Intelligent monitoring method and intelligent monitoring system for rail train brake pad

Abstract

The invention provides an intelligent monitoring method and an intelligent monitoring system for a rail train brake pad. The intelligent monitoring method comprises the following steps: collecting state data of a detected brake pad, wherein the state data comprises temperature, pressure and thickness data of the brake pad; screening and clustering the collected state data, and obtaining brake pad state information according to the processed data; performing brake pad state display and alarm processing according to the brake pad state information; wherein the step of screening and clustering the collected state data and obtaining the state information of the brake pad according to the processed data comprises the following steps: extracting feature data from the state data; performing dimension reduction processing on the feature data by adopting a principal component analysis method and a linear discriminant analysis method; and performing iterative clustering on the data subjected to dimension reduction processing to obtain a classification result of the brake pad state. According to the intelligent monitoring system for implementing the method, the monitoring intelligence and reliability are improved, and safe operation of the train is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of railcar braking monitoring technology, and in particular to a method and system for intelligent monitoring of railcar brake pads based on multi-sensor fusion and K-Means algorithm. Background Technology

[0002] With the development of urban rail transit, the role of trams in urban public transportation is becoming increasingly prominent. As a key component of the tram braking system, the performance and condition of brake pads directly affect the braking effect and operational safety of the train.

[0003] Traditional railcar brake pad inspections rely heavily on periodic manual checks and simple threshold judgments. Manual checks are inherently delayed, making it difficult to monitor brake pad status in real time, and their accuracy is significantly affected by human factors. Simple threshold judgments, based on a single parameter, cannot comprehensively reflect the actual condition of the brake pads under complex operating conditions. For example, under different track gradients, operating speeds, and ambient temperatures, the thickness, temperature, and pressure changes of the brake pads are interconnected and highly variable, making it difficult to accurately determine the brake pad condition using a single threshold.

[0004] Traditional thickness measurement methods mainly rely on a single sensor (such as a laser rangefinder or a contact sensor). These methods have the following limitations in practical applications: laser rangefinders have high requirements for surface flatness and are difficult to adapt to complex surfaces or coated brake pads; contact sensors are easily affected by mechanical wear and installation errors, resulting in a decrease in measurement accuracy; existing methods cannot detect defects inside the brake pad (such as cracks, corrosion, etc.), making it difficult to comprehensively assess the condition of the brake pad. Summary of the Invention

[0005] To address the aforementioned technical problems, at least one embodiment of the present invention provides a method for intelligent monitoring of brake pads on railcars and a corresponding intelligent monitoring system. This method is capable of performing fault diagnosis and analysis of brake pads based on multi-sensor fusion and the K-Means algorithm.

[0006] At least one embodiment of the present invention provides a method for intelligent monitoring of brake pads on railcars, characterized in that the method includes the following steps: Collect the status data of the tested brake pad, including the temperature, pressure and thickness data of the brake pad; The collected status data is filtered and clustered, and the gate status information is derived from the processed data. The brake pad status is displayed and alarms are processed based on the brake pad status information. The process of filtering and clustering the collected status data, and deriving the gate plate status information based on the processed data, includes: Extract feature data from state data; Principal component analysis and linear discriminant analysis were used to reduce the dimensionality of the feature data. The data after dimensionality reduction is iteratively clustered to obtain the classification results of the gate state.

[0007] In some optional embodiments, the dimensionality reduction processing of the feature data using principal component analysis and linear discriminant analysis includes: Principal component analysis was used to obtain the principal component data of the feature data. The principal component data were dimensionality reduced using linear discriminant analysis.

[0008] In some optional embodiments, before extracting the corresponding feature data from the state data as described above, the method further includes: Remove outliers from the collected status data.

[0009] In some optional embodiments, before extracting the corresponding feature data from the state data as described above, the method further includes: The thickness data measured by different types of thickness sensors are weighted and fused, and the weighted fusion result is used as the thickness data of the gate plate.

[0010] In some optional embodiments, the above-mentioned weighted fusion of thickness data measured by different types of thickness sensors, and the use of the weighted fusion result as the thickness data of the gate, includes: Calculate the deviation between the measurement results of different types of thickness sensors; When the deviation does not exceed the preset threshold, the measurement results of different types of thickness sensors are weighted and averaged, and the weighted average value is used as the thickness data of the gate plate. When the deviation exceeds a preset threshold, the measurement result of a thickness sensor capable of simultaneously detecting internal defects in the brake pad is used as the thickness data of the brake pad.

[0011] In some alternative embodiments, the laser rangefinder is used to measure the thickness of the brake pad, and the pulsed eddy current thickness sensor is used to non-contactly measure the thickness of the brake pad and detect internal defects in the brake pad.

[0012] In some optional embodiments, the above-described extraction of feature data from state data includes: The thickness data is used to extract the average thickness to reflect the overall wear condition, the thickness standard deviation to assess wear uniformity, and the thickness change slope to determine the wear rate change. The average temperature is extracted from the temperature data to understand the heating level of the gate, the highest temperature to determine the local overheating situation, and the temperature gradient to assess the uniformity of heat dissipation. The pressure data is used to extract the average pressure value for understanding the average pressure, the peak pressure value for assessing extreme pressure tolerance, and the pressure fluctuation frequency for analyzing pressure stability.

[0013] At least one embodiment of the present invention also provides an intelligent monitoring system for railcar brake pads, characterized in that it includes: The data acquisition module is used to acquire the status data of the tested brake pad, including the temperature, pressure and thickness data of the brake pad; The storage and processing module is used to filter and cluster the status data collected by the data acquisition module, and to derive the gate status information based on the processed data. The display and alarm module is used to display and alarm the status of the gate piece based on the gate piece status information provided by the storage and processing module. The control and management module is connected to the data acquisition module, storage and processing module, and display and alarm module. It is used to monitor the operation, self-test and maintenance of each module, as well as to process fault data and adjust relevant parameters. The storage and processing module includes: The feature extraction unit is used to extract feature data from the state data; The feature selection and dimensionality reduction unit is used to perform dimensionality reduction processing on the feature data according to principal component analysis and linear discriminant analysis. The clustering unit is used to iteratively cluster the feature data after dimensionality reduction to obtain the classification results of the gate state.

[0014] In some optional embodiments, the data acquisition module described above includes a temperature sensor, a pressure sensor, and a thickness sensor, wherein: The thickness sensor includes a laser rangefinder and a pulsed eddy current thickness sensor. The laser rangefinder is used to measure the thickness of the brake pad, and the pulsed eddy current thickness sensor is used to non-contactly measure the thickness of the brake pad and detect internal defects in the brake pad.

[0015] In some optional embodiments, the above data storage and processing module further includes: The weighted detection unit is used to perform weighted fusion of the thickness data measured by the laser rangefinder and the thickness data measured by the pulsed eddy current thickness sensor.

[0016] Compared with the prior art, the intelligent monitoring method and intelligent monitoring system for railcar brake pads provided by the embodiments of the present invention have the following beneficial effects: The embodiments of the present invention propose a multi-sensor fusion thickness detection scheme, which uses a laser rangefinder to measure the surface thickness of the gate plate and a pulsed eddy current thickness sensor to non-contactly measure the thickness of the gate plate and detect internal defects. By performing weighted fusion processing on the laser thickness measurement and pulsed eddy current thickness measurement data, the accuracy and reliability of thickness detection can be significantly improved. The embodiments of this invention employ the K-Means algorithm as an effective clustering algorithm to classify multi-dimensional data from tram brake pad condition monitoring. This overcomes the limitations of traditional detection methods, enabling intelligent and precise monitoring, improving braking system safety, and optimizing train maintenance plans and operational management. Compared with other algorithms used for similar monitoring, the K-Means algorithm has advantages such as higher computational efficiency, relatively simple principles, and strong interpretability, making it more suitable for tram brake pad condition monitoring. The embodiments of the present invention allow users to view historical data of brake pads and summarize wear patterns through the system's human-computer interaction function. They can also adjust K-Means algorithm parameters and alarm thresholds according to actual needs. The system can diagnose and maintain functional units, monitor the operating status of hardware devices and software algorithms in real time, predict and warn of possible faults in advance, ensure stable system operation, improve the intelligence and reliability of railcar brake pad monitoring, and effectively ensure the safe operation of trains. Attached Figure Description

[0017] One or more embodiments are illustrated by way of example with reference to the accompanying drawings, and these illustrative descriptions do not constitute a limitation on the embodiments.

[0018] Figure 1 This is a schematic diagram of the overall architecture of the intelligent monitoring system for railcar brake pads used in an embodiment of the present invention; Figure 2 This is a schematic diagram of the data preprocessing and feature extraction process used in an embodiment of the present invention; Figure 3 This is a schematic diagram of the final clustering distribution of fault data in the K=4 fault diagnosis model used in this embodiment of the invention; Figure 4 This is a schematic diagram of the monitoring system used in the embodiments of the present invention and its self-testing and diagnostic functions. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the various embodiments of the present invention will be described in detail below with reference to the accompanying drawings. However, those skilled in the art will understand that many technical details are presented in the embodiments of the present invention to facilitate a better understanding of the invention. However, the technical solutions claimed in the present invention can be implemented even without these technical details and various variations and modifications based on the following embodiments. The division of the following embodiments is for ease of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined with and referenced by each other without contradiction.

[0020] The implementation details of the above method are described in detail below through examples. The following content is only for the convenience of understanding the implementation details and is not necessary for implementing this solution.

[0021] Example 1 As mentioned above, the present invention aims to provide a method and system for intelligent monitoring of brake pads on railcars.

[0022] Figure 1 This is a diagram illustrating the overall architecture of a railcar brake pad fault monitoring system. Figure 1 As shown, the intelligent monitoring system for railcar brake pads mainly consists of a data acquisition module, a storage and processing module, a display and alarm module, and a control and management module.

[0023] In this embodiment, the high-precision sensor group of the data acquisition module includes thickness, temperature, and pressure sensors.

[0024] In this embodiment, the thickness sensor includes a laser rangefinder and a pulsed eddy current thickness sensor. The laser rangefinder measures the thickness of the gate plate surface by emitting a laser beam with an accuracy of ±0.2 mm; the pulsed eddy current thickness sensor measures the thickness of the gate plate non-contactly by emitting high-frequency electromagnetic pulses to sense eddy current signals on the gate plate surface, and can also detect internal defects (such as cracks, corrosion, etc.) in the gate plate with an accuracy of ±0.1 mm.

[0025] Furthermore, the system employs a weighted fusion method to process the laser thickness measurement and pulsed eddy current thickness measurement data. The specific steps are as follows: 1. Calculate the difference (i.e., deviation) between laser thickness measurement and pulsed eddy current thickness measurement data.

[0026] 2. If the difference does not exceed the preset threshold (e.g., 0.3 mm), the measurement results of the two sensors are weighted and averaged, with the weights dynamically adjusted according to the accuracy and reliability of the sensors.

[0027] 3. If the difference exceeds the preset threshold, pulsed eddy current thickness measurement data will be used first, because it can detect internal defects and is more adaptable to complex surfaces.

[0028] In this embodiment, the temperature sensor array uses thermocouples to monitor the temperature of key parts of the brake pads, with a measurement range of -20℃ to 800℃; the pressure sensor network collects pressure at the brake caliper, with a range of 0 to 80MPa. A high-speed data acquisition card performs multi-channel synchronous acquisition at a sampling frequency of 8kHz, and transmits the sensor signals to the storage and processing module after conversion and conditioning.

[0029] In this embodiment, the large-capacity storage unit in the storage and processing module uses a solid-state drive array to store data. The status data provided by the data acquisition module needs to be preprocessed. First, the outlier removal unit filters out outliers in the thickness, temperature, and pressure data according to the algorithm. Then, the feature extraction unit extracts various features related to thickness, temperature, and pressure. Next, the feature selection and dimensionality reduction unit uses a combination of PCA and LDA to perform dimensionality reduction on the feature data. Finally, the K-Means algorithm processing unit sets the number of clusters k to 4, corresponding to the four states of the brake pad: "normal", "mild wear", "moderate wear", and "severe wear". The K-Means++ method is used to initialize the cluster centers, and the brake pad state classification results are obtained through iterative clustering calculations, and then stored and transmitted.

[0030] In this embodiment, the intelligent display screen in the display and alarm module shows the brake pad status details in graphics and numbers in the cockpit and maintenance room. If the brake pad is severely worn or faulty, the audible and visual alarm device is activated in the cockpit and sends fault information including the brake pad location, fault type, and fault time to the control center and maintenance department.

[0031] In this embodiment, the human-machine interface console in the control and management module is located in the cockpit or maintenance room, and can perform system self-tests, view historical data, and adjust K-Means algorithm parameters and alarm thresholds. The system self-diagnosis and maintenance unit monitors the system's hardware and software operation in real time. When a sensor malfunctions, it displays fault information, cause analysis, and maintenance suggestions; when the software algorithm malfunctions, it checks and optimizes parameters and data to ensure stable system operation.

[0032] Example 2 This embodiment will use specific examples to illustrate the working process of the railcar brake pad fault monitoring method.

[0033] At the moment of train braking, the data acquisition module responds rapidly. The laser distance sensor measures the brake pad surface thickness with an accuracy of ±0.2 mm by emitting a laser beam. The pulsed eddy current thickness sensor, by emitting high-frequency electromagnetic pulses, senses eddy current signals on the brake pad surface, thus measuring the brake pad thickness non-contactly and detecting internal defects (such as cracks and corrosion). The laser and pulsed eddy current thickness measurement data are then fused using the following method: 1. Calculate the difference between laser thickness measurement and pulsed eddy current thickness measurement data.

[0034] 2. If the difference does not exceed the preset threshold (e.g., 0.3 mm), the measurement results of the two sensors are weighted and averaged, with the weights dynamically adjusted according to the accuracy and reliability of the sensors.

[0035] 3. If the difference exceeds the threshold, pulsed eddy current thickness measurement data should be used first because it can detect internal defects and is more adaptable to complex surfaces.

[0036] By using a weighted detection method, the data from laser thickness measurement and pulsed eddy current thickness measurement are fused into a single weighted thickness value: H = w1hlass + w2hedral; where hlass is laser thickness measurement data and hedral is pulsed eddy current thickness measurement data. w1 and w2 are weights. The weighted thickness data H is a single numerical feature representing the overall thickness of the brake pad (i.e., the thickness data of the brake pad).

[0037] A temperature sensor array is distributed at key contact and heat dissipation points between the brake pads and the brake disc. Utilizing the thermoelectric effect of thermocouples, it senses temperature changes and accurately measures temperatures ranging from -20℃ to 800℃. A pressure sensor network is deployed in the brake caliper hydraulic lines and key pressure transmission nodes. Based on the strain gauge principle, the resistance of the strain gauge changes under stress, which is converted into an electrical signal by the circuit, accurately acquiring pressure information from 0 to 80 MPa. A high-speed data acquisition card synchronously acquires analog signals from each sensor at an 8kHz frequency. These signals are amplified and filtered by the built-in conditioning circuitry before being converted into digital signals and transmitted to the data storage and processing module.

[0038] Figure 2 This is a schematic diagram illustrating the process by which the storage and processing module preprocesses and extracts features from the status data collected by the data acquisition module in this embodiment.

[0039] like Figure 2 As shown, after the storage and processing module receives the status data collected by the data acquisition module, it first performs the following preprocessing on the status data to remove outliers from the data.

[0040] The outlier removal unit uses preset thresholds based on statistical laws and physical properties to rigorously screen thickness, temperature, and pressure data, including: For thickness data, considering the brake pad material, normal wear rate, and historical similar working condition data, if a certain thickness value differs from its adjacent value from the estimated normal wear range under the current working condition, or deviates too much from historical data, it is judged as abnormal and removed. For temperature data, based on the heat resistance of the brake pad, heat dissipation mechanism and normal distribution pattern, if the value of a certain measuring point differs significantly from that of adjacent points and does not conform to the temperature gradient change or exceeds the operating limit temperature, anomaly processing shall be performed. For stress data, analyze its dynamic range, peak reasonableness, and fluctuation frequency and amplitude, and remove data that does not conform to the overall trend or exceeds the threshold range.

[0041] Then, the feature extraction unit extracts feature data from the state data after removing outliers, where: The thickness average value is extracted to reflect the overall wear condition; the thickness standard deviation is extracted to assess the wear uniformity; and the thickness change slope is extracted to determine the recent wear rate change. The average temperature, maximum temperature, and temperature gradient are extracted in the temperature dimension to understand the heating level of the gate, monitor local overheating, and evaluate the uniformity of heat dissipation. The pressure dimension extracts the average pressure, peak pressure, and pressure fluctuation frequency to understand the average pressure, assess extreme pressure tolerance, and analyze pressure stability.

[0042] In the actual operation of this system, the execution process of the PCA and LDA algorithms in the feature selection and dimensionality reduction unit is as follows: First, PCA processing is performed. The original data dimension is assumed to be m1, which covers multi-dimensional information such as thickness, temperature, and pressure collected from sensors. PCA generates a series of principal components by eigenvalue decomposition of the data covariance matrix. The number of principal components to be retained is determined based on the cumulative variance contribution rate. For example, if the cumulative variance contribution rate is set to be above 85%, and it is assumed that m1 principal components are ultimately retained after calculation, the data is initially reduced from n dimensions to m1 dimensions, removing the correlation between features. Next, the LDA algorithm is used to process this m1-dimensional data. Since the brake pad status is divided into four cases: "normal", "mild wear", "moderate wear", and "severe wear", according to the LDA principle, its output dimension is at most the number of categories minus 1, that is, it can be projected to a maximum of 3-dimensional space (m2≤3). In actual operation, the final dimension is determined by combining the data distribution and the K-Means clustering effect. For example, after multiple experiments and analyses, the K-Means algorithm achieved the best classification effect when projected into an m²-dimensional space, thus determining the data dimension after LDA processing to be m²-dimensional. Ultimately, the m²-dimensional data obtained after PCA and LDA processing is used as input to the K-Means algorithm. Selecting dimensional features aims to reduce data dimensionality and computational complexity while effectively distinguishing different gate states, thereby improving algorithm efficiency. When determining the optimal feature dimension combination, the K-Means algorithm's clustering results under different dimension combinations are evaluated by calculating metrics such as the silhouette coefficient and the Calinski-Harabasz index, ensuring the K-Means algorithm accurately classifies gate states and improves the overall monitoring system performance.

[0043] Furthermore, due to the different units and dimensions of thickness, temperature, and pressure, the system also employs Z-score normalization to preprocess the data, ensuring that the mean of each feature is 0 and the standard deviation is 1. The normalized data can then be directly input into clustering units for K-Means algorithm cluster analysis.

[0044] Figure 3 This is a schematic diagram of the final clustering distribution of fault data in the fault diagnosis model with K=4 in this embodiment.

[0045] The K-Means algorithm processing unit performs cluster analysis on the standardized multidimensional data. In this embodiment, the preset cluster number k is 4, corresponding to the "normal state", "slight wear state", "moderate wear state" and "severe wear state" of the brake pads, respectively. The K-Means algorithm performs clustering through the following steps: 1. Initialize cluster centers: The K-Means++ algorithm is used to initialize cluster centers, avoiding the problem of traditional K-Means algorithm getting stuck in local optima due to improper selection of initial centers.

[0046] 2. Iterative clustering: Calculate the Euclidean distance of each data point to the cluster center and assign it to the nearest cluster center; update the position of the cluster center and take the mean of all data points in each cluster as the new cluster center.

[0047] 3. Stopping condition: Stop iteration when the change in cluster centers is less than the preset threshold (0.01) or the maximum number of iterations (200 times) is reached.

[0048] The specific clustering steps are as follows: after initialization, iterative clustering calculation phase begins. Each iteration uses the Euclidean distance formula to accurately calculate the distance from each data point to its cluster center and then classifies it. For each data point... and cluster center Its Euclidean distance: .

[0049] Data points are assigned to the clusters of the nearest cluster centers based on distance. After all data points have been assigned, the locations of the cluster centers are recalculated. For example, if cluster K contains m data points... , ,…, New cluster center In the Each feature dimension coordinate: .

[0050] Where 'n' represents the dimension of the data points, i.e., the dimension of the data used for K-Means clustering analysis after PCA and LDA processing. The data point allocation and cluster center update steps are repeated until the change in cluster centers is minimal (less than 0.01) or the preset number of iterations (e.g., 200 times) is reached. Finally, a stable and reliable gate status classification result is obtained, stored in the data storage unit, and transmitted to the status display and alarm module.

[0051] Figure 4This diagram illustrates the self-checking and diagnostic process of the intelligent monitoring system implemented in this project. After receiving the brake pad status classification results, the display and alarm module activates the information display function on the intelligent display screen. Through a high-resolution LCD screen, the brake pad status information is presented to the driver and maintenance personnel in real-time using intuitive graphics (different colored indicator lights and dynamic charts) and digital reports. In the graphical interface, a bright green indicator light and a stable curve indicate that the brake pad is normal, with light wear and good performance; a flashing yellow indicator light and fluctuating chart indicate moderate wear, requiring further monitoring; a continuously flashing red indicator light and drastic chart changes indicate severe wear or malfunction of the brake pad, endangering braking and driving safety. The digital report area lists detailed characteristic parameter values ​​corresponding to the current status, such as brake pad thickness, average temperature, peak pressure, and their variation curves. This display method allows drivers to promptly understand the brake pad status and performance trends, making appropriate adjustments to driving operations or planning maintenance stops; maintenance personnel can quickly diagnose brake pad faults, prepare maintenance tools and spare parts in advance, and improve maintenance efficiency and accuracy.

[0052] When the K-Means algorithm processing unit determines that the brake pads are in a "severely worn or faulty state", the audible and visual alarm device in the driver's cab and maintenance personnel's work area will be activated immediately. At the same time, detailed information including the precise location, type, and time of occurrence of the faulty brake pads will be sent to the train control center and ground maintenance base via the wireless communication system so that the operation plan can be adjusted and maintenance work can be carried out in a timely manner.

[0053] Drivers and maintenance personnel can view historical brake pad data and summarize wear patterns through the human-machine interface console. They can also adjust K-Means algorithm parameters and alarm thresholds according to actual needs. The system's self-diagnosis and maintenance unit monitors the operating status of hardware and software algorithms in real time, predicting and issuing real-time warnings for potential faults. When hardware malfunctions, fault information and maintenance suggestions are displayed on the human-machine interface console; when software algorithms malfunction, parameters and data are automatically checked and optimized to ensure stable system operation, improve the intelligence and reliability of tram brake pad monitoring, and effectively guarantee train safety.

[0054] In the above embodiments, the storage and processing module may be at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the intelligent monitoring method for railcar brake pads in the above embodiments.

[0055] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and will not be described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0056] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0057] Example 3 Another embodiment of the present invention relates to a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the intelligent monitoring method for railcar brake pads described in the above embodiments.

[0058] That is, those skilled in the art will understand that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0059] Example 4 Another embodiment of the present invention relates to a computer program product, including a computer program that, when executed by a processor, implements the intelligent monitoring method for railcar brake pads described in the above embodiments.

[0060] Those skilled in the art will understand that the above embodiments are specific embodiments for implementing the present invention, and in practical applications, various changes in form and detail may be made without departing from the spirit and scope of the present invention.

Claims

1. A method for intelligent monitoring of brake pads on trams, characterized in that, Includes the following steps: Collect the status data of the tested brake pad, including the temperature, pressure and thickness data of the brake pad; The collected status data is filtered and clustered, and the gate status information is derived from the processed data. The brake pad status is displayed and alarms are processed based on the brake pad status information. The process of filtering and clustering the collected status data, and deriving the gate plate status information based on the processed data, includes: Extract feature data from state data; Principal component analysis and linear discriminant analysis were used to reduce the dimensionality of the feature data. The data after dimensionality reduction is iteratively clustered to obtain the classification results of the gate state.

2. The intelligent monitoring method for railcar brake pads according to claim 1, characterized in that, The dimensionality reduction processing of the feature data using principal component analysis and linear discriminant analysis includes: Principal component analysis was used to obtain the principal component data of the feature data. The principal component data were dimensionality reduced using linear discriminant analysis.

3. The intelligent monitoring method for railcar brake pads according to claim 1, characterized in that, Before extracting the corresponding feature data from the state data, the method further includes: Remove outliers from the collected status data.

4. The intelligent monitoring method for railcar brake pads according to claim 1, characterized in that, Before extracting the corresponding feature data from the state data, the method further includes: The thickness data measured by different types of thickness sensors are weighted and fused, and the weighted fusion result is used as the thickness data of the gate plate.

5. The intelligent monitoring method for railcar brake pads according to claim 4, characterized in that, The weighted fusion of thickness data measured by different types of thickness sensors, and the use of the weighted fusion result as the thickness data of the gate plate, includes: Calculate the deviation between the measurement results of different types of thickness sensors; When the deviation does not exceed the preset threshold, the measurement results of different types of thickness sensors are weighted and averaged, and the weighted average value is used as the thickness data of the gate plate. When the deviation exceeds a preset threshold, the measurement result of a thickness sensor capable of simultaneously detecting internal defects in the brake pad is used as the thickness data of the brake pad.

6. The intelligent monitoring method for railcar brake pads according to claim 5, characterized in that, The thickness of the brake pad is measured using the laser rangefinder sensor, and the thickness of the brake pad is measured non-contactly using the pulsed eddy current thickness sensor to detect internal defects in the brake pad.

7. The intelligent monitoring method for railcar brake pads according to claim 1, characterized in that, The extraction of feature data from the state data includes: The thickness data is used to extract the average thickness to reflect the overall wear condition, the thickness standard deviation to assess wear uniformity, and the thickness change slope to determine the wear rate change. The average temperature is extracted from the temperature data to understand the heating level of the gate, the highest temperature to determine the local overheating situation, and the temperature gradient to assess the uniformity of heat dissipation. The pressure data is used to extract the average pressure value for understanding the average pressure, the peak pressure value for assessing extreme pressure tolerance, and the pressure fluctuation frequency for analyzing pressure stability.

8. A smart monitoring system for brake pads on trams, characterized in that, include: The data acquisition module is used to acquire the status data of the tested brake pad, including the temperature, pressure and thickness data of the brake pad; The storage and processing module is used to filter and cluster the status data collected by the data acquisition module, and to derive the gate status information based on the processed data. The display and alarm module is used to display and alarm the status of the gate piece based on the gate piece status information provided by the storage and processing module. The control and management module is connected to the data acquisition module, storage and processing module, and display and alarm module. It is used to monitor the operation, self-test and maintenance of each module, as well as to process fault data and adjust relevant parameters. The storage and processing module includes: The feature extraction unit is used to extract feature data from the state data; The feature selection and dimensionality reduction unit is used to perform dimensionality reduction processing on the feature data according to principal component analysis and linear discriminant analysis. The clustering unit is used to iteratively cluster the feature data after dimensionality reduction to obtain the classification results of the gate state.

9. The intelligent monitoring system for railcar brake pads as described in claim 8, characterized in that, The data acquisition module includes a temperature sensor, a pressure sensor, and a thickness sensor, wherein: The thickness sensor includes a laser rangefinder and a pulsed eddy current thickness sensor. The laser rangefinder is used to measure the thickness of the brake pad, and the pulsed eddy current thickness sensor is used to non-contactly measure the thickness of the brake pad and detect internal defects in the brake pad.

10. The intelligent monitoring system for railcar brake pads as described in claim 9, characterized in that, The data storage and processing module also includes: The weighted detection unit is used to perform weighted fusion of the thickness data measured by the laser rangefinder and the thickness data measured by the pulsed eddy current thickness sensor.

Images