Bridge structure damage identification method and system

Abstract

The invention discloses a bridge structure damage identification method and system, and the method comprises the steps: carrying out the feature orthogonal decomposition of a displacement response measured on a bridge under the load of a moving vehicle, obtaining a principal component matrix, and carrying out the high-frequency power component information filtering of the principal component matrix, and obtaining a modal shape; inputting the modal shape into a preset bridge structure damage identification model, and performing feature extraction on the modal shape by the bridge structure damage identification model to obtain a combined feature; and recognizing the combined features according to a hybrid migration self-organizing signal module in the bridge structure damage recognition model to obtain a damage recognition result corresponding to the combined features. The method can accurately and efficiently improve the requirement of the neural network for adaptive bridge structure damage identification, and increases the accuracy of damage identification.

Description

Technical Field

[0001] The present invention belongs to the technical field of bridge structures, and in particular relates to a bridge structure damage identification method and system. Background Art

[0002] The actual environment in which bridge structures are located is extremely complex and changeable, filled with a large amount of noise interference such as random vibrations caused by vehicle driving and irregular effects of wind loads. This causes a lot of interference information to be mixed into the data collected for damage identification, greatly affecting the accuracy of damage identification.

[0003] Sensor-based monitoring method: Sensor deployment is complex and costly - bridges usually have large spans, and sensors need to be distributed in various key locations. In addition, the procurement, installation, and maintenance costs of the sensors themselves are high. Data collection limitations - the number and layout of sensors are often limited, making it difficult to fully cover all parts of the bridge, especially for some hidden areas. Signal noise problem - the data collected by sensors is often affected by the external environment (such as traffic vibration, temperature changes, wind, etc.), and the signal noise is large, which may lead to inaccurate data analysis results.

[0004] Recognition methods based on image processing: Environmental interference factors - image recognition methods are greatly affected by lighting, weather conditions (such as rain, snow, haze), obstructions, etc., resulting in unstable image quality; large amount of calculation - especially on complex bridge structures, the amount of image data is huge, processing requires a lot of computing resources, and the real-time requirements of image processing algorithms are high; lack of deep detail information - image processing usually focuses on surface damage such as surface cracks, but it is difficult to accurately identify internal damage to the bridge, such as corrosion or fractures, through image analysis. Summary of the invention

[0005] The present invention provides a bridge structure damage identification method and system, which are used to solve the technical problem that damage inside the bridge, such as corrosion or fracture, is difficult to accurately identify through image analysis.

[0006] In a first aspect, the present invention provides a bridge structure damage identification method, comprising: Performing characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filtering the high-frequency dynamic component information of the principal component matrix to obtain the modal vibration shape; Inputting the modal vibration shape into a preset bridge structure damage identification model, the bridge structure damage identification model extracts features from the modal vibration shape to obtain combined features; The combined feature is identified according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

[0007] In a second aspect, the present invention provides a bridge structure damage identification system, comprising: A decomposition module is configured to perform characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filter the high-frequency dynamic component information of the principal component matrix to obtain a modal vibration shape; An extraction module is configured to input the modal vibration shape into a preset bridge structure damage identification model, and the bridge structure damage identification model performs feature extraction on the modal vibration shape to obtain a combined feature; The identification module is configured to identify the combined feature according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

[0008] 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 bridge structure damage identification method of any embodiment of the present invention.

[0009] 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 is caused to execute the steps of the bridge structure damage identification method of any embodiment of the present invention.

[0010] The bridge structure damage identification method and system of the present application proposes a hybrid migration self-organizing signal module to improve the accuracy of the neural network in identifying bridge structure damage under limited data conditions. This module is inserted into the neural network to accurately and efficiently improve the neural network's adaptation to the needs of bridge structure damage identification and increase the accuracy of damage identification. A large model integrated low-rank fine-tuning strategy is also proposed to batch process the data, perform gradient updates on the original training weights, and train the low-rank matrix to capture the bridge structure damage characteristics for adjustment, thereby improving the robustness of the model and continuously optimizing the adaptability and accuracy of the model. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] 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.

[0012] Figure 1 A flow chart of a bridge structure damage identification method provided by an embodiment of the present invention; Figure 2 A structural block diagram of a bridge structure damage identification 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

[0013] 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.

[0014] See also Figure 1 , which shows a flow chart of a bridge structure damage identification method of the present application.

[0015] like Figure 1 As shown in FIG. 1 , the bridge structure damage identification method specifically includes the following steps: Step S101, performing characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filtering the high-frequency dynamic component information of the principal component matrix to obtain a modal vibration shape.

[0016] In this step, a moving load passes through a homogeneous simply supported beam with a length of L, and its dynamic differential equation is: , In the formula, They represent the vertical displacement, density damping coefficient, elastic modulus, and moment of inertia of the simply supported beam, respectively. denote the vertical moving load and moving speed respectively, is the load position; Using the mode superposition method, the formula of mode superposition is expressed as:

[0017] In the formula, is the mode superposition formula, is the coordinate of the nth-order vibration mode, is the vertical displacement response of the nth-order vibration mode simply supported beam under the action of moving load; Specifically: , In the formula, is the vertical moving load, is the density damping coefficient, is the density damping, is the structural natural frequency, is a dimensionless parameter, is the damping ratio, is the excitation frequency of the moving load, is the vibration time, is the offset angle, is the parameter; The analytical solution of the vertical displacement response is decomposed by POD. Since most bridges are made of steel and concrete, the damping coefficient is small and the material density is large, the damping ratio is an infinitesimal quantity, so the damping effect can be ignored, and the formula for the undamped principal component is: , In the formula, is the undamped principal component formula, is the eigenvector of the covariance matrix of the displacement response matrix, is the displacement signal response matrix, is the response 1 coefficient, is the response 2 coefficient, is the response m coefficient, is the response time 1, is the response time k; In order to obtain the modal vibration shape, the dynamic component information needs to be filtered to directly filter out the signal of a specific frequency. The filter formula is expressed as: , In the formula, is the filter formula, is the nth frequency of the bridge, is the sampling frequency, Reason Calculate the filter length, is the nth-order mode shape, is the number of signals, is the total number of signals; The mth-order modal vibration shape can be obtained by filtering the high-frequency dynamic component information with a low-pass filter. This method can obtain high-resolution modal vibration shapes with a small number of sensors. For example, the first three orders of high-resolution modal vibration shapes can be obtained by arranging three sensors.

[0018] Step S102: input the modal vibration shape into a preset bridge structure damage identification model, and the bridge structure damage identification model extracts features from the modal vibration shape to obtain combined features.

[0019] In this step, it is assumed that the input feature map size is , the convolution kernel size is , the output feature map size is , then the computational cost of depth-wise separable convolution is for: , In the formula, and are the width and height of the feature map respectively; and are the width and height of the convolution kernel respectively; M is the number of input channels; N is the number of output channels. The ratio x of the computational complexity of depth-separable convolution to that of standard convolution is: , In the formula, the computational complexity of depthwise separable convolution is equal to the computational complexity of standard convolution. The depth-separable convolution layer reduces parameters and leads to performance degradation. Dynamic convolution sets K convolution kernels with the same scale and number of channels, and extracts the weights through attention. The convolution kernel parameters of this layer are obtained by fusion. Therefore, dynamic separable convolution will bring about additional calculations of attention weight calculation and dynamic weight fusion. The additional calculation amount of attention weight calculation is for: , Additional computational effort when convolution kernels are fused for: , In the formula, K is the number of convolution kernels that need to be fused. Although the dynamic convolution layer causes additional overhead in two parts, it is far less than the computational complexity of static convolution. , so it satisfies The overhead caused by this part is very small. The introduction of dynamic convolution can effectively reduce the number of floating point operations (FLOPs) while improving model performance.

[0020] Step S103, identifying the combined feature according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature.

[0021] In this step, the expression for identifying the combined features is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

[0022] It should be noted that the expression for calculating the update gate variable factor is: , In the formula, To update the gate variable factor coefficient, is the upper bound of the update gate variable factor amplitude, To update the gate variable factor harmonic component, To update the gate variable factor jump component, The upper bound of the jump component amplitude of the update gate variable factor; The expression for calculating the reset gate variable factor is: , In the formula, To reset the gate variable factor, To reset the time domain component of the gate variable factor, To reset the gate variable factor time domain component number, To reset the number of gate variable factors; The expression for calculating the reset state variable factor is: , In the formula, To reset the time, To reset the state coefficient, To reset the state number, To reset the number of Gaussian components of the state variable factor, To reset the Gaussian component weights of the state variable factors, To reset the Gaussian component weights of the state variable factors, To reset the state, is the Gaussian component sequence of the reset state variable factor, is the Gaussian component variance matrix of the reset state variable factor, To reset the state variable factor reset probability, Reward for resetting state variables factor; , In the formula, is the number of hidden state variables, The number of reconstruction components for the hidden state variable factor, To update the hidden state variable factor reconstruction component, To update the hidden state variable factor low-pass signal, To update the hidden state variable factor error component, To update the hidden state variable factor error state.

[0023] In a specific embodiment, before the modal vibration shape is input into a preset bridge structure damage identification model, a dynamic deep separable convolutional neural network is constructed, and the dynamic deep separable convolutional neural network is trained based on a measured data set to obtain a network model, wherein the measured data set includes at least one measured data, a bridge vibration signal in the measured data, and a structural damage label corresponding to the bridge vibration signal; Migrating the network structure and hyperparameters in the dynamic depth separable convolutional neural network to corresponding positions in the network model to obtain a training model; The training model is fine-tuned according to the preset large model integrated low-rank fine-tuning strategy to obtain a bridge structure damage identification model.

[0024] It should be noted that the expression of the loss function in the dynamic depth separable convolutional neural network is: , , , , , In the formula, is the total loss function of the structural fusion loss model, is the weight of the cross entropy loss function, is the cross entropy loss function, is the supervised learning loss function, is the multivariate cross entropy quantity, is the loss weight corresponding to the actual semantic segmentation head, is the multivariate cross entropy loss function, is the loss weight of the domain identification loss, is the actual pixel-level classifier, is the common feature encoder, For a given source domain image, To predict the loss weight corresponding to the semantic segmentation head, To predict the pixel-level classifier, is the number of all-0 matrices, is the true domain label of sample i, is the feature of the target domain, is the discrete feature of the target domain, is the continuous feature of the target domain, is the discriminator, is the domain image set, is the common feature of the source domain image, are the common features of the target domain images.

[0025] Specifically, migrating the network structure and hyperparameters in the dynamic depth separable convolutional neural network to corresponding positions in the network model to obtain a training model includes: Based on the preset hyperparameter dynamic error supervision constraint migration strategy, the network structure and hyperparameters in the dynamic depth separable convolutional neural network are migrated to the corresponding positions in the network model to obtain a training model, wherein the expression of the hyperparameter dynamic error supervision constraint migration strategy is: , In the formula, is the constraint migration formula, is the number of hyperparameter categories, is the channel dimension addition, For information filtering formula, is the amount of error correction information, is the entropy constrained optimization formula, is the index coefficient; , In the formula, is the target entropy quantity, is the standard cross entropy quantity, is the entropy quantity, is the total number of category classification cross entropy, is the number of category classification cross entropy, is the category classification cross entropy, is the total number of category classification errors, is the standard cross entropy, is the minimum sort entropy, To compensate for the target, To cheat weights, To compensate for classification errors, is the target entropy, is the target weight; , In the formula, Reconstruct the sequence number for the index, is the scaling point, is the sequence information, For the splicing sequence, is the hidden sequence, To evaluate the score, For evaluation score compensation, To count down the evaluation score, For unevenness, is the macro-check accuracy, To override the parameters, are overlapping sequences; , In the formula, is the error elasticity coefficient, is the error elasticity, is the simulation correction amount, is the error curvature value, is the error sensitivity, is the channel capacity, is the bit error correction rate; , In the formula, Decompose the weights for the categories, is the learning rate, To confuse categories, is a discrete sequence, is Nth power, is the accuracy, is a discrete category, It is lossless category; , In the formula, is the information oscillation weight, is the information accuracy, For information Mapping volume, For information Model weights, is the information density rate, Scaling angle for information i, To extend the information mapping angle, is the information time series, is the information 1 bijective relation quantity, is the information logic condition weight, For information Model weights, is added along the channel dimension, Scaling sequence for information 2, For information Original coefficients.

[0026] The expression of the large model integrated low-rank fine-tuning strategy is: , In the formula, is the low-rank fine-tuning formula, is the frequency reconstruction factor, To judge the weight, To reconstruct the macro strain, is the regression weight, is the evaluation matrix, is the mth power, To update the number of parameters, is the autocorrelation function value, is T power, is the damage point coefficient; , In the formula, For external motivation, is the dynamic factor, is the node factor, is the amplitude factor, All are factor information. is the factor number; , In the formula, To reconstruct a single frequency, To reconstruct the individual quantities, To reconstruct the correlation function weights, is the weight of the multi-reconstruction correlation function, To reconstruct multiple frequencies; , In the formula, is the regression weight, is the lower bound, is the true value, is the observed value, is the test error, is the root mean square error; , In the formula, is the evaluation matrix, is the parameter membership, is the satisfaction weight transposition, is the second satisfaction weight transposition, is the N-times satisfaction weight transposition, is the satisfaction weight.

[0027] See also Figure 2 , which shows a structural block diagram of a bridge structure damage identification system of the present application.

[0028] like Figure 2 As shown, the bridge structure damage identification system 200 includes a decomposition module 210 , an extraction module 220 and an identification module 230 .

[0029] The decomposition module 210 is configured to perform characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filter the high-frequency dynamic component information of the principal component matrix to obtain a modal vibration shape; An extraction module 220 is configured to input the modal vibration shape into a preset bridge structure damage identification model, and the bridge structure damage identification model performs feature extraction on the modal vibration shape to obtain a combined feature; The identification module 230 is configured to identify the combined feature according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, To summarize, traditional neural networks require a large amount of labeled data for training, which is prone to overfitting. The method of this application is to improve the accuracy of neural network in identifying bridge structure damage under limited data conditions. A hybrid migration self-organizing signal module is proposed. This module is inserted into the neural network to accurately and efficiently improve the neural network's adaptation to the needs of bridge structure damage identification and increase the accuracy of damage identification. A large model integrated low-rank fine-tuning strategy is proposed to batch process the data, perform gradient updates on the original training weights, and train the low-rank matrix to capture the bridge structure damage characteristics for adjustment, thereby improving the robustness of the model and continuously optimizing the adaptability and accuracy of the model.

[0030] 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.

[0031] In some 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 is caused to execute the bridge structure damage identification 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: Performing characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filtering the high-frequency dynamic component information of the principal component matrix to obtain the modal vibration shape; Inputting the modal vibration shape into a preset bridge structure damage identification model, the bridge structure damage identification model extracts features from the modal vibration shape to obtain combined features; The combined feature is identified according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

[0032] 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 application programs required for at least one function; the data storage area may store data created according to the use of the bridge structure 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 bridge structure 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.

[0033] 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 3The example of connecting via a 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 bridge structure damage identification 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 bridge structure damage identification system. The output device 340 may include display devices such as display screens.

[0034] 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.

[0035] As an implementation mode, the electronic device is applied to a bridge structure 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: Performing characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filtering the high-frequency dynamic component information of the principal component matrix to obtain the modal vibration shape; Inputting the modal vibration shape into a preset bridge structure damage identification model, the bridge structure damage identification model extracts features from the modal vibration shape to obtain combined features; The combined feature is identified according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: , In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

[0036] 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.

[0037] 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 bridge structure damage identification method, characterized in that: include: Performing characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filtering the high-frequency dynamic component information of the principal component matrix to obtain the modal vibration shape; The modal vibration shape is input into a preset bridge structure damage identification model, and a principal component analysis module of the bridge structure damage identification model performs feature extraction on the modal vibration shape to obtain a combined feature; The combined feature is identified according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: ， In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

2. A bridge structure damage identification method according to claim 1, characterized in that: in, The expression for calculating the update gate variable factor is: ， In the formula, To update the gate variable factor coefficient, is the upper bound of the update gate variable factor amplitude, To update the gate variable factor harmonic component, To update the gate variable factor jump component, The upper bound of the jump component amplitude of the update gate variable factor; The expression for calculating the reset gate variable factor is: ， In the formula, To reset the gate variable factor, To reset the time domain component of the gate variable factor, To reset the gate variable factor time domain component number, To reset the number of gate variable factors; The expression for calculating the reset state variable factor is: ， In the formula, To reset the time, To reset the state coefficients, To reset the state number, To reset the number of Gaussian components of the state variable factor, To reset the Gaussian component weights of the state variable factors, To reset the Gaussian component weights of the state variable factors, To reset the state, is the Gaussian component sequence of the reset state variable factor, is the Gaussian component variance matrix of the reset state variable factor, To reset the state variable factor reset probability, Reward for resetting state variables factor; ， In the formula, is the number of hidden state variables, The number of reconstruction components for the hidden state variable factor, To update the hidden state variable factor reconstruction component, To update the hidden state variable factor low-pass signal, To update the hidden state variable factor error component, To update the hidden state variable factor error state.

3. A bridge structure damage identification method according to claim 1, characterized in that: Before inputting the modal vibration shape into a preset bridge structure damage identification model, the method further includes: A dynamic deep separable convolutional neural network is constructed, and the dynamic deep separable convolutional neural network is trained based on a measured data set to obtain a network model, wherein the measured data set includes at least one measured data, and one measured data includes a bridge vibration signal and a structural damage label corresponding to the bridge vibration signal; Migrating the network structure and hyperparameters in the dynamic depth separable convolutional neural network to corresponding positions in the network model to obtain a training model; The training model is fine-tuned according to the preset large model integrated low-rank fine-tuning strategy to obtain a bridge structure damage identification model.

4. A bridge structure damage identification method according to claim 3, characterized in that: in, The dynamic depth separable convolutional neural network includes a dynamic convolutional layer, a dynamic depth separable convolutional layer, a global average pooling layer and a fully connected layer; the expression of the loss function in the dynamic depth separable convolutional neural network is: ， ， ， ， ， In the formula, is the total loss function of the structural fusion loss model, is the weight of the cross entropy loss function, is the cross entropy loss function, is the supervised learning loss function, is the multivariate cross entropy quantity, is the loss weight corresponding to the actual semantic segmentation head, is the multivariate cross entropy loss function, is the loss weight of the domain identification loss, is the actual pixel-level classifier, is the common feature encoder, For a given source domain image, To predict the loss weight corresponding to the semantic segmentation head, To predict the pixel-level classifier, is the number of all-0 matrices, is the true domain label of sample i, is the feature of the target domain, is the discrete feature of the target domain, is the continuous feature of the target domain, is the discriminator, is the domain image set, is the common feature of the source domain image, are the common features of the target domain images.

5. A bridge structure damage identification method according to claim 3, characterized in that: The step of migrating the network structure and hyperparameters in the dynamic depth separable convolutional neural network to corresponding positions in the network model to obtain a training model includes: Based on the preset hyperparameter dynamic error supervision constraint migration strategy, the network structure and hyperparameters in the dynamic depth separable convolutional neural network are migrated to the corresponding positions in the network model to obtain a training model, wherein the expression of the hyperparameter dynamic error supervision constraint migration strategy is: ， In the formula, is the constraint migration formula, is the number of hyperparameter categories, is the sum of channel dimensions, For information filtering formula, is the amount of error correction information, is the entropy constrained optimization formula, is the index coefficient; ， In the formula, is the target entropy quantity, is the standard cross entropy quantity, is the entropy quantity, is the total number of category classification cross entropy, is the number of category classification cross entropy, is the category classification cross entropy, is the total number of category classification errors, is the standard cross entropy, is the minimum sort entropy, To compensate for the target, To cheat weights, To compensate for classification errors, is the target entropy, is the target weight; ， In the formula, Reconstruct the sequence number for the index, is the scaling point, is the sequence information, For the splicing sequence, is the hidden sequence, To evaluate the score, For evaluation score compensation, To count down the evaluation score, For unevenness, is the macro-check accuracy, To override the parameters, are overlapping sequences; ， In the formula, is the error elasticity coefficient, is the error elasticity, is the simulation correction amount, is the error curvature value, is the error sensitivity, is the channel capacity, is the bit error correction rate; ， In the formula, Decompose the weights for the categories, is the learning rate, To confuse categories, is a discrete sequence, is Nth power, is the accuracy, is a discrete category, It is lossless category; ， In the formula, is the information oscillation weight, is the information accuracy, For information Mapping volume, For information Model weights, is the information density rate, Scaling angle for information i, To extend the information mapping angle, is the information time series, is the information 1 bijective relation quantity, is the information logic condition weight, For information Model weights, is added along the channel dimension, Scaling sequence for information 2, For information Original coefficients.

6. A bridge structure damage identification method according to claim 3, characterized in that: in, The expression of the large model integrated low-rank fine-tuning strategy is: ， In the formula, is the low-rank fine-tuning formula, is the frequency reconstruction factor, To judge the weight, To reconstruct the macro strain, is the regression weight, is the evaluation matrix, is the mth power, To update the number of parameters, is the autocorrelation function value, is T power, is the damage point coefficient; ， In the formula, For external motivation, is the dynamic factor, is the node factor, is the amplitude factor, All are factor information. is the factor number; ， In the formula, To reconstruct a single frequency, To reconstruct the individual quantities, To reconstruct the correlation function weights, is the weight of the multi-reconstruction correlation function, To reconstruct multiple frequencies; ， In the formula, is the regression weight, is the lower bound, is the true value, is the observed value, is the test error, is the root mean square error; ， In the formula, is the evaluation matrix, is the parameter membership, is the satisfaction weight transposition, is the second satisfaction weight transposition, is the N-times satisfaction weight transposition, is the satisfaction weight.

7. A bridge structure damage identification system, characterized in that: include: A decomposition module is configured to perform characteristic orthogonal decomposition on the displacement response measured on the bridge under the load of the moving vehicle to obtain a principal component matrix, and filter the high-frequency dynamic component information of the principal component matrix to obtain a modal vibration shape; An extraction module is configured to input the modal vibration shape into a preset bridge structure damage identification model, and the bridge structure damage identification model performs feature extraction on the modal vibration shape to obtain a combined feature; The identification module is configured to identify the combined feature according to the hybrid migration self-organizing signal identification module in the bridge structure damage identification model to obtain a damage identification result corresponding to the combined feature, wherein the expression for identifying the combined feature is: ， In the formula, It is a hybrid migration self-organizing signal module. To update the gate variable factor, is the state variable, To reset the last moment status, To update the gate, To update the gate twin variable factor, To update the gate confidence probability, To update the door simulation state, To reset the gate, To reset the gate variable factor, To reset the gate twin variable factor, To reset the gate confidence probability, To reset the door simulation state, To reset the state, To reset the state variable factor, To reset the state twin variable factor, is the element-wise product, To reset the state decomposition function, To reset the state simulation state, To update the hidden state, To update the hidden state variable factor, For updating the door.

8. 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 6.

9. 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 6 is implemented.

Images