Bridge bearing sliding plate wear condition monitoring method and system

Abstract

The application discloses a kind of bridge support sliding plate wear condition monitoring method and system, method includes: S100, with the same sliding plate material in support, after being divided into multiple samples, it is divided into multiple groups, and wear test is carried out on testing machine, and according to test result, training label is marked, and test data is collected to form comparison database;S200, on the middle lining of support, configure force sensor, acoustic emission sensor and data collection instrument, collect the stress and sliding plate wear data of support;S300, after analyzing and processing the stress and sliding plate wear data of support, obtain monitoring data, call corresponding test data from comparison database, compare two data;S400, according to the result of comparison, call training label in test data as the wear condition output of support sliding plate.According to the bridge support sliding plate wear condition monitoring method and system proposed in the application, the actual wear degree of the sliding plate can be accurately judged, the precision of wear evaluation is improved, and the safe and stable operation of the support is ensured.

Description

Technical Field

[0001] This invention belongs to the field of bridge monitoring technology, and more specifically, relates to a method and system for monitoring the wear condition of bridge bearing sliding plates. Background Technology

[0002] In long-span cable-stayed bridge structures, the main girder is constantly subjected to high-frequency, low-amplitude motion under the influence of temperature loads, vehicle loads, and wind loads. The large daily cumulative displacement often leads to excessive wear of the bearing sliding wear plates. Excessive wear of the bearing sliding wear plates is a common problem in long-span cable-stayed bridges. Excessive wear of the sliding wear plates causes steel-to-steel friction within the bearing, significantly increasing the bearing frictional resistance and even altering the bridge's structural stress system, thus having a significant adverse impact on the bridge's normal service life.

[0003] The exposed wear thickness of the bearing sliding wear plate is typically 3mm. Currently, the wear condition of the sliding plate mainly relies on manual on-site inspection using feeler gauges. However, manual feeler gauge measurements are often subjective, subject to uncertainty in the inspection location, and time-sensitive, leading to an inaccurate and incomplete assessment of the wear condition of the bearing sliding wear plate, and prone to misjudgments and omissions. This is especially problematic in long-span cable-stayed bridges where the bearings are typically tension-compression bearings. When the bearings are under tension during service, the exposed thickness of the sliding plate measured with feeler gauges may actually increase, resulting in distorted measurements. The current limitations of manual measurement methods make it impossible to accurately assess the service condition of the bearing sliding plate in real time, which is detrimental to the later maintenance and upkeep of long-span cable-stayed bridge bearings.

[0004] Therefore, a method and system for monitoring the wear condition of bridge bearing sliding plates is needed. By monitoring the wear condition of the exposed thickness of the bearing sliding plate through a data signal acquisition system, the service status of the bearing sliding plate can be obtained intuitively, conveniently and in real time, thereby achieving comprehensive real-time monitoring of the service condition of the bearing sliding wear plate and ensuring the operation of the bearing throughout its entire life cycle. Summary of the Invention

[0005] To address the aforementioned deficiencies or improvement needs of existing technologies, this invention provides a method and system for monitoring the wear condition of bridge bearing sliding plates. By conducting wear tests on different material samples and establishing a comparative database, combined with the installation of multiple sensors in actual operating environments to collect real-world operational data, the method achieves accurate assessment of the wear condition of the bearing sliding plates. Utilizing comparative analysis technology, the test data is matched with on-site monitoring data, and combined with preset training labels, the actual wear degree of the sliding plates is accurately determined, improving the accuracy of wear assessment. Furthermore, this invention provides effective technical support for the timely detection and prevention of bearing failures, contributing to ensuring the safe and stable operation of related facilities.

[0006] To achieve the above objectives, according to a first aspect of the present invention, a method for monitoring the wear condition of bridge bearing sliding plates is provided, comprising the following steps:

[0007] S100. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups. Each group of samples is subjected to abrasion test on the testing machine. According to the test results, a training label is applied. All test data are collected to form a comparison database.

[0008] S200: A force sensor, an acoustic emission sensor, and a data collector are installed on the intermediate liner of the support to collect data on the force and wear of the slide plate during the operation of the support.

[0009] S300: After analyzing and processing the support stress and slide plate wear data, the monitoring data is obtained. The corresponding test data is retrieved from the comparison database and the monitoring data is compared with the test data.

[0010] S400: Based on the comparison results, call the training labels in the test data as the output of the wear status of the support slide plate.

[0011] Furthermore, step S100 specifically includes the following steps:

[0012] S110. The wear stages of the skateboard are divided into three stages: initial normal sliding wear, intermediate dry friction, and later steel-to-steel friction.

[0013] S120. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples, then divide them into multiple large sample groups, and each large sample group is further divided into three small sample groups.

[0014] S130. Provide the characteristics of three wear stages for the samples in the three subgroups of each large sample group, and then conduct wear tests on each large sample group according to different characteristic parameters.

[0015] S140. Acoustic emission sensors are used to collect and record the AE signal data of each sample, and wavelet packet transform is used to filter and perform eigenvalue analysis on the collected AE signals.

[0016] S150. Integrate and analyze the data, label the samples with different test results with training tags, and collect the test data of all samples to form a comparison database.

[0017] Furthermore, step S140 specifically includes the following steps:

[0018] S141. Perform preprocessing on the acquired AE signal data, including DC bias and filtering, and then select a suitable wavelet basis according to the characteristics of the AE signal data.

[0019] S142. Based on the bandwidth and frequency resolution of the AE signal data, select the number of decomposition layers of the signal and decompose the AE signal data into AE sub-signals with multiple frequency bands.

[0020] S143. Extract test features useful for evaluating the wear state of the skateboard from each AE sub-signal, and combine the extracted features into a test feature vector.

[0021] Further, in step S142, the AE signal data decomposition process is as follows:

[0022]

[0023] Among them, cA j [k] represents the approximation coefficients of the j-th layer in the low-frequency component.

[0024] cD j [k is the detail coefficient of the j-th layer in the high-frequency part,

[0025] h[n] is the impulse response of the low-pass filter.

[0026] g[n] is the impulse response of the high-pass filter.

[0027] x[n] is the input signal.

[0028] k is the index of the sub-signal.

[0029] For multi-level decomposition, each level of decomposition will generate two sub-signals, specifically:

[0030]

[0031]

[0032] Furthermore, in step S143, the extracted experimental features include energy feature E. i Root Mean Square (RMS) characteristics i and kurtosis feature K i The energy characteristic E i for:

[0033]

[0034] Among them, c i [n] represents the coefficient of the i-th sub-signal.

[0035] The root mean square feature (RMS) i for:

[0036]

[0037] Where N is the length of the sub-signal;

[0038] The kurtosis feature K i for:

[0039]

[0040] Where, μ i Let be the mean of the i-th sub-signal.

[0041] Furthermore, the experimental feature vector F is composed of the energy feature E i Root Mean Square (RMS) characteristics i It is composed of kurtosis features, specifically:

[0042] F = [E1, E2, ..., E n RMS1, RMS2, ..., RMS n K1, K i2 , ..., K n ] T .

[0043] Furthermore, step S300 specifically includes:

[0044] S310. After processing the skateboard wear data according to the method in step S140, the monitoring data is obtained, and the monitoring feature vector F is selected from the monitoring data. mon ;

[0045] S320, Based on the monitoring feature vector F mon And the distance and similarity metrics between the two are calculated from the experimental feature vector F;

[0046] S330. Using the obtained distance and similarity metrics, a weighted fusion method is used to comprehensively score the monitored data.

[0047] S340. Based on the comprehensive score, find the closest sample from the comparison database and use the wear label of that sample as the result of the monitoring data.

[0048] Further, in step S320, the distance metric dF mon F) is the Euclidean distance d O (F mon F) or Mahalanobis distance d M (F mon F), the Euclidean distance d O (F mon F) is:

[0049]

[0050] Where m is the dimension of the experimental feature vector;

[0051]

[0052] Where S is the covariance matrix of the experimental feature vectors.

[0053] The mean μ of the sample feature vectors in the comparison database is:

[0054]

[0055] The similarity metric is cosine similarity sim(F). mon F i Specifically:

[0056]

[0057] Further, in step S330, the comprehensive score S i Specifically:

[0058] S i =w d ·d(F mon F i )+w s ·(1-sim(F mon F i )),

[0059] Among them, w d and w s These are the weights of the distance metric and the similarity metric, respectively, and w d +w s =1.

[0060] In step S340, the evaluation based on the comprehensive score is as follows:

[0061]

[0062] Among them, L best The wear label is for the closest specimen found from the comparison database.

[0063] According to a second aspect of the present invention, a bridge bearing sliding plate wear monitoring system is provided, comprising:

[0064] Test module: Used to select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups, each group of samples is subjected to abrasion test on the testing machine, training labels are marked according to the test results, and all test data are collected to form a comparison database;

[0065] Data acquisition module: Used to install force sensors, acoustic emission sensors and data collectors on the intermediate liner of the support to collect force and slide plate wear data during the operation of the support;

[0066] Data processing module: used to analyze and process the support stress and slide plate wear data to obtain monitoring data, retrieve the corresponding test data from the comparison database, and compare the monitoring data with the test data;

[0067] Output module: Used to retrieve the training labels from the experimental data based on the comparison results, and output the wear status of the support slide plate;

[0068] Prediction module: Used to build a machine learning model, which is trained based on sample data in the comparison database to predict the wear condition of the skateboard.

[0069] In summary, compared with the prior art, the above-described technical solutions conceived by this invention can achieve the following beneficial effects:

[0070] 1. The monitoring method of the present invention achieves accurate assessment of the wear condition of the support slide plate by conducting wear tests on different material samples and establishing a comparative database, combined with the installation of multiple sensors in the actual use environment to collect real operating data. By using comparative analysis technology to match the test data with the on-site monitoring data, and combined with preset training labels, the actual wear degree of the slide plate can be accurately determined, which improves the accuracy of wear assessment and provides effective technical support for timely detection and prevention of support failures, thus helping to ensure the safe and stable operation of related facilities.

[0071] 2. The monitoring method of the present invention refines the three stages of skateboard wear and designs specific wear tests for each stage to comprehensively simulate the wear process of the skateboard in actual use. It combines acoustic emission sensors to collect AE signals and uses wavelet packet transform technology for signal processing to ensure high-precision identification of wear characteristics. This helps to detect early signs of skateboard wear in a timely manner, prevent major safety accidents, and ensure the safe and efficient operation of equipment.

[0072] 3. The monitoring method of the present invention matches the closest sample wear label from the comparison database based on the score. This not only enables the rapid and accurate identification of the specific wear stage of the slide plate, but also effectively predicts potential faults, providing a scientific basis for timely maintenance measures. This ensures the safe and stable operation of the equipment, reduces unplanned downtime and maintenance costs, and improves overall operational efficiency.

[0073] 4. The monitoring method of the present invention establishes a machine learning model and trains it using sample data in a comparative database, enabling the model to accurately predict the wear state of the slide plate, learn the complex characteristics of different wear stages, and thus detect wear signs earlier in practical applications, provide timely warnings of potential faults, optimize maintenance plans, reduce losses caused by sudden failures, and improve the operational safety and economic benefits of the equipment. Attached Figure Description

[0074] Figure 1 This is a flowchart illustrating a method for monitoring the wear condition of a bridge bearing sliding plate according to an embodiment of the present invention.

[0075] Figure 2 This is a schematic diagram of the specific process of step S100 in a method for monitoring the wear condition of a bridge bearing sliding plate according to an embodiment of the present invention;

[0076] Figure 3 This is a schematic diagram of the specific process of step S140 in a method for monitoring the wear condition of a bridge bearing sliding plate according to an embodiment of the present invention;

[0077] Figure 4 This is a schematic diagram of the specific process of step S300 in a method for monitoring the wear condition of a bridge bearing sliding plate according to an embodiment of the present invention. Detailed Implementation

[0078] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0079] As the bearing continues to operate, the wear of the bearing slide plate can be mainly divided into three stages: initial normal sliding wear, intermediate dry friction after the silicone grease is exhausted, and final steel-to-steel friction after the slide plate is completely worn away. The initial wear stage mainly involves frictional wear of the slide plate under silicone grease lubrication. At this stage, the coefficient of friction is relatively low, and the wear is usually minimal. The intermediate wear stage mainly involves dry friction between the bearing slide plate and the stainless steel plate in the friction pair when the silicone grease in the slide plate's grease reservoir is exhausted. At this stage, the coefficient of friction increases, and the wear increases accordingly. The final wear stage mainly manifests as steel-to-steel friction of the friction pair after the exposed thickness of the slide plate is almost completely worn away. At this stage, the coefficient of friction increases significantly, and the slide plate wear is quite severe.

[0080] Example 1

[0081] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for monitoring the wear condition of bridge bearing sliding plates, which specifically includes the following steps:

[0082] S100. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups. Each group of samples is subjected to abrasion test on the testing machine. According to the test results, a training label is applied. All test data are collected to form a comparison database.

[0083] S200: A force sensor, an acoustic emission sensor, and a data collector are installed on the intermediate liner of the support to collect data on the force and wear of the slide plate during the operation of the support.

[0084] S300: After analyzing and processing the support stress and slide plate wear data, the monitoring data is obtained. The corresponding test data is retrieved from the comparison database and the monitoring data is compared with the test data.

[0085] S400: Based on the comparison results, call the training labels in the test data as the output of the wear status of the support slide plate.

[0086] like Figure 2 As shown, step S100 further includes the following steps:

[0087] S110. The wear stages of the skateboard are divided into three stages: initial normal sliding wear, intermediate dry friction, and later steel-to-steel friction.

[0088] S120. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples, then divide them into multiple large sample groups, and each large sample group is further divided into three small sample groups.

[0089] S130. Provide the characteristics of three wear stages for the samples in the three subgroups of each large sample group, and then conduct wear tests on each large sample group according to different characteristic parameters.

[0090] S140. Acoustic emission sensors are used to collect and record the AE signal data of each sample, and wavelet packet transform is used to filter and perform eigenvalue analysis on the collected AE signals.

[0091] S150. Integrate and analyze the data, label the samples with different test results with training tags, and collect the test data of all samples to form a comparison database.

[0092] In step S110, the initial normal sliding wear stage is characterized by the frictional wear of the slide plate under silicone grease lubrication. During this stage, the friction coefficient is relatively small, and the wear is usually small. The intermediate dry friction stage is characterized by the dry frictional wear of the support slide plate and the stainless steel plate in the friction pair when the silicone grease in the slide plate's grease reservoir is exhausted. During this stage, the friction coefficient increases, and the wear increases accordingly. The later steel-to-steel friction stage is characterized by the steel-to-steel friction of the friction pair after the exposed thickness of the slide plate is almost completely worn away. During this stage, the friction coefficient increases significantly, and the wear of the slide plate is more severe.

[0093] In step S130, the characteristic parameters include surface pressure, stroke, temperature, and displacement velocity. The design value of the bearing surface pressure in a bridge is generally 45 MPa, but there is a certain deviation between the actual surface pressure value and the design value. Therefore, it is necessary to conduct a simulation test on the surface pressure range near the design value to simulate the wear of the slide plate under different surface pressures (30, 35, 40, 45, 50 MPa).

[0094] In step S140, the AE signal data is signal data emitted by an acoustic emission sensor, which is a technique for assessing the health status of a material or structure by detecting elastic waves caused by micro fractures, crack propagation, or other forms of stress release occurring inside the material or structure.

[0095] like Figure 3 As shown, step S140 further includes the following steps:

[0096] S141. Perform preprocessing on the acquired AE signal data, including DC bias and filtering, and then select a suitable wavelet basis according to the characteristics of the AE signal data.

[0097] S142. Based on the bandwidth and frequency resolution of the AE signal data, select the number of decomposition layers of the signal and decompose the AE signal data into AE sub-signals with multiple frequency bands.

[0098] S143. Extract test features useful for evaluating the wear state of the skateboard from each AE sub-signal, and combine the extracted features into a test feature vector.

[0099] In step S142, the AE signal data decomposition process is as follows:

[0100]

[0101]

[0102] Among them, cA j [k] represents the approximation coefficients of the j-th layer in the low-frequency component.

[0103] cD j [k] represents the detail coefficients of the j-th layer in the high-frequency component.

[0104] h[n] is the impulse response of the low-pass filter.

[0105] g[n] is the impulse response of the high-pass filter.

[0106] x[n] is the input signal.

[0107] k is the index of the sub-signal.

[0108] For multi-level decomposition, each level of decomposition will generate two sub-signals, specifically:

[0109]

[0110] In step S143, the extracted experimental features include energy feature E. i Root Mean Square (RMS) characteristics i and kurtosis feature K i The energy characteristic E i for:

[0111]

[0112] Among them, c i [n] represents the coefficient of the i-th sub-signal;

[0113] The root mean square feature (RMS) i for:

[0114]

[0115] Where N is the length of the sub-signal;

[0116] The kurtosis feature K i for:

[0117]

[0118] Where, μ i Let be the mean of the i-th sub-signal.

[0119] The experimental feature vector F is composed of the energy feature E. i Root Mean Square (RMS) characteristics i It is composed of kurtosis features, specifically:

[0120] F = [E1, E2, ..., E n RMS1, RMS2, ..., RMS n K1, N2, ..., K n ] T .

[0121] In step S150, the training labels include: initial wear, intermediate wear, and severe wear.

[0122] like Figure 4 As shown, step S300 further includes the following steps:

[0123] S310. After processing the skateboard wear data according to the method in step S140, the monitoring data is obtained, and the monitoring feature vector F is selected from the monitoring data. mon ;

[0124] S320, Based on the monitoring feature vector Fmon And the distance and similarity metrics between the two are calculated from the experimental feature vector F;

[0125] S330. Using the obtained distance and similarity metrics, a weighted fusion method is used to comprehensively score the monitored data.

[0126] S340. Based on the comprehensive score, find the closest sample from the comparison database and use the wear label of that sample as the result of the monitoring data.

[0127] In step S320, the distance metric d(F) mon F) is the Euclidean distance d O (F mon F) or Mahalanobis distance d M (F mon The Euclidean distance d O (F mon F) is:

[0128]

[0129] Where m is the dimension of the experimental feature vector;

[0130]

[0131] Where S is the covariance matrix of the experimental feature vectors.

[0132] The covariance matrix S of the experimental feature vector is specifically:

[0133]

[0134] Where L represents the number of samples in the comparison database.

[0135] F i To compare the feature vector of the i-th sample in the database,

[0136] μ is the mean of the feature vectors of the samples in the comparison database.

[0137] The mean μ of the sample feature vectors in the comparison database is:

[0138]

[0139] The similarity metric is cosine similarity sim(F). mon F i Specifically:

[0140]

[0141] In step S330, the comprehensive score Si Specifically:

[0142] S i =w d ·d(F mon F i )+w s ·(1-sim(F mon F i )),

[0143] Among them, w d and w s These are the weights of the distance metric and the similarity metric, respectively, and w d +w s =1.

[0144] In step S340, the evaluation based on the comprehensive score is as follows:

[0145]

[0146] Among them, L best The wear label is for the closest specimen found from the comparison database.

[0147] Step S400 is followed by:

[0148] S500. Establish a machine learning model and train it based on the sample data in the comparison database to predict the wear state of the skateboard.

[0149] In step S500, a machine learning model is trained using a support vector machine (SVM), and the machine learning model is as follows:

[0150]

[0151] Where, α i For Lagrange multipliers,

[0152] y i For wear and tear labels,

[0153] K(F i F j ) is the kernel function.

[0154] b is the bias term.

[0155] L represents the number of samples in the comparative database.

[0156] In SVM, the original problem is to find the optimal hyperplane by minimizing an objective function. The objective function for training the machine learning model is:

[0157]

[0158] Where w is the weight vector,

[0159] b is the bias term.

[0160] ξ i As slack variables,

[0161] C is the penalty parameter.

[0162] The constraints on the objective function of its machine learning model training are:

[0163] y i (w·F i +b)≥1-ξ i ξ i ≥0

[0164] To solve the primal problem, we transform it into a dual problem by introducing the Lagrange multiplier α. i Solve the Lagrange function for:

[0165]

[0166] Where, α i For Lagrange multipliers,

[0167] r i For the slack variables, use the Lagrange multipliers.

[0168] The weight vector w is:

[0169]

[0170] The objective function of the dual problem is:

[0171]

[0172] Wherein K(F) i F j ) is the kernel function.

[0173] The constraints on the objective function of the dual problem are as follows:

[0174]

[0175] The bias term b is:

[0176]

[0177] Example 2

[0178] This invention provides another method for monitoring the wear condition of bridge bearing sliding plates, specifically including the following steps:

[0179] S100. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups. Each group of samples is subjected to abrasion test on the testing machine. According to the test results, a training label is applied. All test data are collected to form a comparison database.

[0180] S200: A force sensor, an acoustic emission sensor, and a data collector are installed on the intermediate liner of the support to collect data on the force and wear of the slide plate during the operation of the support.

[0181] S300: After analyzing and processing the support stress and slide plate wear data, the monitoring data is obtained. The corresponding test data is retrieved from the comparison database and the monitoring data is compared with the test data.

[0182] S400: Based on the comparison results, call the training labels in the test data as the output of the wear status of the support slide plate;

[0183] S500. Establish a machine learning model and train it based on the sample data in the comparison database to predict the wear state of the skateboard.

[0184] In step S500, the machine learning model is trained using a random forest. By integrating multiple decision trees, the variance of a single decision tree is reduced, thereby improving the generalization ability of the model. Each decision tree is trained on a different subset of data, and the selection of features at each node is also randomized.

[0185] In step S500, for classification tasks, the prediction results are summarized using majority voting, specifically as follows:

[0186]

[0187] Wherein, P(Y|F mon Given a feature vector of monitoring data, the probability that the wear label is Y.

[0188] T is the number of decision trees in the random forest.

[0189] h t (F mon Let be the prediction result of the t-th decision tree for the feature vector.

[0190] I(h t (F mon )=Y) is an indicator function. The value of this function is 1 when the prediction result of the t-th decision tree is equal to the wear label, and 0 otherwise.

[0191] Example 3

[0192] This invention provides a bridge bearing sliding plate wear monitoring system, specifically comprising:

[0193] Test module: Used to select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups, each group of samples is subjected to abrasion test on the testing machine, training labels are marked according to the test results, and all test data are collected to form a comparison database;

[0194] Data acquisition module: Used to install force sensors, acoustic emission sensors and data collectors on the intermediate liner of the support to collect force and slide plate wear data during the operation of the support;

[0195] Data processing module: used to analyze and process the support stress and slide plate wear data to obtain monitoring data, retrieve the corresponding test data from the comparison database, and compare the monitoring data with the test data;

[0196] Output module: Used to retrieve the training labels from the experimental data based on the comparison results, and output the wear status of the support slide plate;

[0197] Prediction module: Used to build a machine learning model, which is trained based on sample data in the comparison database to predict the wear condition of the skateboard.

[0198] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method of monitoring the wear condition of a bridge bearing slide plate, characterized by, Includes the following steps: S100. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups. Each group of samples is subjected to abrasion test on the testing machine. According to the test results, a training label is applied. All test data are collected to form a comparison database. S200: A force sensor, an acoustic emission sensor, and a data collector are installed on the intermediate liner of the support to collect data on the force and wear of the slide plate during the operation of the support. S300: After analyzing and processing the support stress and slide plate wear data, the monitoring data is obtained. The corresponding test data is retrieved from the comparison database and the monitoring data is compared with the test data. S400: Based on the comparison results, call the training labels in the test data as the output of the wear status of the support slide plate; Step S100 specifically includes the following steps: S110. The wear stages of the skateboard are divided into three stages: initial normal sliding wear, intermediate dry friction, and later steel-to-steel friction. S120. Select the same sliding plate material as the support, divide the sliding plate material into multiple samples, then divide them into multiple large sample groups, and each large sample group is further divided into three small sample groups. S130. Provide the characteristics of three wear stages for the samples in the three subgroups of each large sample group, and then conduct wear tests on each large sample group according to different characteristic parameters. S140. Acoustic emission sensors are used to collect and record the AE signal data of each sample, and wavelet packet transform is used to filter and perform eigenvalue analysis on the collected AE signals. S150. Integrate and analyze the data, label the samples with different test results with training labels, and collect the test data of all samples to form a comparison database. Step S140 specifically includes the following steps: S141. Perform preprocessing on the acquired AE signal data, including DC bias and filtering, and then select a suitable wavelet basis according to the characteristics of the AE signal data. S142. Based on the bandwidth and frequency resolution of the AE signal data, select the number of decomposition layers of the signal and decompose the AE signal data into AE sub-signals with multiple frequency bands. S143. Extract test features useful for evaluating the wear state of the skateboard from each AE sub-signal, and combine the extracted features into a test feature vector.

2. The method of claim 1, wherein, In step S142, the AE signal data decomposition process is as follows: ， ， wherein is the low frequency portion of the approximation coefficients of the layer, For the high frequency part, the details of the layer are coefficient of the layer, impulse response of a low-pass filter, impulse response of a high pass filter, for the input signal, is an index of the sub-signal; For multi-level decomposition, each level of decomposition will generate two sub-signals, specifically: ， 。 3. A method of monitoring the condition of a bridge bearing slide plate according to claim 2, wherein In step S143, the extracted experimental features include energy features. Root Mean Square Characteristics and kurtosis characteristics The energy characteristics for: ， wherein is the coefficient of the th sub-signal; The root mean square characteristic is: ， wherein is the length of the sub-signal; The kurtosis feature is: ， wherein is the mean value of the th sub-signal.

4. The method of claim 3, wherein the method further comprises: The test feature vector is composed of energy features , root mean square features and kurtosis features, specifically: 。 5. A method of monitoring the wear condition of a bridge bearing slide plate according to any one of claims 1-3, characterized in that, Step S300 specifically includes: S310, after the skateboard wear data is processed according to the method in step S140, monitoring data is obtained, and a monitoring feature vector is screened from the monitoring data ; S320、according to the monitoring feature vector and the test feature vector calculate the distance measure and the similarity measure between them; S330. Using the obtained distance and similarity metrics, a weighted fusion method is used to comprehensively score the monitored data. S340. Based on the comprehensive score, find the closest sample from the comparison database and use the wear label of that sample as the result of the monitoring data.

6. A method of monitoring the condition of a bridge bearing slide plate according to claim 5, wherein In step S320, the distance metric Euclidean distance or Mahalanobis distance The Euclidean distance for: ， wherein, is the dimension of the test feature vector; ， wherein is the covariance matrix of the test eigenvectors; the mean of the sample feature vectors in the comparison database is: ； The similarity measure is a cosine similarity In particular: 。 7. The method for monitoring the wear condition of bridge bearing sliding plates according to claim 6, characterized in that, In step S330, the comprehensive score Specifically: ， in, and These are the weights for the distance metric and the similarity metric, respectively. .

8. In step S340, the evaluation based on the comprehensive score is as follows: ， wherein wear label of the closest specimen found in the contrast database.

9. A bridge bearing slide plate wear condition monitoring system for implementing a bridge bearing slide plate wear condition monitoring method as claimed in any one of claims 1 to 7, characterized by, include: Test module: Used to select the same sliding plate material as the support, divide the sliding plate material into multiple samples and then into multiple groups, and conduct abrasion tests on each group of samples on the testing machine. Training labels are applied according to the test results, and all test data are collected to form a comparison database. Data acquisition module: Used to install force sensors, acoustic emission sensors and data collectors on the intermediate liner of the support to collect force and slide plate wear data during the operation of the support; Data processing module: used to analyze and process the support stress and slide plate wear data to obtain monitoring data, retrieve the corresponding test data from the comparison database, and compare the monitoring data with the test data; Output module: Used to retrieve the training labels from the experimental data based on the comparison results, and output the wear status of the support slide plate; Prediction module: Used to build a machine learning model, which is trained based on sample data in the comparison database to predict the wear condition of the skateboard.

Images