Steel rail magnetic flux leakage detection damage classification method and device and electronic equipment

Abstract

The invention provides a steel rail magnetic flux leakage detection flaw classification method and device and electronic equipment.The steel rail magnetic flux leakage detection flaw classification method comprises the steps that magnetic flux leakage signals of the surface and the near surface of a to-be-detected steel rail are collected, and the magnetic flux leakage signals are preprocessed; calling a kernel function, and determining an optimal parameter of the kernel function; performing feature extraction on the magnetic flux leakage signal through a kernel principal component analysis method based on the kernel function to obtain principal component features; and classifying the extracted principal component features through a support vector machine, and determining the damage category of the to-be-detected steel rail. According to the method and the device, the detected magnetic flux leakage signal data can be processed in real time, the flaw classification result can be quickly given, the detection efficiency and accuracy are remarkably improved, and the problem of poor magnetic flux leakage detection accuracy in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rail detection, and in particular to a rail magnetic flux leakage detection damage classification method and device and electronic equipment. BACKGROUND

[0002] As an important part of rail transportation infrastructure, the surface and internal damage of the rail will directly affect the safety of train operation. With the growth of rail transportation mileage, the demand for rail detection and maintenance is increasing. Traditional manual detection methods have low efficiency, high cost, and uneven quality, so automated and intelligent detection methods have become a research hotspot.

[0003] The core principle of magnetic flux leakage detection technology is to identify defects by detecting the change of the magnetic flux leakage field near the surface of the rail. In the field of rail magnetic flux leakage detection, existing technologies mainly rely on magnetic flux leakage detection methods to identify and classify different types of rail damage. This method uses the magnetization of the rail under an external excitation magnetic field. When the magnetic force line passes through the defect area, due to the difference in magnetic permeability between the air inside the defect and the rail, the magnetic force line leaks outside the rail to form a magnetic flux leakage field. By detecting the change of the magnetic flux leakage field near the surface of the rail, defects can be identified. However, this method has some challenges and shortcomings in classifying different types of damage.

[0004] Although the magnetic flux leakage detection technology has the advantages of no pollution, no need for coupling agent, fast, high reliability, etc., there are still some problems in actual application. First, the fluctuation of the sensor lift-off value has a great influence on the measured magnetic flux leakage signal. Second, the non-single mapping relationship between the magnetic flux leakage signal and the defect parameters leads to unsatisfactory results from the above analysis of traditional features. In addition, the existing technology has insufficient recognition accuracy and stability for rail damage, which is difficult to meet the high requirements of high-speed rail safe operation. For example, the ultrasonic phased array technology can detect complex structure welds and can image the results, but cannot completely determine the source and position of the signal.

[0005] Magnetic flux leakage detection technology usually extracts the two-dimensional magnetic flux leakage field signal of the horizontal component Bx and the normal component Bz of the defect magnetic flux leakage field for defect recognition. However, for regular defects with the direction of the external magnetic field perpendicular to the main surface of the defect, the recognition rate is high. When the main plane of the defect is not perpendicular to the external magnetic field, or the actual defect is irregular in shape, the traditional two-dimensional magnetic flux leakage field signal measurement method will not be able to accurately recognize the shape and size of the defect.

[0006] There is no effective solution to the problem of poor magnetic flux leakage detection accuracy in the existing related technology. SUMMARY

[0007] The application provides a rail magnetic flux leakage detection damage classification method, device and electronic equipment to solve the defects of poor magnetic flux leakage detection accuracy in the prior art.

[0008] In a first aspect, the application provides a rail magnetic flux leakage detection damage classification method, comprising: Collecting the magnetic flux leakage signals of the surface and near-surface of the rail to be detected, and preprocessing the magnetic flux leakage signals; Calling a kernel function and determining the optimal parameters of the kernel function; Extracting features from the magnetic flux leakage signals based on the kernel principal component analysis method of the kernel function to obtain principal component features; Classifying the extracted principal component features by a support vector machine to determine the damage category of the rail to be detected.

[0009] According to the rail magnetic flux leakage detection damage classification method provided by the application, the magnetic flux leakage signals of the surface and near-surface of the rail to be detected are collected, comprising: The magnetic flux leakage signals of the surface and near-surface of the rail to be detected are captured by a magnetizer and an array Hall sensor preset on the rail to be detected; The array Hall sensor is located between the sliding shoe of the rail to be detected and the magnetizer.

[0010] According to the rail magnetic flux leakage detection damage classification method provided by the application, the magnetic flux leakage signals are preprocessed, comprising: Extracting the effective frequency band in the magnetic flux leakage signals by a multi-stage band-pass filter to remove high-frequency noise and low-frequency interference components in the magnetic flux leakage signals; Amplifying the amplitude of the magnetic flux leakage signals by an amplification circuit.

[0011] According to the rail magnetic flux leakage detection damage classification method provided by the application, the kernel function is called, and the optimal parameters of the kernel function are determined, comprising: Obtaining magnetic flux leakage sample data, and dividing the magnetic flux leakage sample data into a training set and a test set; Setting the value range of the width parameter of the kernel function; Based on the kernel function, extracting the sample features of the training set by the kernel principal component analysis method, and classifying by the support vector machine; Determining the optimal parameters of the kernel function based on the accuracy of classification, and evaluating the performance of the kernel function by the validation set.

[0012] According to the rail magnetic flux leakage detection damage classification method provided by the application, the features of the magnetic flux leakage signals are extracted by the kernel principal component analysis method based on the kernel function to obtain principal component features, comprising: The leakage magnetic signal is subjected to mean value processing to eliminate DC bias in the leakage magnetic signal. The kernel matrix corresponding to the leakage magnetic signal is determined through the kernel function. Eigenvalues and eigenvectors of the kernel matrix are solved; the eigenvalues are used to represent the direction of principal components, and the eigenvectors are used to represent the weight of principal components. The principal component characteristics of the leakage magnetic signal are determined according to the size of the eigenvalues of the leakage magnetic signal.

[0013] According to the steel rail leakage magnetic detection damage classification method provided by the application, the kernel matrix corresponding to the leakage magnetic signal is determined through the kernel function, which comprises: The leakage magnetic signal is mapped to a high-dimensional feature space through the kernel function. The kernel values between each data point of the leakage magnetic signal are calculated through the kernel function, and a kernel matrix is constructed based on the kernel values.

[0014] According to the steel rail leakage magnetic detection damage classification method provided by the application, the kernel function is a radial basis function.

[0015] According to the steel rail leakage magnetic detection damage classification method provided by the application, the principal component characteristics are classified through a support vector machine to determine the damage category of the steel rail to be detected, which comprises: An optimal classification hyperplane is determined by maximizing the classification interval. The principal component characteristics are classified by combining the optimal classification hyperplane and introducing a slack variable and a kernel technique to determine the damage category of the steel rail to be detected.

[0016] In a second aspect, the application further provides a steel rail leakage magnetic detection damage classification device, which comprises: A collection module is configured to collect leakage magnetic signals on the surface and near the surface of a steel rail to be detected, and to pre-process the leakage magnetic signals. An optimization module is configured to call a kernel function and determine optimal parameters of the kernel function. An extraction module is configured to extract principal component characteristics from the leakage magnetic signals through a kernel principal component analysis method based on the kernel function. A classification module is configured to classify the extracted principal component characteristics through a support vector machine to determine the damage category of the steel rail to be detected.

[0017] In a third aspect, the application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steel rail leakage magnetic detection damage classification method according to the first aspect.

[0018] In a fourth aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the rail flaw classification method based on magnetic flux leakage detection as described in the first aspect above.

[0019] In a fifth aspect, the present application also provides a computer program product comprising a computer program which, when executed by a processor, implements the rail flaw classification method based on magnetic flux leakage detection as described in the first aspect above.

[0020] Compared with the prior art, the present application has the following beneficial effects: The rail flaw classification method based on magnetic flux leakage detection provided by the present application can solve the problem of sensor lift-off value fluctuation. The traditional method is limited by the sensitivity of signal amplitude to lift-off height. The present application extracts the principal component features of the magnetic flux leakage signal, constructs a feature set with lift-off invariance, and combines the kernel space mapping capability of the support vector machine to learn a feature representation that is robust to lift-off changes. The nonlinear characteristics of the kernel function can identify and ignore the feature variation caused by the lift-off change, and focus on the extraction of the essential features of the damage, thereby maintaining stable classification performance under varying detection conditions. The combination of kernel principal component analysis and support vector machine classification can process the detected magnetic flux leakage signal data in real time, quickly give the damage classification result, significantly improve the detection efficiency and accuracy, and solve the problem of poor magnetic flux leakage detection accuracy in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0022] Figure 1 is a flowchart of the rail flaw classification method based on magnetic flux leakage detection provided by the present application; Figure 2 is a process schematic diagram of classifying the magnetic flux leakage signal in the embodiment of the present application; Figure 3 is a rail flaw classification device based on magnetic flux leakage detection provided by the present application; Figure 4 is a structural schematic diagram of an electronic device provided by the present application. DETAILED DESCRIPTION

[0023] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are only a part of, rather than all of, the embodiments of the present application. Based upon the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts should fall into the protective scope of the present application.

[0024] The present application provides a rail magnetic flux leakage detection damage classification method, Figure 1 The present application provides a rail magnetic flux leakage detection damage classification method, Figure 1 The present application provides a rail magnetic flux leakage detection damage classification method, Step S101, collecting the magnetic flux leakage signals of the surface and near-surface of the to-be-tested rail and pre-processing the magnetic flux leakage signals; Step S102, calling a kernel function and determining the optimal parameters of the kernel function; Step S103, extracting the features of the magnetic flux leakage signals by a kernel principal component analysis (KPCA) method based on the kernel function to obtain principal component features; Step S104, classifying the extracted principal component features by a support vector machine (SVM) to determine the damage category of the to-be-tested rail.

[0025] In the present method, the magnetic flux leakage signal collection device can be integrated on the rail detection vehicle. First, the rail detection vehicle moves along the railway under the traction of the train to collect the magnetic flux leakage signals of the to-be-tested rail. Then, the signal conditioning circuit preliminarily processes the magnetic flux leakage signals, and transmits the magnetic flux leakage signals to the high-speed industrial computer through the data acquisition card. Next, the kernel function is called, and the optimal parameters of the kernel function are determined through experiments to improve the effectiveness of the kernel function. Then, the damage features of the to-be-tested rail are extracted by the kernel principal component analysis method to obtain the principal component features. Finally, the principal component features are classified by the support vector machine to obtain the classification results for the reference of the maintenance personnel.

[0026] In the above process, for the sensor lift-off value fluctuation problem, the traditional method is limited by the sensitivity of signal amplitude to lift-off height, the method extracts the principal component feature of the magnetic flux leakage signal, constructs a feature set with lift-off invariance, and combines the kernel space mapping ability of the support vector machine to learn a feature representation that is robust to lift-off changes. The nonlinear characteristics of the kernel function can identify and ignore the feature variation caused by the lift-off change, and focus on the extraction of the essence of the damage, so as to maintain stable classification performance under changing detection conditions. The combination of kernel principal component analysis and support vector machine classification can process the detected magnetic flux leakage signal data in real time, quickly give the damage classification result, significantly improve the detection efficiency and accuracy, and solve the problem of poor magnetic flux leakage detection accuracy in the existing related technology.

[0027] In some embodiments, the step S101 of collecting the magnetic flux leakage signal of the surface and near-surface of the steel rail to be detected includes: capturing the magnetic flux leakage signal of the surface and near-surface of the steel rail to be detected through the magnetizer and the array Hall sensor preset on the steel rail to be detected; and the array Hall sensor is located between the skid shoe and the magnetizer of the steel rail to be detected.

[0028] In the embodiment, the magnetizer and the array Hall sensor are both integrated on the flaw detection vehicle. The design principle of the magnetizer is to cut the defect edge with magnetic lines, so as to generate magnetic leakage at the defect position. The array Hall sensor is located between the magnetizer and the skid shoe, which can collect the magnetic flux leakage signal in real time and pre-process the signal through the signal conditioning circuit.

[0029] In some embodiments, the step S101 of pre-processing the magnetic flux leakage signal includes: extracting the effective frequency band in the magnetic flux leakage signal through a multi-stage band-pass filter to remove high-frequency noise and low-frequency interference components in the magnetic flux leakage signal; and performing amplitude amplification processing on the magnetic flux leakage signal through an amplification circuit.

[0030] Specifically, the magnetic flux leakage signal is easily affected by environmental noise and electromagnetic interference during the collection process, resulting in mixed high-frequency noise and low-frequency interference in the magnetic flux leakage signal. The multi-stage band-pass filter accurately extracts the effective frequency band of the magnetic flux leakage signal, removes high-frequency noise and low-frequency interference components in the signal. By setting appropriate band-pass filter parameters, the characteristic information of the magnetic flux leakage signal can be preserved, while removing useless interference signals and improving the signal-to-noise ratio of the signal.

[0031] The acquired magnetic flux leakage signals typically have low amplitudes, making them difficult to use directly for subsequent analysis and processing. The amplification circuit amplifies the signal appropriately, bringing it to a suitable amplitude range for the data acquisition card. The design of the amplification circuit needs to balance linear amplification and stability of the magnetic flux leakage signal, ensuring that the amplified signal accurately reflects the characteristics of the original signal. After preprocessing by the signal conditioning circuit, the signal data acquisition card transmits the processed magnetic flux leakage signal to a high-speed industrial computer for further analysis. This significantly improves the quality of the magnetic flux leakage signal, providing a reliable data foundation for subsequent feature extraction and classification analysis.

[0032] In some embodiments, step S102, which involves calling a kernel function and determining the optimal parameters of the kernel function, includes: acquiring magnetic flux leakage sample data and dividing the magnetic flux leakage sample data into a training set and a test set; setting the range of values ​​for the width parameter of the kernel function; extracting sample features from the training set using kernel principal component analysis based on the kernel function, and classifying them using a support vector machine; determining the optimal parameters of the kernel function based on the classification accuracy, and evaluating the performance of the kernel function using a validation set.

[0033] For example, firstly, a large amount of magnetic leakage sample data of rail damage of different types and degrees is collected and divided into training and testing sets. Then, the kernel function width parameter is set. The value range is [0.1, 1, 10, 100, 1000]. Then, k-fold cross-validation is used to calculate different... The average classification accuracy at each value. The values ​​are used to train a support vector machine using the training set and evaluate its performance using the validation set. The value that results in the highest classification accuracy is selected. The value is used as the optimal parameter.

[0034] In some embodiments, step S103, which involves extracting features from the leakage magnetic signal using a kernel principal component analysis method based on kernel functions to obtain principal component features, includes: averaging the leakage magnetic signal to eliminate DC bias; determining the kernel matrix corresponding to the leakage magnetic signal using a kernel function; solving for the eigenvalues ​​and eigenvectors of the kernel matrix; using eigenvalues ​​to characterize the direction of the principal components and eigenvectors to characterize the weights of the principal components; and determining the principal component features of the leakage magnetic signal based on the magnitude of the eigenvalues.

[0035] In the embodiment, the extracted characteristic values include the following 9 key parameters: peak-to-peak value, peak distance, peak-to-peak slope, peak area, mean value, root mean square (RMS), kurtosis, crest factor, and skewness. These characteristic values can comprehensively reflect the characteristics of rail damage and provide a rich information base for support vector machine classification.

[0036] Specifically, the kernel matrix corresponding to the magnetic flux leakage signal is determined by the kernel function, including: mapping the magnetic flux leakage signal to a high-dimensional feature space by the kernel function; calculating the kernel value between each data point of the magnetic flux leakage signal by the kernel function, and constructing the kernel matrix based on the kernel value.

[0037] Exemplarily, before feature extraction, the collected magnetic flux leakage signal is first subjected to mean removal processing. The purpose of mean removal processing is to set the mean value of the magnetic flux leakage signal data to zero, eliminate the direct current bias in the magnetic flux leakage signal, and make the magnetic flux leakage signal more suitable for subsequent feature extraction operations. Assuming that the original magnetic flux leakage signal data is , the steps of mean removal processing are as follows:

[0038]

[0039] wherein, represents the mean value of the magnetic flux leakage signal data, n represents the number of data points, represents the magnetic flux leakage signal data after mean removal.

[0040] The core of the kernel principal component analysis method is to map the original data to a high-dimensional feature space by a kernel function. Preferably, the kernel function in the embodiment is a radial basis function (RBF), which has good nonlinear mapping ability. The formula of the kernel function is as follows:

[0041] wherein, represents the kernel function, and both represent the data points after mean removal, represents a width parameter of the kernel function, which determines the smoothness of mapping to a high-dimensional space. Then, the kernel value between the data points is calculated by the kernel function, and the kernel matrix K is constructed. The kernel matrix is an n x n symmetric matrix, and the elements of the kernel matrix are represents the kernel value between data points and The formula of the kernel matrix is as follows:

[0042] wherein, K represents the kernel matrix. The kernel matrix K is subjected to eigen decomposition to solve its eigenvalue and eigenvector. The eigenvalue represents the direction of the principal component, and the eigenvector represents the weight of the principal component. The specific solving formula is as follows:

[0043] wherein, K represents the kernel matrix, represents the corresponding eigenvector, represents the eigenvalue. By solving the above eigen equation, the eigenvalue and eigenvector of the kernel matrix are obtained. Finally, according to the size of the eigenvalue, the eigenvectors corresponding to the first k largest eigenvalues are selected as the principal component features. These principal component features can effectively represent the nonlinear structure of the original data, and the extraction formula of the principal component features is as follows:

[0044] wherein, represents the extracted principal component feature, represents the eigenvector matrix of the kernel matrix, represents the eigenvector.

[0045] In some embodiments, step S104, the extracted principal component features are classified by a support vector machine to determine the damage category of the to-be-tested rail, including: determining an optimal classification hyperplane by maximizing the classification interval; introducing a slack variable and a kernel technique, and combining the optimal classification hyperplane to classify the principal component features to determine the damage category of the to-be-tested rail. In this embodiment, the nonlinear features of the data can be effectively obtained by the nonlinear feature extraction method.

[0046] The classification method is the core link of the rail damage detection, and its accuracy directly determines the performance of the detection system. Since the feature spaces of different damage types may have overlapping regions, the above feature extraction only solves the problem of “how to describe”. Therefore, in this embodiment, as shown in Figure 2 Figure 2 ​is a process diagram for classifying the magnetic flux leakage signal in the embodiment of the present application, which uses support vector machine to classify the principal component features extracted, solving the problem of "how to make decisions". Support vector machine is a supervised learning algorithm based on statistical learning theory, which can effectively classify different damage types by finding the optimal decision boundary. In the case of linear separability, support vector machine finds the optimal classification hyperplane by maximizing the classification interval; in the case of linear non-separability, it converts the non-linear problem into a linear problem for processing by introducing slack variables and kernel skills. This embodiment verifies the influence of different kernel functions and regularization parameters on the classification effect through experiments, and finally determines the best parameters. The experimental results show that support vector machine has high accuracy and stability in rail damage classification, and can effectively identify different types of rail damage.

[0047] In view of the technical difficulty of non-single mapping relationship between magnetic flux leakage signal and defect parameters, the method adopts a multi-dimensional feature fusion strategy, comprehensively represents the diversity of damage by integrating time domain, frequency domain and nonlinear dynamic features. The core advantage of the support vector machine method lies in its structural risk minimization principle based on statistical learning theory, which can effectively handle the intra-class variation and inter-class overlap problems in the feature space by finding the optimal classification hyperplane. In particular, after introducing the kernel skill, the originally complex nonlinear classification problem is converted into a linearly separable problem in a high-dimensional feature space, so that even in the case of different magnetic flux leakage signals generated by the same size defects, the classifier can still establish a reliable classification boundary through the feature transformation implied by the kernel function. This method does not rely on traditional single feature threshold judgment, but automatically discovers the deep correlation between damage features and categories through machine learning, thereby significantly improving the robustness and accuracy of the classification process. These innovations enable the method to effectively overcome the inherent uncertainty problem in magnetic flux leakage detection, providing a more reliable technical solution for rail damage diagnosis.

[0048] In summary, the method combines kernel principal component analysis and support vector machine to achieve efficient and accurate classification of rail damage. Kernel principal component analysis can effectively extract the nonlinear features of the data, mapping the original magnetic flux leakage signal to a high-dimensional feature space, thereby extracting principal component features that can represent damage characteristics. These feature vectors provide an accurate information basis for subsequent damage classification. Support vector machine, based on statistical learning theory, can effectively classify different types of damage features by finding the optimal decision boundary. Even in the case of linear non-separability, support vector machine can convert the problem into a linear problem for processing through kernel skills and other methods. Experimental results show that the method can significantly improve the accuracy and efficiency of rail damage classification, effectively addressing the challenges of existing technologies in rail magnetic flux leakage detection, and providing strong protection for railway transportation safety.

[0049] In practical applications, the method has significant advantages and innovations. First, by optimizing the kernel function and regularization parameter matched by the support vector machine, the accuracy and stability of the rail damage classification are significantly improved, enabling high-precision damage identification capability in complex environments such as high-speed railways. Second, the method is integrated into the rail detection system to achieve automated damage detection and classification, greatly improving detection efficiency, reducing manual intervention, and reducing detection costs. In addition, the rail magnetic flux leakage signal feature classification method based on the support vector machine proposed by the invention extracts nonlinear features through kernel principal component analysis, not only improving the accuracy of classification, but also enhancing the identification ability of different types of damage, effectively distinguishing between scratches, missing blocks, bruises, cracks and other common rail damage types, providing more detailed and accurate damage information for railway maintenance departments, and helping to develop more scientific and reasonable maintenance strategies to ensure the safe and efficient operation of railway transportation.

[0050] The application also provides a rail magnetic flux leakage detection damage classification device. The rail magnetic flux leakage detection damage classification device provided by the application is described below, and the rail magnetic flux leakage detection damage classification device described below can be referred to in conjunction with the rail magnetic flux leakage detection damage classification method described above. Figure 3 The rail magnetic flux leakage detection damage classification device provided by the application is shown in FIG. Figure 3 The device comprises: The acquisition module 301 is configured to acquire the magnetic flux leakage signals of the surface and near-surface of the rail to be tested and pre-process the magnetic flux leakage signals. The optimization module 302 is configured to call the kernel function and determine the optimal parameters of the kernel function. The extraction module 303 is configured to extract the features of the magnetic flux leakage signals by the kernel principal component analysis method based on the kernel function to obtain the principal component features. The classification module 304 is configured to classify the extracted principal component features by the support vector machine to determine the damage category of the rail to be tested.

[0051] In the device, the magnetic flux leakage signal acquisition device can be integrated on the flaw detection vehicle. In use, first, the flaw detection vehicle moves along the railway under the traction of the train, and the magnetic flux leakage signals of the rail to be tested are collected along the road by the acquisition module 301. Then, the signal conditioning circuit preliminarily processes the magnetic flux leakage signals, and transmits the magnetic flux leakage signals to the high-speed industrial computer through the data acquisition card. Next, the optimization module 302 calls the kernel function and determines the optimal parameters of the kernel function through experiments to improve the effectiveness of the kernel function. Then, the extraction module 303 extracts the damage features of the rail to be tested by the kernel principal component analysis method to obtain the principal component features. Finally, the classification module 304 classifies the principal component features by the support vector machine to obtain the classification results for reference by the maintenance personnel.

[0052] In the above process, for the sensor lift-off value fluctuation problem, the traditional method is limited by the sensitivity of signal amplitude to lift-off height, by extracting the principal component features of the magnetic flux leakage signal, a feature set with lift-off invariance is constructed, combined with the kernel space mapping ability of support vector machine to learn the feature representation robust to lift-off change. The nonlinear characteristics of the kernel function can identify and ignore the feature variation caused by the lift-off change, and focus on the extraction of the essence features of the damage, so as to maintain stable classification performance under changing detection conditions. The combination of kernel principal component analysis and support vector machine classification can process the detected magnetic flux leakage signal data in real time, quickly give the damage classification result, significantly improve the detection efficiency and accuracy, and solve the problem of poor magnetic flux leakage detection accuracy in the existing related technology.

[0053] Figure 4 An example of a schematic diagram of the physical structure of an electronic device is shown as Figure 4 The electronic device can include a processor 401, a communications interface 402, a memory 403, and a communications bus 404, wherein the processor 401, the communications interface 402, and the memory 403 communicate with each other through the communications bus 404. The processor 401 can call the logic instructions in the memory 403 to execute the rail magnetic flux leakage damage classification method, which includes: Collecting the magnetic flux leakage signals of the surface and near-surface of the rail to be tested, and pre-processing the magnetic flux leakage signals; Call the kernel function and determine the optimal parameters of the kernel function; Extracting the principal component features of the magnetic flux leakage signals by the kernel principal component analysis method based on the kernel function; Classify the extracted principal component features by support vector machine to determine the damage category of the rail to be tested.

[0054] In addition, the logic instructions in the memory 403 described above can be implemented in the form of a software function unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium, including a plurality of instructions to make a computer device (which can be a personal computer, server, or network device, etc.) execute all or part of the steps of the embodiments of the present application. The foregoing storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disk or optical disk, and various program code storage media.

[0055] In another aspect, the present application also provides a computer program product, the computer program product comprising a computer program, the computer program being stored in a non-transitory computer-readable storage medium, and the computer program being capable of executing the rail magnetic flux leakage detection damage classification method provided by the above-mentioned methods when executed by a processor, the method comprising: collecting a magnetic flux leakage signal of a surface and a near-surface of a rail to be detected, and pre-processing the magnetic flux leakage signal; calling a kernel function, and determining an optimal parameter of the kernel function; extracting features of the magnetic flux leakage signal based on a kernel principal component analysis method of the kernel function, to obtain principal component features; classifying the extracted principal component features by using a support vector machine, to determine a damage category of the rail to be detected.

[0056] In yet another aspect, the present application also provides a non-transitory computer-readable storage medium, which stores a computer program, the computer program being capable of executing the rail magnetic flux leakage detection damage classification method provided by the above-mentioned methods when executed by a processor, the method comprising: collecting a magnetic flux leakage signal of a surface and a near-surface of a rail to be detected, and pre-processing the magnetic flux leakage signal; calling a kernel function, and determining an optimal parameter of the kernel function; extracting features of the magnetic flux leakage signal based on a kernel principal component analysis method of the kernel function, to obtain principal component features; classifying the extracted principal component features by using a support vector machine, to determine a damage category of the rail to be detected.

[0057] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the present embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0058] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0059] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for classifying rail damage based on magnetic flux leakage detection, characterized in that, include: The magnetic flux leakage signals on and near the surface of the rail under test are collected, and the magnetic flux leakage signals are preprocessed. Call the kernel function and determine the optimal parameters for that kernel function; The leakage magnetic field signal is subjected to feature extraction using a kernel principal component analysis method based on the kernel function to obtain principal component features; The extracted principal component features are classified using a support vector machine to determine the damage category of the rail under test.

2. The rail magnetic flux leakage detection damage classification method according to claim 1, characterized in that, Collect magnetic flux leakage signals from the surface and near the surface of the rail under test, including: The magnetic leakage signals on the surface and near surface of the rail under test are captured by a magnetizer and an array of Hall sensors pre-installed on the rail under test. The array of Hall sensors is located between the slipper of the rail under test and the magnetizer.

3. The rail magnetic flux leakage detection damage classification method according to claim 1, characterized in that, The leakage magnetic signal is preprocessed, including: The effective frequency band in the leakage magnetic signal is extracted by a multi-stage bandpass filter to remove high-frequency noise and low-frequency interference components from the leakage magnetic signal. The leakage magnetic signal is amplified by an amplifier circuit.

4. The rail magnetic flux leakage detection damage classification method according to claim 1, characterized in that, Call the kernel function and determine the optimal parameters for the kernel function, including: Obtain magnetic flux leakage sample data, and divide the magnetic flux leakage sample data into a training set and a test set; Define the range of values ​​for the width parameter of the kernel function; Based on the kernel function, the sample features of the training set are extracted using the kernel principal component analysis method, and then classified using the support vector machine. The optimal parameters of the kernel function are determined based on the classification accuracy, and the performance of the kernel function is evaluated using the validation set.

5. The rail magnetic flux leakage detection damage classification method according to claim 1, characterized in that, The leakage magnetic field signal is subjected to feature extraction using a kernel principal component analysis method based on the kernel function to obtain principal component features, including: The leakage magnetic signal is averaged to eliminate the DC bias in the leakage magnetic signal; The kernel matrix corresponding to the leakage magnetic signal is determined by the kernel function. Solve for the eigenvalues ​​and eigenvectors of the kernel matrix; the eigenvalues ​​are used to characterize the orientation of the principal components, and the eigenvectors are used to characterize the weights of the principal components; The principal component features of the leakage magnetic signal are determined based on the magnitude of the characteristic values ​​of the leakage magnetic signal.

6. The rail magnetic flux leakage detection damage classification method according to claim 5, characterized in that, The kernel matrix corresponding to the leakage magnetic signal is determined by the kernel function, including: The leakage magnetic signal is mapped to a high-dimensional feature space using the kernel function. The kernel function is used to calculate the kernel value between each data point of the leakage magnetic signal, and a kernel matrix is ​​constructed based on the kernel value.

7. The rail magnetic flux leakage detection damage classification method according to claim 5, characterized in that, The kernel function is a radial basis function.

8. The rail magnetic flux leakage detection damage classification method according to claim 1, characterized in that, The extracted principal component features are classified using a support vector machine to determine the damage category of the rail under test, including: The optimal classification hyperplane is determined by maximizing the classification margin. By introducing slack variables and kernel tricks, and combining the optimal classification hyperplane to classify the principal component features, the damage category of the rail under test is determined.

9. A rail magnetic flux leakage detection and damage classification device, characterized in that, include: The acquisition module is used to acquire magnetic flux leakage signals from the surface and near-surface of the rail under test, and to preprocess the magnetic flux leakage signals. An optimization module is used to call the kernel function and determine the optimal parameters of the kernel function; The extraction module is used to extract features from the leakage magnetic signal using a kernel principal component analysis method based on the kernel function, thereby obtaining principal component features; The classification module is used to classify the extracted principal component features using a support vector machine to determine the damage category of the rail under test.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the rail magnetic flux leakage detection damage classification method as described in any one of claims 1 to 8.

Images