High-speed rail bearing fault diagnosis method and system

Abstract

The invention discloses a high-speed rail bearing fault diagnosis method and system. The method comprises the following steps: constructing a dynamical model of a bearing according to geometric parameters and fault frequency of the bearing; according to different fault types of the bearing, carrying out local dynamic modeling on the bearing, solving a bearing dynamic equation, and obtaining joint data of the fault bearing; inputting the preprocessed joint data into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model; and acquiring real-time data of the bearing, inputting the real-time data into the high-speed rail bearing fault diagnosis model, and outputting the real-time data by the high-speed rail bearing fault diagnosis model to obtain the fault type of the bearing. The influence of few data samples on the network detection precision can be avoided.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-speed rail bearing detection, and in particular relates to a high-speed rail bearing fault diagnosis method and system. Background Art

[0002] As an important component of the railway vehicle bogie, the bearing must bear the vehicle's load and ensure the smooth rotation of the axle. The stability of its performance directly affects whether the train can run safely. The bearing is composed of different parts, and the fault location is different, and the fault form is also different, which will produce different fault frequencies. Therefore, by processing the collected signals, the type of fault can be inferred, the location of the fault can be identified, and corresponding measures can be taken in time to reduce the frequency of railway accidents and reduce losses.

[0003] Most of the data used in traditional bearing fault diagnosis are obtained in the laboratory, and some are even artificially caused faults, which are far from the real environment. Therefore, knowledge transfer is needed to set up the dynamic model in advance, and transfer the features extracted from the laboratory to the actual data to achieve the general use of the model. Early bearing fault diagnosis methods include ML and DL, but ML has defects such as mutual influence of fault frequency, unstable accuracy, and weak noise resistance. Therefore, ML is gradually replaced by DL. CNN in DL is widely used because of its better noise resistance and robustness, but it itself requires more network layers, large amount of calculation, and requires a large amount of labeled data. In practical applications, most of the network model variants of CNN are used. However, because the recurrent network of RNN is too large, the mutual correlation between information is very high, which is easy to cause data information explosion during model training. Summary of the invention

[0004] The present invention provides a high-speed rail bearing fault diagnosis method and system, which are used to solve the technical problems of insufficient bearing fault experimental data and low network model detection accuracy.

[0005] In a first aspect, the present invention provides a high-speed rail bearing fault diagnosis method, comprising: Construct the dynamic model of the bearing according to the geometric parameters and fault frequency of the bearing; According to different bearing fault types, local dynamic modeling of the bearing is carried out, the bearing dynamic equation is solved, and the joint data of the faulty bearing is obtained; The preprocessed joint data is input into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: , In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: , In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: , In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: , In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; Real-time data of the bearing is obtained, and the real-time data is input into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0006] In a second aspect, the present invention provides a high-speed rail bearing fault diagnosis system, comprising: A construction module configured to construct a dynamic model of the bearing according to geometric parameters and fault frequencies of the bearing; an acquisition module, configured to perform local dynamic modeling on the bearing according to different fault types of the bearing, solve the bearing dynamic equation, and obtain joint data of the faulty bearing; The training module is configured to input the preprocessed joint data into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: , In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: , In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: , In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: , In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; The output module is configured to obtain real-time data of the bearing and input the real-time data into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0007] According to a third aspect, an electronic device is provided, comprising: 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, and the instructions are executed by the at least one processor so that the at least one processor can execute the steps of the high-speed rail bearing fault diagnosis method of any embodiment of the present invention.

[0008] In a fourth aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor executes the steps of the high-speed rail bearing fault diagnosis method of any embodiment of the present invention.

[0009] The high-speed rail bearing fault diagnosis method and system of the present application, when bearing sample data is input, detects the data sample through the fractal attention reconstruction network to determine the type and location of the bearing fault. Compared with the traditional detection method, this method builds a rolling bearing dynamics model based on the bearing's geometric parameters and fault frequency, and generates joint data by combining the dynamics model with the experimental data. The joint data has the same characteristics as the actual experimental data in frequency domain distribution and time domain distribution, which can avoid the small number of data samples affecting the network detection accuracy. By using the fractal attention reconstruction network for fault diagnosis, it is possible to accurately capture microscopic fault features and extract richer signal details. By using a dynamic topological reconstruction method, the diagnostic performance and anti-interference ability of the model under different working conditions are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flow chart of a high-speed rail bearing fault diagnosis method provided by one embodiment of the present invention; Figure 2 A structural block diagram of a high-speed rail bearing fault diagnosis system provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0013] See also Figure 1 , which shows a flow chart of a high-speed rail bearing fault diagnosis method of the present application.

[0014] like Figure 1 As shown, the high-speed rail bearing fault diagnosis method specifically includes the following steps: Step S101, constructing a dynamic model of the bearing according to the geometric parameters and fault frequencies of the bearing.

[0015] In this step, the acquired fault frequencies of the high-speed rail bearings are classified according to the wheel-rail vibration, and the bearing fault types are divided into four types: inner ring fault (IR), outer ring fault (OR), rolling element fault (RB) and normal (N).

[0016] In order to expand the bearing fault data set, the lumped mass method is used for dynamic modeling. The rolling bearing dynamic system is regarded as a four-degree-of-freedom system (only the displacement of the shaft and inner ring, outer ring and bearing base in the two-dimensional coordinate system is considered), and the influence of other factors (such as the number of rolling elements, etc.) is considered in the dynamic modeling to establish a dynamic model. The expression of the inner ring-shaft system in the dynamic model is: In the x-axis direction: , In the formula, is the sum of the masses of the inner ring and the shaft, is the relative displacement of the inner ring in the x direction, is the connection damping between the shafts and the shaft fixings, is the equivalent connection stiffness of the shaft and the shaft fixing point, is the component of the elastic restoring force in the x-axis direction generated by the contact between the inner and outer rings and the rolling elements at time t; In the y-axis direction: , In the formula, is the relative displacement of the inner circle in the y direction, is the acceleration due to gravity, is the wheel-rail excitation coefficient, is the eccentricity, is the angular frequency of the inner ring rotation, For time; The expression of the outer ring-bearing base system in the dynamic model is: In the x-axis direction: , In the formula, is the sum of the masses of the outer ring and the bearing base, is the relative displacement of the outer ring in the x direction, is the connection damping between the bearing base and the ground foundation. is the equivalent connection stiffness between the bearing base and the ground; In the y-axis direction: , In the formula, is the component of the elastic restoring force in the y-axis direction generated by the contact between the inner and outer rings and the rolling elements at time t, is the relative displacement of the outer ring in the y direction.

[0017] It should be noted that, taking the contact between the rolling element and the raceway in the rolling bearing as the linear elastic body restoring force, the expression for calculating the component of the elastic restoring force in the x-axis direction generated by the contact between the inner and outer rings and the rolling element at the time t is: , In the formula, is the load-deformation coefficient, is the load-deformation index, is the position at time t, is the initial position, is the elastic coefficient at time t; The expression for calculating the component of the elastic restoring force in the y-axis direction generated by the contact between the inner and outer rings and the rolling element at the time t is: , In the formula, is the displacement after contact, is the initial displacement.

[0018] In this embodiment, compared with the traditional single modeling method of bearing dynamics, this paper proposes to adopt four-degree-of-freedom coupling modeling, which takes into account the multi-dimensional coupling characteristics, randomness and volatility of the system, and establishes a more accurate parameter mapping relationship by introducing dynamic nonlinear parameters, making the constructed dynamic model more realistic.

[0019] Step S102, according to different bearing fault types, local dynamic modeling is performed on the bearing, the bearing dynamic equation is solved, and joint data of the faulty bearing is obtained.

[0020] In this step, local modeling is performed according to different fault types of the bearing (outer ring fault, inner ring fault). When the fault is introduced, the contact deformation at the rolling element for: , In the formula, is the offset angle, is the offset distance, It is the distance that the rolling bearing enters the fault zone during the entire operating cycle after the fault is introduced.

[0021] Outer race fault: , In the formula, is the angular position of the outer race fault, is the central angle corresponding to the outer ring fault, let is the outer ring diameter. Since the fault size is very small compared to the outer raceway diameter, It can be approximately written as: , is the outer raceway diameter.

[0022] Inner race fault: , In the formula, is the angular position of the inner race fault, is the central angle corresponding to the outer ring fault, let is the inner ring diameter. Since the fault size is very small compared to the inner raceway diameter, It can be approximately written as: .

[0023] Specifically, select appropriate bearings and measure the geometric parameters and fault frequencies of the bearings as shown in Tables 1 and 2, set appropriate rolling bearing simulation model parameters, and obtain the number of rolling elements N and the diameter of the rolling elements. , pitch diameter , frequency conversion The constructed dynamic model is combined with the bearing fault data to generate a large number of data sets.

[0024] Table 1 Failure frequency of rolling bearings , Table 2 Geometric parameters of rolling bearings , Use Simulink or Matlab to build a dynamic model, use the variable step size simulation algorithm ode45 to solve the dynamic equations, obtain the vibration response of the rolling bearing, and pre-process the joint data generated by the dynamic model (standardization operation, data enhancement) and input it into the fractal attention reconstruction network for training.

[0025] Step S103, input the preprocessed joint data into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model.

[0026] In this step, the bearing fault signal is input into the input layer of the fractal attention reconstruction network, and after being standardized, it is output to the convolution layer. The fractal attention reconstruction network uses a fractal convolution kernel, adaptively calculates the fractal dimension according to the complexity of the input bearing fault signal, dynamically adjusts the convolution kernel size and weight, and uses a non-integer order fractal convolution kernel to perform multi-scale feature mapping on the signal; Among them, the calculation expression of the adaptive fractal dimension in the fractal attention reconstruction network is: , In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: , In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: , In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: , In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, Variable parameters associated with the feature The expression for calculating the complexity of the input bearing fault signal is: , In the formula, is the number of network nodes, is the entropy of the graph, The complexity of the input bearing fault signal; The feature map is input into the pooling layer, and the pooled feature map is input into the fractal attention reconstruction module, which calculates the self-similarity and complexity of the features based on fractal geometry theory, where the self-similarity formula is: , The value at point x can be applied again by scaling x and translating it The function is used to calculate, by scaling x, calculating the value of F respectively, and taking the average of the two.

[0027] The expression for calculating the complexity of the feature is: , In the formula, is the complexity of the feature, is the feature distribution probability, is the feature category; The local features are integrated according to the fully connected layer to obtain the global features, which are applied to the fractal attention reconstruction network for multiple trainings. The training parameters are retained to obtain the high-speed rail bearing fault diagnosis model.

[0028] Step S104, acquiring real-time data of the bearing, and inputting the real-time data into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0029] In summary, the method of the present application, when bearing sample data is input, detects the data sample through the fractal attention reconstruction network to determine the type and location of the bearing fault. Compared with the traditional detection method, this method builds a rolling bearing dynamics model based on the bearing's geometric parameters and fault frequency, and generates joint data by combining the dynamics model with the experimental data. The joint data has the same characteristics as the actual experimental data in frequency domain distribution and time domain distribution, which can avoid the small number of data samples affecting the network detection accuracy. By using the fractal attention reconstruction network for fault diagnosis, it is possible to accurately capture microscopic fault features and extract richer signal details. By using a dynamic topological reconstruction method, the diagnostic performance and anti-interference ability of the model under different working conditions are improved.

[0030] See also Figure 2 , which shows a structural block diagram of a high-speed rail bearing fault diagnosis system of the present application.

[0031] like Figure 2 As shown, the high-speed rail bearing fault diagnosis system 200 includes a construction module 210, an acquisition module 220, a training module 230 and an output module 240.

[0032] Wherein, the construction module 210 is configured to construct a dynamic model of the bearing according to the geometric parameters and fault frequency of the bearing; An acquisition module 220 is configured to perform local dynamic modeling on the bearing according to different fault types of the bearing, solve the bearing dynamic equation, and obtain joint data of the faulty bearing; The training module 230 is configured to input the preprocessed joint data into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model; The output module 240 is configured to obtain real-time data of the bearing and input the real-time data into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0033] It should be understood that Figure 2 Modules and references documented in Figure 1 Therefore, the operations and features described above for the method and the corresponding technical effects are also applicable to Figure 2 The modules in it will not be described in detail here.

[0034] In other embodiments, the embodiments of the present invention further provide a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor executes the high-speed rail bearing fault diagnosis method in any of the above method embodiments; As an implementation mode, the computer-readable storage medium of the present invention stores computer-executable instructions, and the computer-executable instructions are configured as follows: Construct the dynamic model of the bearing according to the geometric parameters and fault frequency of the bearing; According to different bearing fault types, local dynamic modeling of the bearing is carried out, the bearing dynamic equation is solved, and the joint data of the faulty bearing is obtained; The preprocessed joint data is input into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: , In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: , In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: , In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: , In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; Real-time data of the bearing is obtained, and the real-time data is input into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0035] The computer-readable storage medium may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required by at least one function; the data storage area may store data created according to the use of the high-speed rail bearing fault diagnosis system, etc. In addition, the computer-readable storage medium may include a high-speed random access memory, and may also include a memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the high-speed rail bearing fault diagnosis system via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0036] Figure 3 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention, such as Figure 3 As shown, the device includes: a processor 310 and a memory 320. The electronic device may also include: an input device 330 and an output device 340. The processor 310, the memory 320, the input device 330 and the output device 340 may be connected via a bus or other means. Figure 3 The example of the connection via bus is taken. The memory 320 is the above-mentioned computer-readable storage medium. The processor 310 executes various functional applications and data processing of the server by running the non-volatile software programs, instructions and modules stored in the memory 320, that is, the high-speed rail bearing fault diagnosis method of the above-mentioned method embodiment is realized. The input device 330 can receive input digital or character information, and generate key signal input related to user settings and function control of the high-speed rail bearing fault diagnosis system. The output device 340 may include display devices such as display screens.

[0037] The electronic device can execute the method provided by the embodiment of the present invention, and has the functional modules and beneficial effects corresponding to the execution method. For technical details not described in detail in this embodiment, please refer to the method provided by the embodiment of the present invention.

[0038] As an implementation mode, the electronic device is applied to a high-speed rail bearing fault diagnosis system and is used for a client, and includes: at least one processor; and a memory connected to the at least one processor in communication; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: Construct the dynamic model of the bearing according to the geometric parameters and fault frequency of the bearing; According to different bearing fault types, local dynamic modeling is performed on the bearing, the bearing dynamic equation is solved, and the joint data of the faulty bearing is obtained; The preprocessed joint data is input into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: , In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: , In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: , In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: , In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; Real-time data of the bearing is obtained, and the real-time data is input into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

[0039] Through the description of the above implementation modes, those skilled in the art can clearly understand that each implementation mode can be implemented by means of software plus a necessary general hardware platform, or of course by hardware. Based on such an understanding, the above technical solution can essentially or in other words be embodied in the form of a software product that contributes to the prior art. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods of each embodiment or some parts of the embodiment.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A high-speed rail bearing fault diagnosis method, characterized in that: include: Construct the dynamic model of the bearing according to the geometric parameters and fault frequency of the bearing; According to different bearing fault types, local dynamic modeling is performed on the bearing, the bearing dynamic equation is solved, and the joint data of the faulty bearing is obtained; The preprocessed joint data is input into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: ， In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: ， In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: ， In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: ， In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; Real-time data of the bearing is obtained, and the real-time data is input into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

2. A high-speed rail bearing fault diagnosis method according to claim 1, characterized in that: The expression of the inner ring-shaft system in the dynamic model is: In the x-axis direction: ， In the formula, is the sum of the masses of the inner ring and the shaft, is the relative displacement of the inner ring in the x direction, is the connection damping between the shafts and the shaft fixings, is the equivalent connection stiffness of the shaft and the shaft fixing point, is the component of the elastic restoring force in the x-axis direction generated by the contact between the inner and outer rings and the rolling elements at time t; In the y-axis direction: ， In the formula, is the relative displacement of the inner circle in the y direction, is the acceleration due to gravity, is the wheel-rail excitation coefficient, is the eccentricity, is the angular frequency of the inner ring rotation, For time; The expression of the outer ring-bearing base system in the dynamic model is: In the x-axis direction: ， In the formula, is the sum of the masses of the outer ring and the bearing base, is the relative displacement of the outer ring in the x direction, is the connection damping between the bearing base and the ground foundation. is the equivalent connection stiffness between the bearing base and the ground; In the y-axis direction: ， In the formula, is the component of the elastic restoring force in the y-axis direction generated by the contact between the inner and outer rings and the rolling elements at time t, is the relative displacement of the outer ring in the y direction.

3. A high-speed rail bearing fault diagnosis method according to claim 2, characterized in that: in, The contact between the rolling element and the raceway in the rolling bearing is regarded as the linear elastic body restoring force, and the expression for calculating the component of the elastic restoring force in the x-axis direction generated by the contact between the inner and outer rings and the rolling element at the time t is: ， In the formula, is the load-deformation coefficient, is the load-deformation index, is the position at time t, is the initial position, is the elastic coefficient at time t; The expression for calculating the component of the elastic restoring force in the y-axis direction generated by the contact between the inner and outer rings and the rolling element at the time t is: ， In the formula, is the displacement after contact, is the initial displacement.

4. A high-speed rail bearing fault diagnosis method according to claim 1, characterized in that: The pre-processed joint data is input into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, including: The bearing fault signal is input into the input layer of the fractal attention reconstruction network, and after being standardized, it is output to the convolution layer. The fractal attention reconstruction network uses a fractal convolution kernel, adaptively calculates the fractal dimension according to the complexity of the input bearing fault signal, dynamically adjusts the convolution kernel size and weight, and uses a non-integer order fractal convolution kernel to perform multi-scale feature mapping on the signal; The feature map is input into the pooling layer, and the pooled feature map is input into the fractal attention reconstruction module. The fractal attention reconstruction module calculates the self-similarity and complexity of the feature based on the fractal geometry theory. The expression for calculating the complexity of the feature is: ， In the formula, is the complexity of the feature, is the feature distribution probability, is the feature category; According to the fully connected layer, local features are integrated to obtain global features, which are then applied to the fractal attention reconstruction network for multiple trainings. The training parameters are retained to obtain a bearing fault diagnosis model. The fractal attention reconstruction network is applied to multiple trainings. The training parameters are retained to obtain a high-speed rail bearing fault diagnosis model.

5. A high-speed rail bearing fault diagnosis method according to claim 4, characterized in that: in, The expression for calculating the complexity of the input bearing fault signal is: ， In the formula, is the number of network nodes, is the entropy of the graph, is the complexity of the input bearing fault signal.

6. A high-speed rail bearing fault diagnosis system, characterized in that: include: A construction module configured to construct a dynamic model of the bearing according to geometric parameters and fault frequencies of the bearing; an acquisition module, configured to perform local dynamic modeling on the bearing according to different fault types of the bearing, solve the bearing dynamic equation, and obtain joint data of the faulty bearing; The training module is configured to input the preprocessed joint data into a preset fractal attention reconstruction network for training to obtain a high-speed rail bearing fault diagnosis model, wherein the adaptive fractal dimension calculation expression in the fractal attention reconstruction network is: ， In the formula, is the adaptive fractal dimension, is the basic fractal dimension, a preset constant, is the learning rate, is the number of input data points, is the value of the i-th point, is the scale of the ith point, is the value of the i-1th point, is the scale of the i-1th point, is an estimate of the local fractal dimension between adjacent points; The expression of the non-integer order fractal convolution kernel in the fractal attention reconstruction network is: ， In the formula, is the gamma function, is the convolution kernel weight, is the order, For time, is the input signal, is a non-integer order parameter; The expression for adaptive convolution kernel weight adjustment in the fractal attention reconstruction network is: ， In the formula, is the softmax activation function, is the sample fractal dimension, is the weight of the convolution kernel, is the learning rate, is the loss gradient; The expression of the multi-scale feature map in the fractal attention reconstruction network is: ， In the formula, is the m-th layer multi-scale feature map, is the scale quantity, is the number of convolution kernels for each scale, is the weight corresponding to the scale and convolution kernel, is a non-integer order fractal convolution kernel, is a non-integer order parameter, are variable parameters related to the feature; The output module is configured to obtain real-time data of the bearing and input the real-time data into the high-speed rail bearing fault diagnosis model, and the high-speed rail bearing fault diagnosis model outputs the bearing fault type.

7. An electronic device, characterized in that: include: 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, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images