A method and system for identifying crack damage in railway ballastless track

Abstract

The invention discloses a method and system for identifying crack damage in a ballastless track of a railway. The method comprises: constructing a spatial vibration analysis model of a railway vehicle-track coupling system, proposing a crack trajectory approximation method, introducing the actual generation position of cracks in the ballastless track into the model, and considering the time dimension, setting a crack dynamic diffusion model including crack attenuation and frequency response, and finally constructing a crack damage constitutive matrix to accurately simulate and predict the behavior of the crack; obtaining vibration signals of a track plate when a train passes by before and after the crack is generated, using a multi-level fluctuation feature analysis method to establish a crack damage feature index to extract damage features; inputting the extracted features into an improved neural network model for training, and outputting the position and damage degree of cracks in the ballastless track of the railway, providing a solution to the problems of randomness of cracks and difficulty in predicting cracking paths, and improving the accuracy of crack identification in the ballastless track of the railway.

Description

Technical Field

[0001] The invention belongs to the technical field of railway ballastless track detection, and in particular relates to a railway ballastless track crack damage identification method and system. Background Art

[0002] With the continuous operation of high-speed railways, ballastless tracks are also facing challenges of various damages. These damages are mainly caused by factors such as train dynamic loads, thermal expansion and geological settlement. Cracks and gaps in the track slabs will not only reduce the smoothness, rigidity and comfort of the track, but in severe cases, they may even cause the track slabs to arch and rupture, seriously threatening the operation safety of high-speed trains. Traditional manual inspection methods rely on human observation and empirical judgment, which is not only inefficient, but also the accuracy and reliability of the results are limited by the skills and experience of the inspectors, and are prone to omissions and misjudgments. Image-based inspection technology, although it has made rapid progress, still has some limitations, such as sensitivity to ambient light, complex image processing algorithms and high data volume requirements, which may lead to slow processing speed and high consumption of computing resources in practical applications.

[0003] By accurately identifying crack damage, the railway department can quickly take targeted maintenance measures to prevent cracks from further expanding or causing other safety hazards. Optimizing the crack damage identification method can also extend the service life of ballastless track and further reduce the operating costs of the railway department. Therefore, accurate and rapid identification of crack damage on ballastless track is of great significance to ensure the safe operation of high-speed rail. Summary of the invention

[0004] The present invention provides a method and system for identifying crack damage in a ballastless railway track, provides a solution to the problems of randomness of cracks and difficulty in predicting cracking paths, improves the accuracy of identifying cracks in the ballastless railway track and the precision of identifying the degree of damage, and is conducive to the investigation of cracks in the ballastless railway track.

[0005] In a first aspect, the present invention provides a method for identifying crack damage in a ballastless railway track, comprising:

[0006] Constructing a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of ballastless track cracks into the spatial vibration analysis model according to a preset crack trajectory approximation rule;

[0007] Considering the time dimension, a crack dynamic diffusion model including crack attenuation and frequency response is set up, and a crack damage constitutive matrix is ​​constructed to simulate and predict crack behavior;

[0008] According to the crack damage constitutive matrix, the vibration signal of the track plate when the train passes by before and after the crack is generated is obtained, a crack damage characteristic index is established by using a preset multi-level fluctuation characteristic analysis strategy, and crack damage characteristics are extracted;

[0009] The extracted crack damage features are input into an improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

[0010] In a second aspect, the present invention provides a railway ballastless track crack damage identification system, comprising:

[0011] A construction module is configured to construct a spatial vibration analysis model of a railway vehicle-track coupling system, and introduce the actual generation position of the ballastless track crack into the spatial vibration analysis model according to a preset crack trajectory approximation rule;

[0012] a setup module configured to consider the time dimension, set up a crack dynamic propagation model including crack attenuation and frequency response, and construct a crack damage constitutive matrix for simulating and predicting crack behavior;

[0013] An extraction module is configured to obtain vibration signals of the track plate when a train passes by before and after the crack is generated, establish crack damage characteristic indicators using a preset multi-level fluctuation characteristic analysis strategy, and perform crack damage characteristic extraction;

[0014] The output module is configured to input the extracted crack damage characteristics into the improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

[0015] In 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 perform the steps of the method for identifying crack damage of a railway ballastless track of any embodiment of the present invention.

[0016] 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 method for identifying crack damage in a railway ballastless track of any embodiment of the present invention.

[0017] The railway ballastless track crack damage identification method and system of the present application construct a spatial vibration analysis model of a railway vehicle-track coupling system, propose a crack trajectory approximation method, introduce the actual generation position of the ballastless track crack into the model, and consider the time dimension, set up a crack dynamic diffusion model including crack attenuation and frequency response, and finally construct a crack damage constitutive matrix to accurately simulate and predict the behavior of the crack; obtain the vibration signal of the track plate when the train passes before and after the crack is generated, and use a multi-level fluctuation feature analysis method to establish a crack damage feature index (MFDI) for damage feature extraction; input the extracted features into an improved neural network model for training, and output the location and damage degree of the railway ballastless track crack, which provides a solution to the randomness of cracks and the difficulty in predicting the cracking path, improves the accuracy of railway ballastless track crack identification, and provides a certain theoretical reference for the railway ballastless track detection field. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] 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.

[0019] Figure 1 A flow chart of a railway ballastless track crack damage identification method provided by an embodiment of the present invention;

[0020] Figure 2 A structural block diagram of a railway ballastless track crack damage identification system provided by an embodiment of the present invention;

[0021] Figure 3 It is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0022] 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.

[0023] See also Figure 1 , which shows a flow chart of a railway ballastless track crack damage identification method of the present application.

[0024] like Figure 1 As shown, the railway ballastless track crack damage identification method specifically includes the following steps:

[0025] Step S101, constructing a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of the ballastless track crack into the spatial vibration analysis model according to a preset crack trajectory approximation rule.

[0026] In this step, the establishment of a spatial vibration analysis model is a prerequisite. It is necessary to simulate the real crack damage in the spatial vibration analysis model to extract a more realistic track plate vibration signal, and to set a crack position closer to the real one according to the preset crack trajectory approximation rule. Compared with the random setting of the existing crack position, the actual position of the ballastless track crack is introduced into the spatial vibration analysis model in a more accurate and realistic description of the occurrence of cracks. Specifically, an approximate parameter is established. Approximate crack trajectory , close to the real crack trajectory , the objective function is:

[0027] ,

[0028] In the formula, is the approximate objective function, is the approximate parameter, t is the time variable, is the dynamic learning rate, is the dynamic discount factor, is the time attenuation coefficient, is the amplitude adjustment parameter, is the amplitude, is the angular frequency, is the phase, is the maximum value of the next state, is the coordinate of the crack trajectory at any time, is the coordinate of the crack trajectory at the next moment;

[0029] Establishing 3D crack approximate trajectory matrix Close to the real crack trajectory matrix ,Right now:

[0030] ,

[0031] In the formula, is the approximate coordinate of the first crack trajectory at the initial moment, is the approximate coordinate of the second crack trajectory at the initial moment, is the approximate coordinate of the Nth crack trajectory at the initial moment, is the approximate coordinate of the first crack trajectory at time t, is the approximate coordinate of the second crack trajectory at time t, is the approximate coordinate of the Nth crack trajectory at the tth time, is the coordinate of the first real crack trajectory at the initial moment, is the coordinate of the second real crack trajectory at the initial moment, is the coordinate of the Nth real crack trajectory at the initial moment, is the coordinate of the first real crack trajectory at time t, is the coordinate of the second real crack trajectory at time t, is the coordinate of the Nth real crack trajectory at the tth moment.

[0032] Step S102 , considering the time dimension, setting a crack dynamic diffusion model including crack attenuation and frequency response, and constructing a crack damage constitutive matrix for simulating and predicting crack behavior.

[0033] In this step, the crack propagation equation of the crack dynamic diffusion model is:

[0034] ,

[0035] In the formula, is the crack diffusion function, x, y, z are the spatial coordinates, t is the time variable, is the frequency variable, is a small quantity approaching zero in the limiting process, is the initial integration variable, is the integration variable, is the time attenuation coefficient, is the attenuation coefficient of frequency;

[0036] The width constraint equation for the crack width in the crack dynamic diffusion model is:

[0037] ,

[0038] In the formula, , , are the crack widths in the x, y, and z directions, respectively. is the elastic cracking strain, is the inelastic cracking strain, , , are the normal angles of the crack in the x, y, and z directions, respectively. , , are the tangential angles of the cracks in the x, y, and z directions, respectively. , , are the stress components in the x, y, and z directions, respectively. is the scale factor.

[0039] It should be noted that the crack damage constitutive matrix can not only capture factors such as crack density, direction, and length, but also comprehensively consider the time variable compared to the existing crack damage constitutive matrix, making the model more comprehensive and accurate in predicting and analyzing the behavior of materials in complex application environments. The expression of the crack damage constitutive matrix is:

[0040] ,

[0041] In the formula, is the crack damage constitutive matrix, E is the elastic modulus, is Poisson's ratio, Determine cracking parameters for stress or strain states, is the crack density, is the fracture anisotropy parameter, is the crack length, is the crack direction, is the stress intensity factor, is the material toughness, R is the crack extension resistance, is the interaction force between cracks, t is the time variable, is the time impact adjustment index, is the time attenuation coefficient.

[0042] Step S103, obtaining the vibration signal of the track plate when a train passes by before and after the crack is generated according to the crack damage constitutive matrix, establishing crack damage characteristic indicators using a preset multi-level fluctuation characteristic analysis strategy, and performing crack damage feature extraction.

[0043] In this step, the acquired track plate vibration signal S is decomposed by waveform transformation, and the i-th decomposition layer is obtained. sub-bands, the track plate vibration signal S is expressed as:

[0044] ,

[0045] In the formula, represents the dynamic response of the vibration signal of the i-th layer decomposition node j, Decompose the nodes at the i-th level -1 dynamic response of the vibration signal;

[0046] Based on the principle of multi-scale analysis, the time domain energy of the track plate vibration signal in each frequency band is calculated, and the waveform transformation time-frequency characteristic energy vector is constructed, namely:

[0047] ,

[0048] In the formula, The vibration signal on the i-th layer decomposition node j is centered at the sampling time point and the time interval is The energy of the internal signal, N is the number of accelerations collected in an observation window corresponding to a time point, is the width of the observation window;

[0049] The time-frequency characteristic energy vector of the waveform transformation of the track plate vibration signal at the i-th decomposition layer for:

[0050] ,

[0051] In the formula, Decompose the nodes at the i-th level The vibration signal on -1 is centered at the sampling time point and the time interval is The energy of the internal signal;

[0052] The crack damage characteristic index is constructed, and the crack damage characteristics are extracted from the track plate vibration signal. The expression of the crack damage characteristic index is:

[0053] ,

[0054] In the formula, is the characteristic index of crack damage, and are the energy of the Pth characteristic frequency band of the structure in healthy and damaged states, and are the average values ​​of all frequency band energies of the structure in healthy and damaged states, respectively, and k is the number of damaged frequency bands.

[0055] Step S104, inputting the extracted crack damage features into the improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

[0056] In this step, the neural network feature vector is constructed, and the expression is:

[0057] ,

[0058] In the formula, T is the feature vector set, is the first crack damage characteristic index value, is the second crack damage characteristic index value, is the characteristic index value of the nth crack damage;

[0059] Complex numbers and fractional transformations are used in each layer of the neural network, and multidimensional tensor operations and higher-order mathematical structures are introduced. The difference is that compared with many traditional methods, high-order multidimensional fractional transformation networks can process multidimensional data and perform complex transformations in multiple dimensions. Fractional functions and complex coefficients are used for data transformation, which provides more complex and flexible model expression capabilities than traditional linear or polynomial methods, and more flexible adjustment and optimization of its parameters to adapt to different data characteristics and task requirements. This adaptive ability exceeds the capabilities of many fixed transformations or static networks. Effect: High-order multidimensional fractional transformation networks can more accurately capture the intrinsic structure and pattern of data, especially when the data has complex dynamic characteristics. This enhanced representation ability can improve the accuracy and efficiency of data analysis and processing tasks, and is suitable for processing nonlinear relationships and highly complex data environments. In practical applications such as image recognition, speech processing, or complex system modeling, they can provide better performance and higher robustness. Due to its multidimensional and high-order characteristics, this network can adapt to a wide range of application scenarios to improve operating efficiency and greatly increase the expressive power of the model to capture more complex data patterns and dependencies. The expression of the network structure of the neural network is:

[0060] ,

[0061] In the formula, is a multidimensional complex input, is the global constant coefficient, is the weight-adjusted complex coefficient, Adjust the complex coefficients for the levels, is the integer power of the upper level, is the integer power of the upper level, is the level cyclic coefficient, J is the index of a specific polynomial term, and m is the dimension of the index;

[0062] The parameterized multidimensional linear transformation expression is:

[0063] ,

[0064] In the formula, To indicate the The parameterized linear transformation function between layers, is the output data passed down from the previous layer, is the tensor parameter of the previous layer of the neural network structure layer, It is the tensor parameter of the next level of the neural network structure layer;

[0065] Propose a composite loss function containing multiple regularization factors , Difference: The composite loss function introduces multiple regularization terms to increase the control over the model weights and structure, which is more complex and comprehensive than the common single regularization. Through hierarchical (each layer has its own unique regularization term) and global regularization terms, the composite loss function fine-tunes the parameters of each layer and maintains the consistency of global parameter optimization, which is not common in traditional loss functions. The regularization term is specially designed to ensure the legitimacy of the fractional transformation and prevent mathematically impermissible operations caused by improper parameter settings;

[0066] Effect: Through multiple regularization terms, the composite loss function effectively limits the complexity of the model and reduces the risk of overfitting. The multiple regularization strategy ensures the stability of the model during training and reduces the numerical calculation problems that may occur during training, such as gradient explosion or disappearance. Through adjustment, model developers can flexibly adjust the influence of different regularization terms according to specific application requirements and data characteristics to achieve more detailed parameter optimization. In addition, the design of this composite loss function also helps the performance of the model on novel or unseen data. Through appropriate regularization strategies, the adaptability and robustness of the model can be enhanced, so that it can better cope with data changes and challenges that may be encountered in practical applications.

[0067] The expression is:

[0068] ,

[0069] In the formula, is the main loss function, n is the total number of layers in the neural network, , are the regularization factors, is the first complex coefficient, is the post-complex coefficient, is the output data passed down from the previous layer, is an integer power, is the number of polynomial terms in the kth level, is the highest power in each term, The data converted for the structure layer;

[0070] Among them, the main loss function The expression is:

[0071] ,

[0072] In the formula, G is the number of samples, C is the number of categories, is the dynamic weight factor, is an indication of whether the i-th sample belongs to category c, The model predicts the probability that the i-th sample belongs to category c, is the gap between the predicted probability and the actual label, is the position of the actual label of the kth layer;

[0073] Optimizing the neural network parameter weights specifically involves introducing the framework of multidimensional tensor field theory to deal with high-dimensional parameter optimization problems, where each parameter is regarded as a field point in a high-dimensional space, and their interactions are described by a complex multi-body field equation. The field dynamics equation can be decomposed into multiple small, relatively independent sub-problems for parallel processing, expressed as:

[0074] ,

[0075] In the formula, To define the field, is the potential energy function, For the field The gradient in parameter space, is the field interaction coefficient, is the factorial of k, is the coefficient of the interaction of higher-order derivatives, m is the order of the derivative term, M is the maximum order of the derivative term, For the field of Derivatives, To define the i-th value in the field, To define the jth value in the field, is the jth value in the definition field;

[0076] The dynamic equation of the field is:

[0077] ,

[0078] In the formula, is potential energy, For the field The functional derivative of .

[0079] In summary, the method of the present application constructs a spatial vibration analysis model of a railway vehicle-track coupling system, proposes a crack trajectory approximation method, introduces the actual location of cracks in the ballastless track into the model, considers the time dimension, sets a crack dynamic diffusion model including crack attenuation and frequency response, and finally constructs a crack damage constitutive matrix to accurately simulate and predict the behavior of cracks; obtains the vibration signal of the track plate when the train passes before and after the crack is generated, and uses a multi-level fluctuation feature analysis method to establish a crack damage feature index (MFDI) for damage feature extraction; the extracted features are input into an improved neural network model for training, and the location and damage degree of cracks in the ballastless track are output, which provides a solution to the randomness of cracks and the difficulty in predicting cracking paths, improves the accuracy of crack identification in the ballastless track, and provides a certain theoretical reference for the field of ballastless track detection.

[0080] See also Figure 2 , which shows a structural block diagram of a railway ballastless track crack damage identification system of the present application.

[0081] like Figure 2 As shown, the railway ballastless track crack damage identification system 200 includes a construction module 210, a setting module 220, an extraction module 230 and an output module 240.

[0082] Among them, the construction module 210 is configured to construct a spatial vibration analysis model of the railway vehicle-track coupling system, and introduce the actual generation position of the ballastless track crack into the spatial vibration analysis model according to the preset crack trajectory approximation rule; the setting module 220 is configured to consider the time dimension, set a crack dynamic diffusion model including crack attenuation and frequency response, and construct a crack damage constitutive matrix for simulating and predicting crack behavior; the extraction module 230 is configured to obtain the vibration signal of the track plate when the train passes before and after the crack is generated according to the crack damage constitutive matrix, use a preset multi-level fluctuation feature analysis strategy to establish a crack damage feature index, and extract the crack damage feature; the output module 240 is configured to input the extracted crack damage feature into the improved neural network model, and the improved neural network model outputs the location and damage degree of the railway ballastless track crack.

[0083] 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.

[0084] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor is caused to execute the railway ballastless track crack damage identification method in any of the above method embodiments;

[0085] 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:

[0086] Constructing a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of ballastless track cracks into the spatial vibration analysis model according to a preset crack trajectory approximation rule;

[0087] Considering the time dimension, a crack dynamic diffusion model including crack attenuation and frequency response is set up, and a crack damage constitutive matrix is ​​constructed to simulate and predict crack behavior;

[0088] Obtain the vibration signal of the track plate when the train passes by before and after the crack is generated, use the preset multi-level fluctuation characteristic analysis strategy to establish the crack damage characteristic index, and extract the crack damage characteristics;

[0089] The extracted crack damage features are input into an improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

[0090] 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 and an application required by at least one function; the data storage area may store data created according to the use of the railway ballastless track crack damage identification 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 railway ballastless track crack damage identification 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.

[0091] 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 electronic 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 In the example, the connection through the bus is taken as an example. 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 railway ballastless track crack damage identification method of the above-mentioned method embodiment is implemented. 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 railway ballastless track crack damage identification system. The output device 340 may include display devices such as display screens.

[0092] 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.

[0093] As an implementation mode, the electronic device is applied to a railway ballastless track crack damage identification 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:

[0094] Constructing a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of ballastless track cracks into the spatial vibration analysis model according to a preset crack trajectory approximation rule;

[0095] Considering the time dimension, a crack dynamic diffusion model including crack attenuation and frequency response is set up, and a crack damage constitutive matrix is ​​constructed to simulate and predict crack behavior;

[0096] According to the crack damage constitutive matrix, the vibration signal of the track plate when the train passes by before and after the crack is generated is obtained, a crack damage characteristic index is established by using a preset multi-level fluctuation characteristic analysis strategy, and crack damage characteristics are extracted;

[0097] The extracted crack damage features are input into an improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

[0098] 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.

[0099] 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 method for identifying crack damage in ballastless railway track, characterized in that: include: Constructing a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of cracks in the ballastless track into the spatial vibration analysis model according to a preset crack trajectory approximation rule, the construction of a spatial vibration analysis model of a railway vehicle-track coupling system, and introducing the actual generation position of cracks in the ballastless track into the spatial vibration analysis model according to a preset crack trajectory approximation rule comprises: Establish an approximate parameter Approximate crack trajectory , close to the real crack trajectory , the objective function is: ， In the formula, is the approximate objective function, is the approximate parameter, t is the time variable, is the dynamic learning rate, is the dynamic discount factor, is the time attenuation coefficient, is the amplitude adjustment parameter, is the amplitude, is the angular frequency, is the phase, is the maximum value of the next state, is the coordinate of the crack trajectory at any time, is the coordinate of the crack trajectory at the next moment; Establishing 3D crack approximate trajectory matrix Close to the real crack trajectory matrix ,Right now: ， In the formula, is the approximate coordinate of the first crack trajectory at the initial moment, is the approximate coordinate of the second crack trajectory at the initial moment, is the approximate coordinate of the Nth crack trajectory at the initial moment, is the approximate coordinate of the first crack trajectory at time t, is the approximate coordinate of the second crack trajectory at time t, is the approximate coordinate of the Nth crack trajectory at the tth time, is the coordinate of the first real crack trajectory at the initial moment, is the coordinate of the second real crack trajectory at the initial moment, is the coordinate of the Nth real crack trajectory at the initial moment, is the coordinate of the first real crack trajectory at time t, is the coordinate of the second real crack trajectory at time t, is the coordinate of the Nth real crack trajectory at the tth time; Considering the time dimension, a crack dynamic diffusion model including crack attenuation and frequency response is set, and a crack damage constitutive matrix for simulating and predicting crack behavior is constructed. The crack propagation equation of the crack dynamic diffusion model is: ， In the formula, is the crack diffusion function, x, y, z are the spatial coordinates, t is the time variable, is the frequency variable, is a small quantity approaching zero in the limiting process, is the initial integration variable, is the integration variable, is the time attenuation coefficient, is the attenuation coefficient of frequency; The width constraint equation for the crack width in the crack dynamic diffusion model is: ， In the formula, , , are the crack widths in the x, y, and z directions, respectively. is the elastic cracking strain, is the inelastic cracking strain, , , are the normal angles of the crack in the x, y, and z directions, respectively. , , are the tangential angles of the cracks in the x, y, and z directions, respectively. , , are the stress components in the x, y, and z directions, respectively. is the scale factor; The expression of the crack damage constitutive matrix is: ， In the formula, is the crack damage constitutive matrix, E is the elastic modulus, is Poisson's ratio, Determine cracking parameters for stress or strain states, is the fracture density, is the fracture anisotropy parameter, is the crack length, is the crack direction, is the stress intensity factor, is the material toughness, R is the crack extension resistance, is the interaction force between cracks, t is the time variable, is the time impact adjustment index, is the attenuation coefficient of time; According to the crack damage constitutive matrix, the vibration signal of the track plate when the train passes by before and after the crack is generated is obtained, a crack damage characteristic index is established by using a preset multi-level fluctuation characteristic analysis strategy, and crack damage characteristics are extracted; The extracted crack damage features are input into an improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

2. A railway ballastless track crack damage identification method according to claim 1, characterized in that: The method of obtaining the vibration signal of the track plate when a train passes by before and after the crack is generated, establishing the crack damage characteristic index by using a preset multi-level fluctuation characteristic analysis strategy, and extracting the crack damage characteristic comprises: The obtained track plate vibration signal S is decomposed by waveform transformation, and the i-th decomposition layer is obtained sub-bands, the track plate vibration signal S is expressed as: ， In the formula, represents the dynamic response of the vibration signal of the i-th layer decomposition node j, Decompose the nodes at the i-th level -1 dynamic response of the vibration signal; Based on the principle of multi-scale analysis, the time domain energy of the track plate vibration signal in each frequency band is calculated, and the waveform transformation time-frequency characteristic energy vector is constructed, namely: ， In the formula, The vibration signal on the i-th layer decomposition node j is centered at the sampling time point and the time interval is The energy of the internal signal, N is the number of accelerations collected in an observation window corresponding to a time point, is the width of the observation window; The time-frequency characteristic energy vector of the waveform transformation of the track plate vibration signal at the i-th decomposition layer for: ， In the formula, Decompose the nodes at the i-th level The vibration signal on -1 is centered at the sampling time point and the time interval is The energy of the internal signal; The crack damage characteristic index is constructed, and the crack damage characteristics are extracted from the track plate vibration signal. The expression of the crack damage characteristic index is: ， In the formula, is the characteristic index of crack damage, and are the energy of the Pth characteristic frequency band of the structure in healthy and damaged states, and are the average values ​​of all frequency band energies of the structure in healthy and damaged states, respectively, and k is the number of damaged frequency bands.

3. A railway ballastless track crack damage identification method according to claim 1, characterized in that: The extracted crack damage features are input into the improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway, including: Construct the neural network feature vector, the expression is: ， In the formula, T is the feature vector set, is the first crack damage characteristic index value, is the second crack damage characteristic index value, is the characteristic index value of the nth crack damage; Complex numbers and fractional transformations are used in each layer of the neural network, and multidimensional tensor operations and higher-order mathematical structures are introduced. The network structure of the neural network is expressed as: ， In the formula, is a multidimensional complex input, is the global constant coefficient, is the weight-adjusted complex coefficient, Adjust the complex coefficients for the levels, is the integer power of the upper level, is the integer power of the upper level, is the level cyclic coefficient, J is the index of a specific polynomial term, and m is the dimension of the index; The parameterized multidimensional linear transformation expression is: ， In the formula, To indicate the The parameterized linear transformation function between layers, is the output data passed down from the previous layer, is the tensor parameter of the previous layer of the neural network structure layer, It is the tensor parameter of the next level of the neural network structure layer; Determine the composite loss function containing multiple regularization factors , the expression is: ， In the formula, is the main loss function, n is the total number of layers in the neural network, , are the regularization factors, is the first complex coefficient, is the post-complex coefficient, is an integer power, is the number of polynomial terms in the kth level, is the highest power in each term, The data converted for the structure layer; Among them, the main loss function The expression is: ， In the formula, G is the number of samples, C is the number of categories, is the dynamic weight factor, is an indication of whether the i-th sample belongs to category c, The model predicts the probability that the i-th sample belongs to category c, is the gap between the predicted probability and the actual label, is the position of the actual label of the kth layer; Optimizing the neural network parameter weights specifically involves introducing the framework of multidimensional tensor field theory to deal with high-dimensional parameter optimization problems, where each parameter is regarded as a field point in a high-dimensional space, and their interactions are described by a multi-body field equation. The field dynamics equation is decomposed into multiple small, relatively independent sub-problems for parallel processing, expressed as: ， In the formula, To define the field, is the potential energy function, For the field The gradient in parameter space, is the field interaction coefficient, is the factorial of k, is the coefficient of the interaction of higher-order derivatives, m is the order of the derivative term, M is the maximum order of the derivative term, For the field of Derivatives, To define the i-th value in the field, To define the jth value in the field, is the jth value in the definition field; The dynamic equation of the field is: ， In the formula, is potential energy, For the field The functional derivative of .

4. A railway ballastless track crack damage identification system according to any one of the methods described in claims 1-3, characterized in that: include: A construction module is configured to construct a spatial vibration analysis model of a railway vehicle-track coupling system, and introduce the actual generation position of the ballastless track crack into the spatial vibration analysis model according to a preset crack trajectory approximation rule; a setup module configured to consider the time dimension, set up a crack dynamic propagation model including crack attenuation and frequency response, and construct a crack damage constitutive matrix for simulating and predicting crack behavior; An extraction module is configured to obtain a vibration signal of the track plate when a train passes by before and after the crack is generated according to the crack damage constitutive matrix, establish a crack damage characteristic index using a preset multi-level fluctuation characteristic analysis strategy, and perform crack damage characteristic extraction; The output module is configured to input the extracted crack damage characteristics into the improved neural network model, and the improved neural network model outputs the location and damage degree of the cracks in the ballastless track of the railway.

5. 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 execute the method described in any one of claims 1 to 3.

6. 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 3 is implemented.

Images