Track fastener looseness identification method based on ultra-weak fiber grating vibration optical cable

Abstract

The invention discloses a rail fastener looseness identification method based on an ultra-weak fiber grating vibration optical cable, and relates to the field of rail engineering. The method comprises the following steps: building a steel rail vibration signal acquisition system based on an ultra-weak fiber grating vibration optical cable to obtain signals of steel rail vibration under various working conditions; acquiring an optimal modal component of the steel rail vibration signal under each working condition; determining influence indexes of steel rail vibration, and calculating time domain characteristic values of the steel rail vibration signals under various working conditions; and constructing a track fastener looseness identification model based on the convolutional neural network and the bidirectional long-short-term memory network to obtain a track fastener looseness identification result. According to the invention, the limitation that a plurality of vibration optical cables cannot be simultaneously used for rail long-line distance detection is broken through, the flexibility of the identification method is improved, the arrangement of the device is simple, and a large number of vibration signals can be collected for rail fastener looseness identification, so that the reliability of rail fastener looseness identification is improved.

Description

Technical Field

[0001] The invention relates to the field of track engineering, and in particular to a method for identifying loose track fasteners based on ultra-weak fiber grating vibration optical cables. Background Art

[0002] Distributed acoustic sensing system (DAS) has the advantages of high precision, wide distribution and low cost. With the development of technology, it has gradually attracted the attention of various fields and has been widely used in signal recognition and monitoring. Since each grid point on the ultra-weak fiber Bragg grating vibration cable has the ability to sense, its spatial resolution can reach several times the advantage compared with traditional ordinary fiber sensors and electronic vibration sensors, and has broad application prospects in the field of complex environment monitoring. With the continuous development of fiber optic sensing technology, fiber optic monitoring technology can be mainly divided into three generations. The first generation is the traditional quasi-distributed monitoring technology based on fiber Bragg grating, and the second generation is the distributed monitoring technology based on Brillouin optical time domain reflection technology. Among them, the traditional Bragg grating displacement sensor can realize remote real-time monitoring, but can only realize quasi-distributed monitoring of limited points; the sensing cable based on Brillouin optical time domain reflection technology can realize long-distance distributed monitoring, but there are problems such as high cost, lower measurement accuracy than quasi-distributed, and difficulty in real-time monitoring. The latest third-generation fiber Bragg grating array sensing technology can realize large-capacity fiber Bragg grating sensor multiplexing due to its narrow bandwidth and low reflectivity, thereby realizing intensive and precise distributed monitoring.

[0003] Tracks are one of the important components of the lifeline of urban infrastructure. Track fasteners are an important part of the track structure, and their service status directly determines the operation and maintenance status of the train. Therefore, fastener fault diagnosis is an important part of railway facility maintenance. Track fastener failures are mainly manifested in fastener defects and looseness. Defects are manifested in missing parts, installation errors, etc., and looseness is manifested in fastener separation, fastener spring bar falling off, etc. In the past, the loosening of track fasteners was often detected by a single sensor fixed-point detection method, such as an acceleration vibration sensor. Although this disease detection method can meet the loosening identification of a single track fastener, it is difficult to achieve overall detection for long-line tunnels and scenarios where multiple fasteners are loose. Summary of the invention

[0004] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for identifying loose rail fasteners based on ultra-weak fiber grating vibration optical cables.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] A method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cables comprises the following steps:

[0007] S1. Build a rail vibration signal acquisition system based on ultra-weak fiber Bragg grating vibration optical cable to obtain rail vibration signals under various working conditions;

[0008] S2. Based on the whale optimization method and the variational modal decomposition method, the rail vibration signal under each working condition is decomposed, the optimal modal component of each working condition is extracted, and the optimal modal component of the rail vibration signal under each working condition is obtained;

[0009] S3, determining the influence index of rail vibration, and calculating the time domain characteristic value of the rail vibration signal under each working condition according to the optimal modal component of the rail vibration signal under each working condition and the influence index of the rail vibration;

[0010] S4. A track fastener loosening identification model is constructed based on a convolutional neural network and a bidirectional long short-term memory network. The track fastener loosening identification result is obtained according to the time domain eigenvalues ​​of the rail vibration signal under various working conditions and the track fastener loosening identification model.

[0011] Furthermore, in step S1, a rail vibration signal acquisition system is built based on the ultra-weak fiber Bragg grating vibration cable, specifically: determining the working section of the ultra-weak fiber Bragg grating vibration cable, serially welding the ultra-weak fiber Bragg grating vibration cable, arranging the ultra-weak fiber Bragg grating vibration cable by slotting and burying, connecting it to a vibration demodulator, and setting the excitation point position on the track plate of the rail to build the rail vibration signal acquisition system.

[0012] Further, in step S1, the working conditions include fasteners being fully tightened, single fastener failure on one side of the track, single fastener failure on both sides of the track, interval failure of fasteners on one side of the track, complete failure of fasteners on both sides of the track, external intrusion of foreign objects on the track, and pedestrians passing by the track.

[0013] Further, step S2 includes the following steps:

[0014] S21. Based on the whale optimization method, the envelope entropy is used as the fitness function to initialize the position vector of the whale group, calculate the fitness of each whale, and select the iterative formula for iterative update by judging the size of the convergence factor to determine the optimization parameters of the variational mode decomposition;

[0015] S22. Based on the variational modal decomposition method, the rail vibration signal is decomposed according to the optimization parameters of the variational modal decomposition, the optimal modal components of each working condition are extracted, and the optimal modal components of the rail vibration signal under each working condition are obtained.

[0016] Further, in step S3, the influencing indexes of the rail vibration include the maximum value, minimum value, average value, effective value, standard deviation, kurtosis, skewness, waveform index and pulse index of the rail vibration.

[0017] Further, in step S4, the track fastener loosening identification model includes an input layer, a sequence folding layer, a convolution layer, a pooling layer, a Flattern layer, a BiLSTM layer, a dropout layer and a fully connected layer connected in sequence; the input layer is used to convert the time domain characteristic values ​​of the rail vibration signal under various working conditions into one-dimensional sequence data of characteristic parameters, and the sequence folding layer is used to convert the one-dimensional sequence data of the characteristic parameters into tensor data of the characteristic parameters; the convolution layer is used to convolve the tensor data of the characteristic parameters to extract the features of the tensor data; the pooling layer is used to filter the features of the tensor data to obtain filtered feature data; the Flattern layer is used to flatten the filtered feature data to obtain filtered feature one-dimensional data; the BiLSTM layer is used to obtain the front and back temporal dependencies of the filtered feature one-dimensional data to extract the deep features of the filtered feature one-dimensional data; the dropout layer is used to set the probability in the forward propagation process to select the deep features of the filtered feature one-dimensional data; the fully connected layer maps the feature space to the sample label space based on the deep features of the selected filtered feature one-dimensional data to obtain the track fastener loosening identification result.

[0018] Furthermore, a BN layer is set between the convolution layer and the pooling layer. The BN layer is used to standardize, scale and offset the features of the tensor data to speed up the training process and prevent gradient disappearance or gradient explosion.

[0019] The present invention has the following beneficial effects:

[0020] (1) The present invention builds a rail vibration signal acquisition system based on ultra-weak fiber Bragg grating vibration optical cables to obtain rail vibration signals under various working conditions. For the first time, ultra-weak fiber Bragg grating vibration optical cables are used to identify the looseness of rail fasteners by serial welding, breaking through the limitation that multiple vibration optical cables cannot be used for long-distance rail detection at the same time, and improving the flexibility of the identification method;

[0021] (2) The present invention builds a rail vibration signal acquisition system based on ultra-weak fiber Bragg grating vibration optical cable. The device has simple layout and high reliability. It can collect a large number of vibration signals for loose rail fastener identification, thereby improving the reliability of loose rail fastener identification;

[0022] (3) The present invention builds a rail vibration signal acquisition system based on an ultra-weak fiber Bragg grating vibration cable to obtain the rail vibration signal under various working conditions, and then decomposes the rail vibration signal under various working conditions based on the whale optimization method and the variational mode decomposition method, extracts the optimal mode component of each working condition, obtains the optimal mode component of the rail vibration signal under each working condition, and then determines the influencing index of the rail vibration. According to the optimal mode component of the rail vibration signal under each working condition and the influencing index of the rail vibration, the time domain eigenvalue of the rail vibration signal under each working condition is calculated. Finally, based on the convolutional neural network and the bidirectional long short-term memory network, a track fastener loosening identification model is constructed. According to the time domain eigenvalue of the rail vibration signal under each working condition and the track fastener loosening identification model, the track fastener loosening identification result is obtained, which solves the problems of high detection cost, low detection efficiency and inability to perform long-distance distributed and accurate monitoring of track fasteners in the existing track fastener loosening identification method, thereby meeting the use requirements of modern railways. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic flow chart of a method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cables;

[0024] Figure 2 A schematic diagram of the locations of the excitation points provided for the present invention;

[0025] Figure 3 This is a time domain vibration waveform diagram collected by the present invention when the fastener is fully fastened;

[0026] Figure 4 This is a time domain vibration waveform diagram collected by the present invention when a single fastener on one side of the track fails;

[0027] Figure 5 This is a time domain vibration waveform diagram collected by the present invention when a single fastener on both sides of the track fails;

[0028] Figure 6 The time domain vibration waveform diagram collected by the present invention when the interval of the single-side fastener of the track fails;

[0029] Figure 7 This is a time domain vibration waveform diagram collected by the present invention when all the fasteners on both sides of the track fail;

[0030] Figure 8 The time domain vibration waveform diagram collected by the present invention when there is external invasion of foreign matter on the track;

[0031] Fig. 9 The time domain vibration waveform diagram collected by the present invention when a pedestrian passes by on the track;

[0032] Fig.10This is a time domain vibration waveform diagram of the environmental noise on the track collected by the present invention;

[0033] Fig.11 This is a frequency domain vibration waveform diagram collected by the present invention when the fastener is fully fastened;

[0034] Fig.12 This is a frequency domain vibration waveform diagram collected by the present invention when a single fastener on one side of the track fails;

[0035] Fig.13 This is a frequency domain vibration waveform diagram collected by the present invention when a single fastener on both sides of the track fails;

[0036] Fig.14 The frequency domain vibration waveform diagram collected by the present invention when the interval of the single-side fastener of the track fails;

[0037] Fig.15 This is a frequency domain vibration waveform diagram collected by the present invention when all the fasteners on both sides of the track fail;

[0038] Fig.16 This is a frequency domain vibration waveform diagram collected by the present invention when there is external invasion of foreign objects on the track;

[0039] Fig.17 The frequency domain vibration waveform diagram collected by the present invention when a pedestrian passes by on the track;

[0040] Fig.18 The frequency domain vibration waveform diagram of the environmental noise on the track collected by the present invention;

[0041] Fig.19 It is a schematic diagram of the structure of the track fastener loosening identification model of the present invention;

[0042] Fig. 20 This is a statistical diagram of the results of rail fastener identification using the test set in the present invention;

[0043] Fig.21 This is a schematic diagram of the test set recognition results after the confusion matrix is ​​introduced into the present invention. DETAILED DESCRIPTION

[0044] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0045] like Figure 1As shown, a method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable includes steps S1-S4, which are as follows:

[0046] S1. A rail vibration signal acquisition system is built based on ultra-weak fiber Bragg grating vibration cables to obtain rail vibration signals under various working conditions.

[0047] In an optional embodiment of the present invention, the present invention builds a rail vibration signal acquisition system based on an ultra-weak fiber grating vibration cable, specifically: determining the working section of the ultra-weak fiber grating vibration cable, serially welding the ultra-weak fiber grating vibration cable, arranging the ultra-weak fiber grating vibration cable in a slotted and buried manner, connecting it to a vibration demodulator, and setting an excitation point position on the track plate of the rail to build a rail vibration signal acquisition system.

[0048] The working conditions include fasteners being fully tightened, failure of a single fastener on one side of the track, failure of a single fastener on both sides of the track, failure of interval fasteners on one side of the track, failure of all fasteners on both sides of the track, external intrusion of foreign objects on the track, and pedestrians passing by on the track.

[0049] like Figure 2 As shown, the present invention sets multiple excitation point positions on the track plate, strikes the track plate by hammering, and collects a large number of rail vibration signals by a vibration demodulator. The present invention performs repeated load addition experiments for each working condition, and each group of experiments is repeated 200 times to obtain the rail vibration signals under each working condition, and the signals are specifically time domain signals.

[0050] like Figure 3-10 As shown, the present invention provides an evaluation diagram of the time domain relationship of ultra-weak fiber Bragg grating vibration optical cable. Figure 3 It is a time domain vibration waveform diagram collected by the present invention when the fastener is fully tightened, the horizontal axis Time is time, the vertical axis Amplitude is amplitude, and Fasteners fully tightened means the fastener is fully tightened; Figure 4 This is a time domain vibration waveform diagram collected by the present invention when a single fastener on one side of the track fails. Unilateral failure of fasteners refers to the failure of a single fastener on one side. Figure 5 The time domain vibration waveform diagram collected by the present invention when a single fastener on both sides of the track fails, Bilateral failure of fasteners refers to the failure of a single fastener on both sides; Figure 6 It is a time domain vibration waveform diagram collected by the present invention when the single-side fastener interval of the track fails, and Failure of fastener intervals is the single-side fastener interval failure; Figure 7It is a time domain vibration waveform diagram collected by the present invention when all the fasteners on both sides of the track fail. All the fasteners failed means that all the fasteners on both sides fail. Figure 8 It is a time domain vibration waveform diagram collected by the present invention when there is external invasion of foreign objects on the track, and foreigninvasion means there is external invasion of foreign objects; Fig. 9 It is a time domain vibration waveform diagram collected by the present invention when a pedestrian passes by on the track, and Pedestrian trespassing means that a pedestrian passes by; Fig.10 It is a time domain vibration waveform diagram of environmental noise on the track collected by the present invention, and environmental noise is environmental noise.

[0051] like Figure 11-18 As shown, the present invention provides a reference diagram for evaluating the time domain relationship of ultra-weak fiber Bragg grating vibration optical cable. Fig.11 This is a frequency domain vibration waveform diagram collected by the present invention when the fastener is fully fastened, where the horizontal axis Frequency is the frequency; Fig.12 This is a frequency domain vibration waveform diagram collected by the present invention when a single fastener on one side of the track fails;

[0052] Fig.13 This is a frequency domain vibration waveform diagram collected by the present invention when a single fastener on both sides of the track fails; Fig.14 The frequency domain vibration waveform diagram collected by the present invention when the interval of the single-side fastener of the track fails; Fig.15 This is a frequency domain vibration waveform diagram collected by the present invention when all the fasteners on both sides of the track fail; Fig.16 This is a frequency domain vibration waveform diagram collected by the present invention when there is external invasion of foreign objects on the track; Fig.17 The frequency domain vibration waveform diagram collected by the present invention when a pedestrian passes by on the track; Fig.18 This is a frequency domain vibration waveform diagram of the environmental noise on the track collected by the present invention.

[0053] S2. Based on the whale optimization method and variational modal decomposition method, the rail vibration signal under each working condition is decomposed, the optimal modal component of each working condition is extracted, and the optimal modal component of the rail vibration signal under each working condition is obtained.

[0054] In an optional embodiment of the present invention, the maximum number of iterations of the whale optimization method in the present invention is set to 50, and the population size is set to 30, which can provide moderate search diversity and computational burden while ensuring the calculation speed to find a better solution.

[0055] Step S2 includes the following steps:

[0056] S21. Based on the whale optimization method, envelope entropy is used as the fitness function, the position vector of the whale group is initialized, the fitness of each whale is calculated, and the iterative formula is selected by judging the size of the convergence factor for iterative update to determine the optimization parameters of the variational mode decomposition.

[0057] The convergence factor in the present invention is a dynamically adjusted parameter. As the number of iterations increases, the convergence factor will decrease to ensure that the whale optimization method has stronger exploration capabilities in the early stages and enhanced development capabilities in the later stages. The convergence factor is expressed as a linearly or nonlinearly decreasing parameter, which is used to control the values ​​of other key parameters and affect the movement direction and step size of the whale. The value of the convergence factor will affect the updated position of the whale. When the position of the whale is updated, the fitness of the whale at the new position needs to be recalculated. Therefore, the convergence factor will affect the calculation results of the fitness of each whale. A large convergence factor allows the whale to make large jumps in the search space, so that a better solution can be found; while a small convergence factor will prompt the whale to search in the local vicinity of the solution, refine the search process, and find the local optimal solution.

[0058] The present invention initializes the position vector of the whale group and calculates the fitness of each whale. If a whale individual appears with a better position than the current optimal solution, it is determined as the new optimal solution and the whale position is updated, which is expressed as:

[0059] D=|CX * (t)-X(t)|

[0060] X(t+1)=X * (t)-A·D

[0061] Where: D is the step size, t is the current iteration number, A is the first coefficient vector, C is the second coefficient vector, X * (t) is the position of the optimal solution obtained up to the current iteration, X(t) is the whale position at the current iteration, X(t+1) is the whale position at the next iteration, and A·D is the encirclement step size.

[0062] The first coefficient vector and the second coefficient vector are expressed as:

[0063] A=2a·ra

[0064] C=2·r

[0065] Where: r is a random number between [0,1], a is the convergence factor, which decreases linearly from 2 to 0 during the iteration process, that is,

[0066] a=2-2t / t max

[0067] Where: t max is the maximum number of iterations.

[0068] S22. Based on the variational modal decomposition method, the rail vibration signal is decomposed according to the optimization parameters of the variational modal decomposition, the optimal modal components of each working condition are extracted, and the optimal modal components of the rail vibration signal under each working condition are obtained.

[0069] S3. Determine the influencing index of rail vibration, and calculate the time domain eigenvalue of the rail vibration signal under each working condition according to the optimal modal component of the rail vibration signal under each working condition and the influencing index of rail vibration.

[0070] In an optional embodiment of the present invention, the influencing indexes of rail vibration include maximum value, minimum value, average value, effective value, standard deviation, kurtosis, skewness, waveform index and pulse index of rail vibration.

[0071] S4. A track fastener loosening identification model is constructed based on a convolutional neural network and a bidirectional long short-term memory network. The track fastener loosening identification result is obtained according to the time domain eigenvalues ​​of the rail vibration signal under various working conditions and the track fastener loosening identification model.

[0072] In an optional embodiment of the present invention, the present invention constructs a track fastener loosening recognition model based on a convolutional neural network and a bidirectional long short-term memory network, such as Fig.19 As shown, the rail fastener loosening recognition model includes an input layer, a sequence folding layer, a convolution layer, a pooling layer, a Flattern layer, a BiLSTM layer, a dropout layer and a fully connected layer which are connected in sequence, and a BN layer is set between the convolution layer and the pooling layer.

[0073] The input layer is used to convert the time domain eigenvalues ​​of the rail vibration signal under various working conditions into one-dimensional sequence data of characteristic parameters.

[0074] The sequence folding layer is used to convert the one-dimensional sequence data of the feature parameters into the tensor data of the feature parameters. Specifically, a sliding window is constructed, and the vibration sequence of the input audio signal is moved through the sliding window to extract a subsequence at each position. In this way, the one-dimensional sequence data of the feature parameters is converted into the tensor data of the feature parameters.

[0075] The convolution layer is used to convolve the tensor data of the feature parameters and extract the features of the tensor data.

[0076] The BN layer is used to standardize, scale and offset the features of tensor data to accelerate the training process and prevent gradient vanishing or gradient exploding. The present invention standardizes the features of tensor data so that its mean is 0 and its variance is 1 to reduce the change in data distribution. The present invention scales and offsets the standardized data by introducing learnable parameter scale factors and offset factors, so that the model can restore the flexibility of the original data distribution, thereby accelerating the training process and preventing gradient vanishing or gradient exploding.

[0077] The pooling layer is used to filter the features of tensor data and obtain filtered feature data.

[0078] The Flattern layer is used to flatten the filtered feature data to obtain the one-dimensional data of the filtered features.

[0079] The BiLSTM layer is used to obtain the temporal dependency of the filtering feature one-dimensional data before and after, so as to extract the deep features of the filtering feature one-dimensional data. The BiLSTM layer is specifically a 1-layer BiLSTM network layer with 30 neurons.

[0080] The dropout layer is used to set the probability in the forward propagation process to select the deep features of the one-dimensional data of the filter feature. The dropout layer can make the track fastener loosening identification model more generalized, reduce the dependence of the data on local features, thereby ensuring sparsity and reducing the risk of overfitting.

[0081] The fully connected layer maps the feature space to the sample label space based on the deep features of the selected filtered feature one-dimensional data to obtain the rail fastener loosening identification results. The fully connected layer can reduce the influence of feature position on the classification results and improve the robustness of the rail fastener loosening identification model.

[0082] like Fig. 20 As shown, the present invention uses the time domain eigenvalue pairs of rail vibration signals under various working conditions to test, and the accuracy of obtaining the track fastener loosening identification model for track fastener identification is 97.1429%. In contrast, the recognition accuracy of existing methods (such as traditional methods based on visual detection or conventional sensors) is between 85% and 90%. Compared with existing methods, the present invention has better accuracy, reliability and stability, specifically: accuracy, the recognition accuracy of the present invention is 97.1429%, which is higher than the 85%-90% of the existing methods, showing higher detection accuracy; reliability, due to the use of ultra-weak fiber grating technology, the present invention can more accurately capture tiny vibration changes, thereby improving the detection reliability of track fastener loosening; stability, compared with existing methods, the present invention has better adaptability and stability to vibration signals under various working conditions, reducing the possibility of false detection and missed detection. Through the above comparison, it can be seen that the present invention has significant advantages in the identification of loose track fasteners.

[0083] like Fig.21 As shown, in order to more clearly demonstrate the recognition results of the rail fastener loosening identification model for each category under each working condition, the present invention introduces a confusion matrix to analyze the prediction results of the rail fastener loosening identification model. The main diagonal elements of the confusion matrix represent the accuracy of the recognition of each type of working condition, and the secondary diagonal elements represent the error rate of each type of working condition.

[0084] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0085] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0086] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0087] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

[0088] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable, characterized in that: The following steps are involved: S1. Build a rail vibration signal acquisition system based on ultra-weak fiber Bragg grating vibration optical cable to obtain rail vibration signals under various working conditions; S2. Based on the whale optimization method and the variational modal decomposition method, the rail vibration signal under each working condition is decomposed, the optimal modal component of each working condition is extracted, and the optimal modal component of the rail vibration signal under each working condition is obtained; S3, determining the influence index of rail vibration, and calculating the time domain characteristic value of the rail vibration signal under each working condition according to the optimal modal component of the rail vibration signal under each working condition and the influence index of the rail vibration; S4. A track fastener loosening identification model is constructed based on a convolutional neural network and a bidirectional long short-term memory network. The track fastener loosening identification result is obtained according to the time domain eigenvalues ​​of the rail vibration signal under various working conditions and the track fastener loosening identification model.

2. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 1, characterized in that: In step S1, a rail vibration signal acquisition system is built based on the ultra-weak fiber Bragg grating vibration cable, specifically: determining the working section of the ultra-weak fiber Bragg grating vibration cable, serially welding the ultra-weak fiber Bragg grating vibration cable, arranging the ultra-weak fiber Bragg grating vibration cable by slotting and burying, connecting it to a vibration demodulator, and setting the excitation point position on the track plate of the rail to build the rail vibration signal acquisition system.

3. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 1, characterized in that: In step S1, the working conditions include fasteners being fully tightened, a single fastener on one side of the track failing, a single fastener on both sides of the track failing, an interval fastener on one side of the track failing, all fasteners on both sides of the track failing, foreign objects intruding from outside on the track, and pedestrians passing by on the track.

4. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 1, characterized in that: Step S2 includes the following steps: S21. Based on the whale optimization method, the envelope entropy is used as the fitness function to initialize the position vector of the whale group, calculate the fitness of each whale, and select the iterative formula for iterative update by judging the size of the convergence factor to determine the optimization parameters of the variational mode decomposition; S22. Based on the variational modal decomposition method, the rail vibration signal is decomposed according to the optimization parameters of the variational modal decomposition, the optimal modal components of each working condition are extracted, and the optimal modal components of the rail vibration signal under each working condition are obtained.

5. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 1, characterized in that: In step S3, the influencing indexes of the rail vibration include the maximum value, minimum value, average value, effective value, standard deviation, kurtosis, skewness, waveform index and pulse index of the rail vibration.

6. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 1, characterized in that: In step S4, the rail fastener loosening identification model includes an input layer, a sequence folding layer, a convolution layer, a pooling layer, a Flattern layer, a BiLSTM layer, a dropout layer and a fully connected layer connected in sequence; the input layer is used to convert the time domain characteristic values ​​of the rail vibration signal under various working conditions into one-dimensional sequence data of characteristic parameters, and the sequence folding layer is used to convert the one-dimensional sequence data of the characteristic parameters into tensor data of the characteristic parameters; the convolution layer is used to convolve the tensor data of the characteristic parameters to extract the features of the tensor data; the pooling layer is used to filter the features of the tensor data to obtain filtered feature data; the Flattern layer is used to flatten the filtered feature data to obtain filtered feature one-dimensional data; The BiLSTM layer is used to obtain the temporal dependency of the filtered feature one-dimensional data to extract the deep features of the filtered feature one-dimensional data; The dropout layer is used to set the probability during the forward propagation process to select the deep features of the filtered one-dimensional data; The fully connected layer maps the feature space to the sample label space based on the deep features of the selected filtered feature one-dimensional data to obtain the loose rail fastener recognition results.

7. The method for identifying loose rail fasteners based on ultra-weak fiber Bragg grating vibration optical cable according to claim 6, characterized in that: A BN layer is set between the convolution layer and the pooling layer. The BN layer is used to standardize, scale, and offset the features of the tensor data to speed up the training process and prevent gradient disappearance or gradient explosion.

Images