Method for judging virtual and real consistency of brake shoe of brake

Abstract

The invention discloses a method for judging the virtual-real consistency of a brake shoe of a brake. The method comprises the following steps: arranging a sensor near the brake shoe; acquiring brake shoe stress data by using a sensor; establishing a geometric model of the brake shoe, discretizing the geometric structure of the brake shoe through finite element analysis, and establishing boundary constraints; adding a load condition to the established geometric model of the brake shoe, and solving stress distribution of the brake shoe under different working conditions through finite elements; performing inversion on the constructed brake shoe model by using the collected brake shoe stress data; and judging the virtual-real consistency of the brake shoe according to an inversion result, and obtaining a final virtual-real consistency result according to a comprehensive consistency evaluation result and a preset threshold value. According to the invention, the overall data is inverted based on the local data of the stress field, and the verification of the real-time and dynamic consistency is realized by combining the multi-dimensional consistency evaluation method, so that the precision and reliability of the health assessment of the part are effectively improved, and an innovative solution is provided for the fault early warning and intelligent maintenance of the equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twins, and particularly to a method for evaluating the virtual-real consistency of brake shoes. Background Art

[0002] With the rapid development of China's economic construction, high-rise buildings have been built rapidly. Elevators have become an indispensable facility in people's daily production and life and are special equipment directly related to people's life and property safety. As an important friction braking component, the performance of brake shoes is crucial for the overall braking effect and equipment safety. Traditional methods for evaluating the health of brake shoes mainly rely on physical detection means, such as regular inspections, vibration analysis, visual monitoring, etc. However, these methods have the following limitations: First, the number of physical sensors is limited and cannot comprehensively monitor. Currently, most systems use a limited number of stress or temperature sensors to monitor the state of brake shoes. These sensors are usually arranged in local areas of the brake shoes and are difficult to comprehensively reflect the stress distribution and performance changes of the entire brake shoe. Second, there is a lack of real-time and continuous virtual-real consistency verification methods. Traditional virtual-real consistency verification is mostly offline or periodic analysis and lacks real-time performance. Although some methods can achieve limited real-time detection, they still cannot meet the requirements for real-time health management in high-speed and complex operating environments. Third, there is a lack of comprehensive evaluation criteria and a unified evaluation framework. Existing methods have made breakthroughs in certain specific dimensions but still have not formed a complete and systematic evaluation framework. Most existing evaluation methods focus on single data sources and cannot comprehensively evaluate multiple dimensions such as global consistency, local consistency, and dynamic consistency, resulting in one-sidedness and low accuracy of evaluation results. Summary of the Invention

[0003] Therefore, the purpose of the present invention is to provide a method for evaluating the virtual-real consistency of brake shoes, which adopts a reasonable sensor layout and real-time data fusion, and inversely calculates the overall data based on the local data of the stress field. Combining with a multi-dimensional consistency evaluation method, it realizes the verification of real-time and dynamic consistency, overcomes the problems of data limitations, error amplification, and lack of real-time performance existing in the prior art, effectively improves the accuracy and reliability of component health evaluation, and provides an innovative solution for equipment fault warning and intelligent maintenance.

[0004] To achieve the above purpose, a method for evaluating the virtual-real consistency of brake shoes provided by the present invention includes the following steps: S1. Arrange sensors near the brake shoes according to a preset layout strategy; collect brake shoe stress data using the sensors; S2. Constructing a brake shoe model, including establishing a brake shoe geometric model, discretizing the brake shoe's geometric structure through finite element analysis, generating a mesh, and establishing boundary constraints; adding load conditions to the established brake shoe geometric model, and solving the brake shoe stress distribution under different working conditions through finite element analysis; S3, using the collected brake shoe stress data to inverse the constructed brake shoe model; S4. According to the inversion results, the brake shoe virtual-real consistency is judged, and a multi-scale consistency evaluation criterion is set to perform consistency evaluation; the final virtual-real consistency result is obtained based on the comprehensive consistency evaluation results and the preset threshold.

[0005] Further preferably, the preset layout strategy includes arranging a plurality of optical fiber sensors at a preset interval at the upper and lower ends of the brake shoe side, and the preset interval is set according to the sensor acquisition range including the entire brake shoe side stress change range.

[0006] Further preferably, in S3, the constructed brake shoe model is inverted using the collected brake shoe stress data, including: S301, based on the stress distribution of the brake shoe under different working conditions obtained by finite element solution; representing the corresponding stress field data under different working conditions as a stress matrix; S302, using the POD algorithm to centralize the stress matrix, calculating the covariance of any two centralized matrices, and calculating the correlation between the stress fields under different working conditions; S303, performing eigenvalue decomposition on all covariance matrices to obtain eigenvalues and eigenvectors; sorting the eigenvalues from large to small; selecting the eigenvectors corresponding to the eigenvalues that meet the sorting requirements as POD modes; using the POD modes to represent the main modes of the brake shoe stress field, and decomposing the stress field into a set of linear combinations of modes; S304. Reconstruct the stress field using modal dimensionality reduction, and infer the stress field distribution of the entire brake shoe from limited sensor data.

[0007] Further preferably, in S4, the brake shoe virtual-real consistency evaluation is performed based on the inversion result, including: A multi-scale consistency evaluation criterion is set to perform consistency evaluation, and the multi-scale consistency evaluation criterion is measured using a comprehensive consistency evaluation; the comprehensive consistency evaluation includes standardizing and weighting the scores of the global consistency evaluation, local consistency evaluation, and dynamic consistency evaluation to obtain a final comprehensive consistency evaluation result; the final virtual-real consistency result is obtained based on the comprehensive consistency evaluation result and a preset threshold.

[0008] Further preferably, the global consistency evaluation is calculated based on the mean square error (MSE) of the stress calculated according to the following formula (1): ; Formula (1) Wherein, is the stress value predicted by the model; is the stress value measured actually; is the number of measurement points; is the global consistency score.

[0009] Further preferably, the local consistency evaluation includes the following calculation steps: S401. Calculate the maximum error (Max Error) to evaluate the error magnitude in the local area: S402. Calculate the local weighted error and assign higher weights to the stress values in the areas where the brake pads are prone to wear and fatigue; Wherein, is the number of measurement points in the local area, is the weight of each point in the local area.

[0010] S403. Local consistency score : When the maximum error is closer to zero, it indicates better local consistency; the scoring range is [0, 1], and the closer the local consistency score value is to 1, the better the global consistency.

[0011] Further preferably, the dynamic consistency is compared using time series and includes the following calculation steps: S411. Set two groups of strain time series, where one group is the actual measurement data , and the other group is the model prediction data ; Calculate the local distance metric between the two strain time series: ; Wherein: are the values of the time series at the th moment respectively; S412. Define the cumulative distance matrix to record the shortest cumulative distance from corresponding to , and calculate the cumulative distance matrix through a recursive formula: S413. Calculate the DTW distance using the cumulative distance matrix and convert the DTW distance into a similarity score: Wherein: The closer it is to 1, the higher the similarity between time series, and the more accurate the dynamic response prediction of the model; The closer it is to 0, the lower the similarity between time series, and the less accurate the dynamic response prediction of the model.

[0012] Furthermore, the calculation of the cumulative distance matrix includes the following steps: S4121. Seek the shortest path according to the cumulative distance matrix using the following three assumptions; Assumption 1. Move downwards: It represents the step size from to while keeping unchanged (time delay); it usually means that the change of time series lags behind the time series ; Assumption 2. Move rightwards: It represents the step size from to while keeping unchanged (time advance); it usually means that the change of time series leads the time series ; Assumption 3. Move diagonally: It represents approaching along both to directions simultaneously for time synchronization; S4122. Define the boundary conditions The initial value of the cumulative distance matrix is set as: For the first column and the first row of the matrix: , ; where ，x k (k = 1, 2..i) represents any value in the first column of the matrix, y k (k = 1, 2..j) represents any value in the first row of the matrix; S4123. Calculate the optimal path and the DTW distance Using the cumulative distance matrix, determine the optimal alignment path. The optimal alignment path is the path from backtracking to in reverse, and calculate the final DTW distance: The smaller this distance metric value is, the higher the degree of alignment of the two time series in time and the greater the similarity.

[0013] Further preferably, the comprehensive consistency evaluation includes the following calculation steps: Standardize the obtained global consistency evaluation score, local consistency evaluation score, and dynamic consistency evaluation score according to historical data; Use the standardized evaluation scores as input features P, and adopt the least squares method to fit and solve the weights of the evaluation scores as follows, and output the comprehensive consistency score Q using a linear regression model; are the weights to be solved.

[0014] The method for judging the virtual and real consistency of the brake shoe disclosed in the present application has the following advantages compared with the prior art: The prior art usually relies on a single stress sensor or partial simulation model to evaluate the health state of the brake shoe, which may lead to the amplification of local errors or the inability to comprehensively reflect the stress distribution of the entire brake shoe. However, the present invention adopts a global sensor layout and a multi-dimensional consistency judgment method, and can more comprehensively and accurately judge the consistency between the virtual model and the actual sensor data by comprehensively comparing the stress data on each side of the brake shoe, greatly improving the accuracy of the judgment.

[0015] The present invention combines a finite element analysis model with real-time sensor data, and realizes the comparison and synchronization of time series through methods such as dynamic time warping (DTW), and can more accurately reflect the stress evolution under different working conditions. Compared with the traditional method, this method can achieve higher adaptability, especially in the face of complex and dynamically changing working conditions, and can effectively improve the reliability and accuracy of the virtual model.

[0016] Traditional virtual and real consistency verification mostly relies on offline analysis or regular inspection, and cannot achieve continuous and real-time verification. The present invention can obtain stress data in real time through stress sensors arranged on different sides of the brake shoe, and compare it with the virtual model to monitor the health state of the brake shoe at any time. This real-time verification provides a more timely and effective means for the fault warning of the equipment, reducing the risk of equipment failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 is a schematic flow chart of the method for judging the virtual and real consistency of the brake shoe provided by the present invention.

[0018] Figure 2 This is the model diagram of the brake shoe provided by the present invention. Detailed implementation manners

[0019] The present invention will be further described in detail below with reference to the accompanying drawings and specific implementation manners.

[0020] As Figure 1 shown, the method for judging the virtual-real consistency of the brake shoe provided by an embodiment of the present invention on the one hand includes the following steps: S1. Arrange sensors near the brake shoe according to a preset layout strategy; collect the stress data of the brake shoe by using the sensors; The preset layout strategy includes arranging a plurality of fiber optic sensors at the upper and lower ends of the side surface of the brake shoe at a preset interval, and the preset interval is set according to the stress change range of the entire side surface of the brake shoe included in the sensor collection range. Specifically, it includes: According to Figure 2 shown, focus on arranging at the positions near the contact surface of the side surface of the brake shoe, because the contact area is usually the place where the stress is most concentrated. Select the upper and lower parts (i.e., both ends of the brake shoe) near the contact area of the side surface of the brake shoe for sensor arrangement. The position marked as A in the figure is the position for arranging the sensor.

[0021] Key areas: For brake shoe components with a relatively long size, the central position and two end areas of the brake shoe can be selected for arrangement. These places are usually areas where the stress distribution changes greatly.

[0022] Arrangement quantity: Arrange 3-5 fiber optic strain sensors, and cover the stress state of the entire side surface of the brake shoe through reasonable spacing distribution to capture possible stress concentration points or uneven stress conditions.

[0023] Specific process: Arrange stress sensors → signal acquisition and processing module → data transmission module (Wi-Fi / 4G / Ethernet) → cloud platform API / MQTT reception → data storage (database) → data processing and analysis (cloud computing). Thus, the stress change of the brake shoe is monitored by arranging stress sensors.

[0024] S2. Build a brake shoe model, including establishing a geometric model of the brake shoe, discretizing the geometric structure of the brake shoe through finite element analysis to generate a mesh, and establishing boundary constraints; adding load conditions to the established geometric model of the brake shoe, and solving the stress distribution of the brake shoe under different working conditions through finite element; In order to infer the stress of the entire brake shoe from a small amount of sensor data, it is first necessary to establish a stress calculation model of the brake shoe. According to the known geometric structure and mechanical properties of the brake shoe, the modeling is carried out through the following steps: Geometric modeling: First, a geometric model of the brake shoe is created. Finite element analysis is used to discretize the brake shoe's geometry and generate a mesh. The model incorporates boundary conditions, such as the brake shoe's material, geometry, and mechanical behavior in the working environment.

[0025] Mechanical modeling: According to the working principle of the elevator brake, load conditions are introduced into the brake shoe model, such as the influence of braking force, temperature change, friction and other factors. Based on mechanical theory, the stress model of the brake shoe is constructed.

[0026] Stress Field Solution: Finite element software is used to numerically solve the brake shoe's stress field and determine its stress distribution under different operating conditions. This process involves both static and dynamic analysis. First, a static analysis is performed to calculate the brake shoe's stress distribution at rest. Next, a dynamic analysis is performed to simulate how stress changes over time during braking, taking into account time effects (such as strain during braking).

[0027] S3. Inverse analysis of the constructed brake shoe model using the collected brake shoe stress data; including: S301, based on the stress distribution of the brake shoe under different working conditions obtained by finite element solution; representing the corresponding stress field data under different working conditions as a stress matrix; S302, using the POD algorithm to centralize the stress matrix, calculating the covariance of any two centralized matrices, and calculating the correlation between the stress fields under different working conditions; S303, performing eigenvalue decomposition on all covariance matrices to obtain eigenvalues and corresponding eigenvectors; sorting the eigenvalues from large to small; selecting the eigenvectors corresponding to the eigenvalues that meet the sorting requirements as POD modes, using the POD modes to represent the main modes of the brake shoe stress field, and decomposing the stress field into a set of linear combinations of modes; S304. Reconstruct the stress field using modal dimensionality reduction, and infer the stress field distribution of the entire brake shoe from limited sensor data.

[0028] S301 specifically includes obtaining the stress field data of the brake shoe under different working conditions through finite element analysis, which is represented as a matrix, where each column represents the stress field under a different working condition. The stress matrix of the brake shoe is as follows: in: is a stress matrix containing stress field data under each working condition; Is a vector, indicating the working condition The dimension of this vector is equal to the number of nodes of the brake shoe.

[0029] In S302, it specifically includes: The first step of POD is to centralize the stress matrix, aiming to eliminate the average effect in the stress data and ensure that the POD decomposition focuses on the main fluctuations of the data. For the stress field of each working condition subtract the mean stress field of all working conditions , where: to obtain the centralized data matrix: POD extracts the main features of the data by performing covariance analysis on the stress matrix. First, calculate the covariance matrix of the centralized stress matrix , which represents the correlation between stress fields under different working conditions: where: is the covariance matrix, with the dimension of (where is the dimension of each stress field vector, equal to the number of grid nodes); is the inner product of the data matrix, used to measure the similarity between stress field data.

[0030] In S303, it specifically includes: Perform eigenvalue decomposition on the covariance matrix to obtain eigenvalues and eigenvectors: where: is the eigenvalue of the covariance matrix, representing the variance corresponding to each eigenvector; is the eigenvector of the covariance matrix, representing the main patterns of the data.

[0031] By sorting the eigenvectors and eigenvalues of the covariance matrix, select the first few dominant eigenvectors ; these eigenvectors are the POD modes, thus capturing the main features of the stress field. The POD modes are derived as follows: These modes represent the main patterns of the brake shoe stress field, and decompose the stress field into a linear combination of a set of modes.

[0032] In S304, it specifically includes that in the actual elevator working condition, due to the limited component layout of the elevator, only a small number of sensors can be arranged to obtain the data source, so use as few modes as possible to reconstruct the stress field. By selecting the eigenvectors corresponding to the largest several eigenvalues , for dimensionality reduction representation. The reconstruction of the stress field is expressed as: Wherein: is the coefficient of each mode, representing the working condition and the weight of each mode under the working condition; is the number of selected POD modes.

[0033] Through the above method, the stress field distribution of the entire brake shoe is deduced from limited sensor data.

[0034] Based on the above method, a method for obtaining the model of the brake shoe stress field through a small number of sensors is proposed. To ensure the real-time reliability of the stress distribution of the brake shoe model, a method based on the global layout of sensors is proposed: First, stress sensors are evenly arranged on the four sides of the whole brake shoe. Since a small number of sensors can only monitor the stress changes on one side of the brake shoe, the stress distribution on the other three sides is verified through global layout, and finally the credibility of the model is verified, which further provides a verification basis for the evaluation of the virtual-real consistency of the brake shoe.

[0035] Sensor position: Stress sensors are evenly arranged on the four sides of the brake shoe to ensure that there is sensor data on each side. Considering the cost of sensors and the limitations of actual deployment, usually a small number of sensors (such as 3-5 sensors) are selected to be arranged on each side.

[0036] Data transmission: The local stress data collected by the sensors is transmitted to the central processing system in real time through wireless or wired transmission methods.

[0037] Position selection: According to the geometric characteristics of the brake shoe, representative areas (such as areas with large stress gradients) are selected to arrange sensors, especially in areas with large forces, such as the contact area or areas where local stress concentration may occur.

[0038] S4. Evaluate the virtual-real consistency of the brake shoe according to the inversion results, and set multi-scale consistency evaluation criteria for consistency evaluation; obtain the final virtual-real consistency result according to the comprehensive consistency evaluation result and the preset threshold.

[0039] Set multi-scale consistency evaluation criteria for consistency evaluation. The multi-scale consistency evaluation criteria include: global consistency evaluation, local consistency evaluation, dynamic consistency evaluation, time consistency evaluation, and comprehensive consistency evaluation; The comprehensive consistency evaluation includes normalizing and weighted calculating the scores of each consistency evaluation to obtain the final comprehensive consistency evaluation result; obtain the final virtual-real consistency result according to the comprehensive consistency evaluation result and the preset threshold.

[0040] On the global scale, evaluate the consistency of the strain distribution on the entire brake shoe surface or key contact areas. Global consistency mainly focuses on the large-scale strain change trend and is applicable to checking whether the model can capture the overall physical behavior.

[0041] Evaluation criterion: The root mean square error (MSE) based on the strain field measures the consistency of the predicted strain field of the model with the actual strain field in the overall trend.

[0042] The specific steps are as follows: Quantify the global consistency by calculating the mean square error (MSE) of the stress.

[0043] Among them, is the stress value predicted by the model; is the stress value measured actually; is the number of measurement points.

[0044] Regarding the global consideration of the brake shoe, in actual applications, each area of the brake shoe has an impact on the overall performance. Therefore, first, from the perspective of equal regional importance, evaluate the overall error of the model.

[0045] Global consistency score : ; The score is calculated by the mean square error, and the score range is [0, 1]. The closer the score value is to 1, the better the global consistency.

[0046] 4.2 Local consistency (microscopic scale) On the local scale, focus on the key positions where the strain sensors are located, especially the highly sensitive areas such as strain concentration regions, stress concentration points, and contact surfaces. Local consistency is more refined and is especially suitable for verifying the model's prediction ability for details.

[0047] Evaluation criterion: Local maximum absolute error (Max Error) and local root mean square error (Local RMSE). If the local error is greater than a certain preset threshold, it indicates that the model fails to accurately capture the key strain changes. The specific steps are as follows: 1. Calculate the maximum error (Max Error) to evaluate the error magnitude of the local area: 2. Calculate the local weighted error, and assign higher weights to the stress values in the areas where the brake lining is prone to wear and fatigue.

[0048] Among them, is the number of measurement points in the local area, is the weight of each point in the local area.

[0049] 3. Local consistency score : When the maximum error is closer to zero, it indicates better local consistency; the scoring range is [0, 1], and the closer the local consistency score value is to 1, the better the global consistency.

[0050] 4.3 Dynamic Consistency The strain field of the brake shoe is not only static, but also changes over time. Therefore, it is crucial to evaluate the dynamic strain consistency. The dynamic consistency evaluates the change trend of the stress field at different time steps, and focuses on whether the change of stress over time is consistent with the actual measured change trend. Time series comparison is used for comparison.

[0051] By comparing the curves of strain data changing over time, the accuracy of the model under dynamic response is evaluated. This can be achieved through time series analysis methods, such as Dynamic Time Warping (DTW) and Cross-correlation analysis.

[0052] Judgment criterion: Use the Dynamic Time Warping (DTW) method to compare the actually measured strain time series with the strain time series predicted by the model, and check the temporal consistency between the two. DTW can effectively align data at different time steps and calculate the similarity.

[0053] The specific steps are as follows: 1. Define the distance metric: First, set two groups of strain time series, one group from the actual measurement data , and the other group from the model prediction ; Then define the local distance metric between the two strain time series: Where: are the values of the time series at the th moment respectively.

[0054] 2. Define the cumulative distance matrix DTW records the shortest cumulative distance from corresponding to to by defining a cumulative distance matrix

[0055] The cumulative distance matrix is calculated through a recursive formula: Where: To find the shortest path, three possible paths are proposed.

[0056] Move down: represents the step from to while keeping Unchanged (time delay); usually represents the time series The change lags behind the time series ; Shift to the right: Indicates the step size from to while keeping unchanged (time advance); usually represents the time series The change leads the time series ; Diagonal movement: Indicates approaching in both directions simultaneously along to for time synchronization.

[0057] 3. Define boundary conditions The initial value of the cumulative distance matrix is set as: For the first column and the first row of the matrix: , ; where ，x k (k = 1, 2..i) represents any value in the first column of the matrix, y k (k = 1, 2..j) represents any value in the first row of the matrix.

[0058] 4. Calculate the optimal path and DTW distance After calculating the cumulative distance matrix, the optimal alignment path corresponds to the path from backtracking to This path minimizes the cumulative distance between the two time series and represents the best path for time series alignment.

[0059] The final DTW distance is the value of the cumulative distance matrix : The smaller this distance metric value is, the higher the degree of alignment between the two time series in time and the greater the similarity.

[0060] To evaluate the accuracy of the model under dynamic response, the DTW distance can be converted into a similarity score. The smaller the DTW distance, the higher the similarity of the time series. The distance is converted into a similarity score using the following formula: ; where: The closer it is to 1, the higher the similarity between the time series and the more accurate the dynamic response prediction of the model; The closer to 0, the lower the similarity between time series, and the less accurate the dynamic response prediction of the model.

[0061] Furthermore, to comprehensively evaluate the accuracy of the virtual model, the consistency results at different scales and dimensions above are fused into a comprehensive score. The comprehensive consistency evaluation includes: standardizing the obtained global consistency evaluation score, local consistency evaluation score, and dynamic consistency evaluation score according to historical data; Taking the standardized evaluation scores as input features P, the following formula uses the least squares method to fit and solve the weights of each evaluation score, and a linear regression model is used to output the comprehensive consistency score Q.

[0062] First, for each consistency evaluation (global consistency, local consistency, dynamic consistency), a consistency score needs to be obtained through different methods (such as mean square error, DTW, trend matching, etc.). The values of these scores represent the consistency level of the model at each scale.

[0063] Set the calculation results of each consistency score as follows: Global consistency score: ; Local consistency score: ; Dynamic consistency score: .

[0064] Then, since the dimensions and ranges of different evaluation indicators are different, in order to avoid a certain indicator having too much influence on the comprehensive score, the consistency scores of each item are standardized. The scores of each indicator are unified to the same dimension for subsequent weighted calculation.

[0065] Min-Max standardization: Where: are the minimum and maximum values of this indicator respectively, is the actual score of this indicator; through the standardization method, it is ensured that each consistency score is within the same range.

[0066] In order to fuse the consistency scores at different scales and dimensions into a comprehensive score, a weight needs to be assigned to each consistency score. These weights reflect the relative importance of different consistency scales to the model accuracy. Therefore, a data-driven method is proposed to solve the weighted coefficients through a linear regression model, establish a linear relationship between the model performance and each consistency score, and use regression analysis to automatically solve the weights.

[0067] Therefore, a linear regression model is established: using the consistency score as the input feature P and the performance score of the model as the output Q; The matrix P contains data of various consistency metrics in the following form: ; The target vector contains data of the performance scores: ; A linear regression model is fitted by the least squares method: ; where, is the performance score of the model, are the weights to be solved, is the error term.

[0068] Secondly, the weights are solved using the least squares method : where X is the matrix containing the consistency scores and Y is the vector of the performance scores, are the obtained weights.

[0069] 4.4.4 Comprehensive Score Calculation Based on the standardized consistency scores and weights, a comprehensive consistency score is obtained through fusion using the weighted average method: ; and it satisfies ; The following is the specific process of the comprehensive evaluation: Firstly, determine the score range: The value of the comprehensive score usually falls between 0 and 1, representing the overall accuracy of the virtual model in all consistency dimensions.

[0070] When it indicates a perfect match between the virtual stress field of the model and the actual measurement data.

[0071] When it indicates a complete mismatch between the virtual stress field of the model and the actual measurement data.

[0072] Then, set the judgment threshold: A suitable threshold can be set according to the braking conditions of different types of brakes to judge the consistency of the model. For example: : indicating good model consistency; : indicating average model consistency and may require further optimization; : indicating poor model consistency and model adjustment is required.

[0073] Obviously, the above embodiments are only examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or alterations can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. And the obvious changes or alterations derived therefrom still fall within the protection scope of the present invention.

Claims

1. A method for evaluating the virtual and real consistency of a brake shoe, characterized in that: It includes the following steps: S1. Arrange sensors near the brake shoe according to a preset layout strategy; collect brake shoe stress data using the sensors; S2. Construct a brake shoe model, including establishing a brake shoe geometric model, discretizing the geometric structure of the brake shoe through finite element analysis to generate a mesh, and establishing boundary constraints; add load conditions to the established brake shoe geometric model, and solve for the stress distribution of the brake shoe under different working conditions through finite element analysis; S3. Invert the constructed brake shoe model using the collected brake shoe stress data; S4. Conduct a virtual-real consistency evaluation of the brake shoe according to the inversion result, and set a multi-scale consistency evaluation criterion for the consistency evaluation; obtain the final virtual-real consistency result based on the comprehensive consistency evaluation result and a preset threshold.

2. The method for judging the virtual and real consistency of the brake shoe according to claim 1, characterized in that The preset layout strategy includes setting multiple fiber optic sensors at the upper and lower ends of the side of the brake shoe at a preset spacing, and the preset spacing is set according to the sensor acquisition range to cover the entire stress change range of the side of the brake shoe.

3. The method for judging the virtual and real consistency of the brake shoe according to claim 1, characterized in that, In S3, inverting the constructed brake shoe model using the collected brake shoe stress data includes: S301. According to the stress distribution of the brake shoe under different working conditions obtained after finite element solution; represent the corresponding stress field data under different working conditions as a stress matrix; S302. Use the POD algorithm to centralize the stress matrix, calculate the covariance for any two centralized matrices, and calculate the correlation between stress fields under different working conditions; S303. Perform eigenvalue decomposition on all covariance matrices to obtain eigenvalues and eigenvectors; sort the eigenvalues from largest to smallest; select the eigenvectors corresponding to the eigenvalues that meet the sorting requirements as POD modes, use the POD modes to represent the main modes of the brake shoe stress field, and decompose the stress field into a linear combination of a set of modes; S304. Use modal dimension reduction to reconstruct the stress field and deduce the stress field distribution of the entire brake shoe from limited sensor data.

4. The method for judging the virtual and real consistency of the brake shoe according to claim 1, wherein, In S4, conducting a virtual-real consistency evaluation of the brake shoe according to the inversion result includes: Set a multi-scale consistency evaluation criterion for the consistency evaluation, and the multi-scale consistency evaluation criterion is measured by comprehensive consistency evaluation; the comprehensive consistency evaluation includes standardizing and weighting the scores of global consistency evaluation, local consistency evaluation, and dynamic consistency evaluation to obtain the final comprehensive consistency evaluation result; obtain the final virtual-real consistency result based on the comprehensive consistency evaluation result and a preset threshold.

5. The method for judging the virtual and real consistency of the brake shoe according to claim 4, wherein The global consistency evaluation is calculated based on the following formula to calculate the mean square error MSE of stress: ； ； Among them, is the stress value predicted by the model; is the stress value measured actually; is the number of measurement points; is the global consistency score.

6. The method for judging the virtual and real consistency of the brake shoe according to claim 4, characterized in that The local consistency evaluation includes the following calculation steps: S401. Calculate the maximum error Max Error to evaluate the error magnitude in the local area: S402. Calculate the local weighted error and assign higher weights to the stress values in the areas where the brake lining is prone to wear and fatigue; Among them, is the number of measurement points in the local area, is the weight of each point in the local area; is the stress value predicted by the model; is the actually measured stress value; S403, Local Consistency Scoring : When the maximum error is closer to zero, it indicates better local consistency; the scoring range is [0,1], and the closer the local consistency scoring value is to 1, the better the global consistency.

7. The method for judging the virtual and real consistency of the brake shoe according to claim 4, characterized in that The dynamic consistency is compared using time series and includes the following calculation steps: S411. Set two groups of strain time series, where one group is actual measurement data , and the other group is model prediction data ; Calculate the local distance metric between two strain time series: ; where: are the values of the time series at the th moment respectively; S412. Define the cumulative distance matrix to record the shortest cumulative distance from corresponding to . Calculate the cumulative distance matrix through a recursive formula: S413. Calculate the DTW distance using the cumulative distance matrix and convert the DTW distance into a similarity score: Wherein: The closer it is to 1, the higher the similarity between time series, and the more accurate the dynamic response prediction of the model; The closer it is to 0, the lower the similarity between time series, and the less accurate the dynamic response prediction of the model.

8. The method for evaluating the virtual and real consistency of the brake shoe according to claim 7, characterized in that The calculation of the cumulative distance matrix includes the following steps: S4121. Seek the shortest path according to the cumulative distance matrix using the following three assumptions; Hypothesis 1: Moving downward: Indicates the step size from to while keeping unchanged, indicating that the change in the time series lags behind the time series ; Hypothesis 2: Move to the right: denote from...to... the step size of, keeping unchanged, denoting that the time series change of leads the time series ; Hypothesis 3, diagonal movement: It means approaching in both to two directions simultaneously for time synchronization; S4122. Define boundary conditions The initial value of the cumulative distance matrix is set to: ; For the first column and the first row of the matrix: , ; Among them ，x k (k = 1, 2..i) represents any value in the first column of the matrix, y k (k = 1, 2..j) represents any value in the first row of the matrix; S4123. Calculate the optimal path and the DTW distance; Using the cumulative distance matrix, determine the optimal alignment path, which is the path from backtracking in reverse to , and calculate the final DTW distance: The smaller the value of this distance metric, the higher the degree of alignment of the two time series in time and the greater the similarity.

9. The method for judging the virtual and real consistency of the brake shoe according to claim 4, wherein The comprehensive consistency evaluation includes the following calculation steps: According to the historical data, standardize the obtained global consistency evaluation score, local consistency evaluation score, and dynamic consistency evaluation score; Use the standardized evaluation scores as input features P, and use the least squares method to fit and solve the weights of the evaluation scores according to the following formula, and output the comprehensive consistency score Q using a linear regression model; Among them, is the comprehensive consistency score of the model, is the weight to be solved, is the error term; is the global consistency score; is the local consistency score; is the dynamic consistency score.

Images