrochemical corrosion data.

[0106] Evaluation unit 704 is used to input the abrupt change detection data of the electrochemical corrosion data into the finite element model for analysis and evaluation, so as to obtain the electrochemical corrosion evaluation information of all bearings.

[0107] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0108] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for synchronous electrochemical corrosion analysis of multiple bearings in an electrified train, characterized in that, include: Obtain electrochemical corrosion data of bearings for a preset number of grease types; All the electrochemical corrosion data are filtered using Kalman filtering to obtain filtered electrochemical corrosion data. The filtered electrochemical corrosion data is input into a preset mutation detection model for mutation detection to obtain mutation detection data of electrochemical corrosion data. The abrupt change detection data of the electrochemical corrosion data were input into the finite element model for analysis and evaluation to obtain the electrochemical corrosion evaluation information of all bearings; The process of inputting the filtered electrochemical corrosion data into a preset mutation detection model for mutation detection includes: A Bayesian model is constructed based on the filtered electrochemical corrosion data, and the posterior probability distribution of the running length for each preset time step is calculated based on the filtered electrochemical corrosion data. The probability of a mutation occurring at each time step is calculated based on the posterior probability distribution of the running length of each preset time step, and the mutation time is determined based on the probability of a mutation occurring at each time step to obtain the first mutation detection result. Based on the filtered electrochemical corrosion data, least squares fitting is performed, and the slope of the fitted line is calculated. The second mutation detection result is obtained based on the slope of the fitted line. The first mutation detection result and the second mutation detection result are input into the judgment model to obtain the mutation detection data of the final electrochemical corrosion data; The process includes performing least-squares fitting based on the filtered electrochemical corrosion data and calculating the slope of the fitted line, including: Based on the first mutation detection result and the second mutation detection result, joint sparse coding processing is performed to obtain the fusion coefficient matrix representing the two detection results; Based on the fusion coefficient matrix, structural similarity measurement is performed, and the coupling similarity weight vector is obtained by calculating the relevant structural indices between mutation time series. Nonlinear feature reconstruction is performed based on the coupled similarity weight vector to obtain a multi-scale mutation feature vector set; The decision-making and discrimination processing is performed based on the multi-scale mutation feature vector group to obtain the mutation detection data of the final electrochemical corrosion data.

2. The method for analyzing the synchronous electrochemical corrosion of multiple bearings in electrified trains according to claim 1, characterized in that... Obtain electrochemical corrosion data of bearings for a preset number of grease types, including: The servo motor of the electrified train is kept at a constant speed. Different types of grease are used on a preset number of bearings. An initial overvoltage is set, and the overvoltage value is increased by a set step size. The current passing through the bearing and the bearing temperature parameters are monitored in real time. The current and temperature parameters of the bearing are uploaded and stored as electrochemical corrosion data of the bearing under a preset number of grease types.

3. The method for analyzing the synchronous electrochemical corrosion of multiple bearings in electrified trains according to claim 1, characterized in that... All the electrochemical corrosion data are filtered using Kalman filtering to obtain filtered electrochemical corrosion data, including: Set the initial state and covariance matrix of the Kalman filter, and predict the state and covariance of the next preset time step based on the initial state and covariance matrix; Electrochemical corrosion data are input into a preset formula for updating the covariance matrix of the state estimate and processed to obtain the updated state estimate and covariance at each preset time step. The updated state estimate and covariance at each preset time step are used as the filtered electrochemical corrosion data.

4. A multi-bearing synchronous electrochemical corrosion analysis system for electrified trains, characterized in that, include: The acquisition unit is used to acquire electrochemical corrosion data of bearings under a preset number of grease types; The filtering unit is used to filter all the electrochemical corrosion data based on Kalman filtering to obtain filtered electrochemical corrosion data. The detection unit is used to input the filtered electrochemical corrosion data into a preset mutation detection model for mutation detection, and obtain mutation detection data of the electrochemical corrosion data; The evaluation unit is used to input the abrupt change detection data of the electrochemical corrosion data into the finite element model for analysis and evaluation, and to obtain the electrochemical corrosion evaluation information of all bearings. The detection unit includes: The first detection subunit is used to construct a Bayesian model based on the filtered electrochemical corrosion data, and to calculate the posterior probability distribution of the running length of each preset time step based on the filtered electrochemical corrosion data. The second detection subunit is used to calculate the probability of a mutation occurring at each time step based on the posterior probability distribution of the running length of each preset time step, and to determine the mutation time based on the probability of a mutation occurring at each time step, thereby obtaining the first mutation detection result. The third detection subunit is used to perform least squares fitting based on the filtered electrochemical corrosion data, calculate the slope of the fitted line, and obtain the second mutation detection result based on the slope of the fitted line. The fourth detection subunit is used to input the first mutation detection result and the second mutation detection result into the judgment model to obtain the mutation detection data of the final electrochemical corrosion data; The fourth detection subunit includes: The fifth detection subunit is used to perform joint sparse coding processing based on the first mutation detection result and the second mutation detection result to obtain a fusion coefficient matrix representing the two detection results; The sixth detection subunit is used to perform structural similarity measurement processing based on the fusion coefficient matrix, and obtains the coupling similarity weight vector by calculating the relevant structural indicators between the mutation time series. The seventh detection subunit is used to perform nonlinear feature reconstruction processing based on the coupling similarity weight vector to obtain a multi-scale mutation feature vector set; The eighth detection subunit is used to perform decision discrimination processing based on the multi-scale mutation feature vector group to obtain the mutation detection data of the final electrochemical corrosion data.

5. The multi-bearing synchronous electrochemical corrosion analysis system for electrified trains according to claim 4, characterized in that, The acquisition unit includes: The first acquisition subunit is used to control the servo motor of the electrified train to keep the speed constant, apply different types of grease to a preset number of bearings, set an initial overvoltage, increase the overvoltage value by a set step size, and monitor the current passing through the bearing and the bearing temperature parameters in real time. The second acquisition subunit is used to upload and store the current passing through the bearing and the bearing temperature parameters as electrochemical corrosion data of the bearing under a preset number of grease types.

6. The multi-bearing synchronous electrochemical corrosion analysis system for electrified trains according to claim 4, characterized in that, The filtering unit includes: The first filtering subunit is used to set the initial state and covariance matrix of the Kalman filter, and predict the state and covariance of the next preset time step based on the initial state and covariance matrix. The second filtering subunit is used to input the electrochemical corrosion data into the preset formula for the covariance matrix of the updated state estimate for processing, so as to obtain the updated state estimate and covariance at each preset time step. The third filtering subunit is used to take the updated state estimate and covariance of each preset time step as the filtered electrochemical corrosion data.

Images