A safety status evaluation method and system for highway bridges

Abstract

The present invention discloses a method and system for evaluating the safety state of highway bridges, relating to the technical field of highway bridge state evaluation, including: collecting the time-series vibration data at multiple key nodes of the highway bridge and the overall surface image data; inputting each time-series vibration data into an improved generative adversarial network to generate denoised time-series vibration data, and then performing a preliminary state evaluation based on the denoised data to obtain an initial evaluation value; wherein, the generator in the network adopts an encoder-decoder structure, and a channel attention mechanism is added between feature encoding and decoding; identifying all cracks in the surface image data and dividing the areas of the cracks, calculating a secondary evaluation value according to the cracks in each area, and using the secondary evaluation value to assist in adjusting the initial evaluation value; for each adjusted evaluation value, integrating based on the support degree between data to obtain an overall evaluation result. The present invention can effectively improve the accuracy and efficiency of the safety state evaluation of highway bridges.

Description

Technical Field

[0001] The present invention relates to the technical field of highway bridge condition assessment, and particularly to a safety condition assessment method and system for highway bridges. Background Art

[0002] During the use of highway bridges, they are affected by traffic loads, environmental factors, natural disasters, etc. for a long time, which may cause structural damages or performance degradation problems such as cracks, deformations, and corrosion in the bridges, resulting in a decrease in the stability of highway bridges and affecting their safety. Therefore, to ensure the safety of highway bridges, it is necessary to conduct a safety condition assessment on highway bridges, and then carry out effective maintenance according to the assessment results to extend the service life of highway bridges and reduce the risk of traffic interruption and accidents.

[0003] Regarding the safety condition of highway bridges, traditional assessment methods mainly rely on regular inspections, embedded sensor detections, and manual inspections, etc. However, this method is costly, time-consuming and laborious, and it is easy to overlook hidden defects. To avoid the disadvantages of traditional methods, in the prior art, considering that there will be abnormal vibrations when the stability of the highway bridge structure changes, a method of obtaining highway bridge vibration data through detection and performing a safety condition assessment based on this vibration data is proposed. However, this method still has certain problems, that is:

[0004] (1) Usually, the data obtained by detecting through vibration sensors has noise, which makes it inaccurate to directly evaluate the safety of highway bridges based on the vibration data. To ensure the accuracy of the final assessment, it is necessary to denoise the vibration data. However, existing denoising methods such as mean filtering and wavelet transform all have certain disadvantages and cannot meet the current requirements for effectively denoising vibration data. For example, in mean filtering, the selection of the filtering window affects the final filtering and denoising effect, and this filtering window is usually manually selected, which cannot guarantee the filtering effect; wavelet transform requires multiple decomposition analyses of the data, its calculation cost is high, and the efficiency is low.

[0005] (2) Vibration data is often obtained based on vibration sensors distributed at key nodes on key sections of the bridge. These nodes are discrete. If there are no corresponding sensors configured at the damaged parts, it is impossible to determine whether local damage has occurred, which will affect the accuracy of judging the overall safety condition of the highway bridge; and if sensors are widely equipped on the highway bridge, not only is the cost high, the installation is time-consuming and laborious, but also the processing of a large amount of vibration data will further affect the efficiency of data analysis and processing. Summary of the Invention

[0006] To address the deficiencies of the above-mentioned existing technologies, the present invention provides a method and system for evaluating the safety status of highway bridges. By collecting multi-dimensional monitoring data of highway bridges: vibration data and surface image data, and jointly analyzing the multi-dimensional monitoring data, a more accurate and comprehensive evaluation of the safety status of highway bridges is achieved, solving the problem of poor accuracy in existing evaluations based solely on single data.

[0007] In a first aspect, the present invention provides a method for evaluating the safety status of highway bridges.

[0008] A method for evaluating the safety status of highway bridges includes:

[0009] Collecting time-series vibration data at multiple key nodes of the highway bridge and overall surface image data;

[0010] Inputting each time-series vibration data into a trained improved generative adversarial network to generate denoised time-series vibration data, and then performing a preliminary status evaluation based on the denoised data to obtain an initial evaluation value; in the improved generative adversarial network, the generator adopts an encoder-decoder structure and a channel attention mechanism is added between feature encoding and decoding;

[0011] Identifying all cracks in the surface image data, dividing the cracks into regions, calculating a secondary evaluation value based on the cracks in each region, and using the secondary evaluation value to assist in adjusting the initial evaluation value;

[0012] For each adjusted evaluation value, integrating based on the support degree between data to obtain an overall evaluation result of the safety status of the highway bridge.

[0013] In a second aspect, the present invention provides a system for evaluating the safety status of highway bridges.

[0014] A system for evaluating the safety status of highway bridges includes:

[0015] A data acquisition module for collecting time-series vibration data at multiple key nodes of the highway bridge and overall surface image data;

[0016] An initial evaluation module for inputting each time-series vibration data into a trained improved generative adversarial network to generate denoised time-series vibration data, and then performing a preliminary status evaluation based on the denoised data to obtain an initial evaluation value; in the improved generative adversarial network, the generator adopts an encoder-decoder structure and a channel attention mechanism is added between feature encoding and decoding;

[0017] A secondary adjustment module for identifying all cracks in the surface image data, dividing the cracks into regions, calculating a secondary evaluation value based on the cracks in each region, and using the secondary evaluation value to assist in adjusting the initial evaluation value;

[0018] An overall evaluation module is used to integrate the adjusted evaluation values based on the support degree between the data to obtain the overall evaluation result of the safety state of the highway bridge.

[0019] In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the steps of the method described in the first aspect are completed.

[0020] In a fourth aspect, the present invention also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by the processor, the steps of the method described in the first aspect are completed.

[0021] The above one or more technical solutions have the following beneficial effects:

[0022] 1. The present invention provides a method and system for evaluating the safety state of a highway bridge. By collecting multi-dimensional monitoring data of the highway bridge, that is, one-dimensional time-series vibration data at each key node and two-dimensional surface image data of the whole, the two-dimensional monitoring data are jointly analyzed to solve the problem of poor evaluation accuracy in safety evaluation based only on single-dimensional data. In the above solution, on the one hand, the time-series vibration data is used for preliminary monitoring and evaluation. On the other hand, an image recognition algorithm is used to identify the surface crack condition of the highway bridge, and the preliminary evaluation result is adjusted according to the local crack condition, so as to make up for the disadvantages of uneven distribution and inability to comprehensively reflect local losses that may exist in the vibration data, and further avoid the problem that the evaluation value with local errors affects the accuracy of the final global evaluation, and can also avoid the problem of low accuracy in single-dimensional evaluation using only single vibration data or crack conditions.

[0023] 2. The present invention uses an improved generative adversarial network to filter and denoise the time-series vibration data. The improved generative adversarial network includes a generator and a discriminator. The generator adopts a U-NET structure (encoder-decoder structure), and a channel attention mechanism is added between each corresponding encoding layer and decoding layer to enhance the effect of feature fusion and improve the convergence speed of the network model. The time-series vibration data is input into the trained network, and the generator extracts deep features and generates vibration data that is closer to the pure signal after denoising. Compared with the existing signal denoising methods such as time-domain-based, frequency-domain-based, and time-frequency analysis-based methods, it can achieve more efficient denoising and better denoising effect.

[0024] 3. Considering the high complexity of one-dimensional time-series vibration data, some important features such as partial time correlation are often ignored during feature extraction and analysis. Therefore, the present invention adopts a data conversion method to convert one-dimensional time-series vibration data into two-dimensional image data, and then analyzes and evaluates based on the two-dimensional image data, which can effectively increase the time correlation. The neural network can also evaluate and diagnose according to the distribution of each pixel block of the image. Compared with one-dimensional time-series data, two-dimensional images can highlight the internal features of the data, and the distribution rules of pixel blocks of various data images are clear, which is more conducive to the neural network to identify and improve the accuracy and efficiency of the final judgment.

[0025] 4. In the present invention, a secondary evaluation value based on crack recognition is used to assist in adjusting the initial evaluation value. Among them, the data consistency analysis method is adopted to analyze the consistency relationship between the initial evaluation and the secondary evaluation, and the initial evaluation result is assisted to be adjusted according to this relationship, so as to consider various influencing factors in multiple aspects and dimensions, ensure that the final evaluation result is closer to the actual situation, and ensure the accuracy of the evaluation; in addition, by analyzing the support degree between data, each evaluation value is integrated to obtain a total evaluation result that can cover the overall situation of the highway bridge, so as to accurately divide the safety level of the highway bridge safety status and provide data support for subsequent highway bridge maintenance measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The accompanying drawings forming a part of this specification are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention.

[0027] Figure 1 It is the overall flowchart of the safety status evaluation method for the highway bridge described in the embodiment of the present invention;

[0028] Figure 2 It is the structural schematic diagram of the improved adversarial generation network in the embodiment of the present invention;

[0029] Figure 3 It is the schematic diagram of the principle of the attention module in the improved adversarial generation network in the embodiment of the present invention;

[0030] Figure 4 It is the flow schematic diagram of crack target detection in the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0031] It should be noted that the following detailed description is exemplary only for the purpose of describing specific embodiments, aiming to provide further explanation of the present invention and not intended to limit the exemplary embodiments according to the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present invention belongs. In addition, it should also be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0032] Embodiment 1

[0033] This embodiment provides a method for evaluating the safety state of highway bridges, as Figure 1 shown, including the following steps:

[0034] Step S1: Collect time-series vibration data at multiple key nodes of the highway bridge and overall surface image data.

[0035] Step S1.1: Collect time-series vibration data at multiple key nodes of the highway bridge.

[0036] Specifically, vibration sensors are arranged at the key nodes of the critical sections of the highway bridge. The vibration sensors can use highly sensitive seismographs or acceleration sensors. Real-time monitoring is carried out using these vibration sensors, that is, monitoring the time-series vibration data at multiple key nodes of the highway bridge under a set traffic load within a set time period. The key nodes refer to the connection nodes between the bridge deck of the highway bridge and each pier; each vibration sensor is electrically connected to a data collector, which has a high sampling rate and can collect and capture the instantaneous vibration data of the bridge under a set traffic load such as when a vehicle passes, so as to obtain the time-series vibration data.

[0037] Preferably, after obtaining multiple time-series vibration data, linear interpolation is used to process each time-series vibration data to avoid data loss caused by fluctuations in the acquisition frequency during the process of collecting vibration data and ensure the accuracy of data collection.

[0038] Step S1.2: Collect overall surface image data of the highway bridge.

[0039] Specifically, use a drone or other equipment to obtain the surface images of the overall outer surface of the highway bridge. In fact, the overall structure of the highway bridge includes various component members, such as bridge span structures (such as bridge decks, etc.), bearings, piers, abutments, main girders, cantilevers, pier and abutment foundations, etc. In this step, the surface images of each component member of the highway bridge are obtained.

[0040] Step S2: Input each piece of time-series vibration data into the trained improved generative adversarial network to generate denoised time-series vibration data, and then perform a preliminary state assessment based on the denoised data to obtain an initial assessment value. Among them, in the improved generative adversarial network, the generator adopts an encoder-decoder structure, and a channel attention mechanism is added between feature encoding and decoding.

[0041] Specifically, for the time-series vibration data of each key node, use the improved adversarial generative network for denoising processing, convert the denoised time-series vibration data into a two-dimensional image, and input the two-dimensional image into the trained state assessment model to output the preliminary assessment value at each key node. In this embodiment, according to the layout order of n vibration sensors on the highway bridge from left to right (or from right to left), a sequence of preliminary assessment values at multiple key nodes is obtained. Among them, n represents the number of vibration sensors. represents the preliminary assessment value corresponding to the i-th key node.

[0042] Step S2.1: For the time-series vibration data of each key node, use the improved adversarial generative network for denoising processing to generate denoised time-series vibration data.

[0043] In the process of using sensors to measure and obtain time-series vibration data, there will inevitably be noise interference. For example, environmental anomalies may cause similar abnormal increases or drifts in the data collected by multiple sensors, sensor equipment itself has defects resulting in abnormal collected data, and large differences in accuracy between the deployed sensors may cause abnormal data such as data jumps, trends, missing, and drifts in the data collected by some sensors. Therefore, when evaluating the safety state of a highway bridge based on time-series vibration data, it is first necessary to exclude the noise interference caused by environmental anomalies and sensor anomalies through data cleaning and denoising processing to avoid errors or anomalies in subsequent analysis.

[0044] Since the existing denoising methods have low processing efficiency and poor accuracy, in this embodiment, a machine learning method is adopted to propose a vibration data denoising method based on an improved adversarial generative network model (GAN). This network is used to identify the noise data (or called noise signals) in the collected data and effectively suppress the noise signals, which can quickly and effectively remove the noise in the original time-series vibration data, and then perform an accurate safety state assessment based on the denoised time-series vibration data.

[0045] Specifically, for the time-series vibration data (also called vibration signals), in order to better retain useful signals and suppress noise signals, this embodiment adopts a denoising method based on an attention mechanism-improved generative adversarial network. As Figure 2As shown in the figure, the improved generative adversarial network includes a generator and a discriminator. First, the generator is used to remove the noise component in the noisy signal (i.e., the original time series vibration signal) and establish the mapping relationship between the noisy signal and the denoised signal. Second, the denoised signal generated by the generator and the theoretically pure signal are input into the discriminator for classification. If the output label of the discriminator is closer to 1, it is considered that the input data signal is closer to the pure signal; otherwise, the output label is closer to 0, that is, it is considered that the input data signal is closer to the denoised signal.

[0046] By pre-training the above improved generative adversarial network, the denoised signal generated by the generator is closer to the pure signal, so that the discriminator cannot judge whether the input is a pure signal. On this basis, the trained generator can generate a denoised vibration signal for the input original vibration signal with noise.

[0047] Among them, in the above improved adversarial generative network model, the generator adopts the U-NET structure (i.e., the encoder-decoder structure), including a feature encoding unit and a feature decoding unit. Among them, the feature encoding unit includes multiple encoding layers, and each encoding layer includes a downsampling unit. The input noisy signal undergoes multiple one-dimensional convolutions through multiple downsampling units to continuously compress and extract features. The feature decoding unit includes multiple decoding layers, and each decoding layer includes an upsampling unit. The compressed deep features are gradually restored to features of the same scale as the input signal through the one-dimensional transposed convolution of the upsampling unit. Moreover, the sizes of the feature maps extracted in the feature encoding unit and the feature decoding unit correspond one by one, and are spliced through skip connections to fuse low-dimensional features (i.e., encoded feature maps) and high-dimensional features (i.e., decoded feature maps). Finally, a denoised signal is generated based on the fused features.

[0048] Furthermore, to enhance the fusion effect of low-dimensional features and high-dimensional features, the U-NET structure of the generator is improved, and a channel attention mechanism is added between feature encoding and feature decoding, as Figure 3 shown. For each corresponding encoding layer and decoding layer, the fusion method of the extracted encoded feature map and decoded feature map is as follows:

[0049] (1) The encoded feature map x e output by the encoding layer and the decoded feature map x d output by the decoding layer are respectively convolved with a 1×1 convolution kernel with a stride of 1 to obtain two weight vectors W e and W d ; among them, the sizes of the encoded feature map x e and the decoded feature map x d are (B, H, C), where B, H, and C represent the length, width, and dimension respectively. In this embodiment, H = 256;

[0050] (2) Add the weight vectors W e and W d , and then activate the result using the LeakyRelu activation function to obtain the activated result W a1 ;

[0051] (3) The activated result is convolved with a 1×1 convolutional kernel and activated through Sigmoid to obtain the attention weight layer W a2 . This attention weight layer W a2 is resampled and then multiplied by the decoded feature map x d output by the current decoding layer to complete the attention activation of the decoded feature map x d .

[0052] By adding a channel attention mechanism to the U-NET structure of the generator as described above, the effect of feature fusion can be further enhanced, the convergence speed of the network model can be increased, and the training loss value can be made lower. Based on the improved generative adversarial network completed by the above training, the generator is used to process the input time-series vibration data to generate vibration data that is closer to the pure signal after denoising. This data denoising method avoids the complex data analysis and processing process, and has higher efficiency and better denoising effect.

[0053] Step S2.2: Convert the time-series vibration data after denoising processing into a two-dimensional image, and input the two-dimensional image into the state evaluation model to obtain an initial evaluation value.

[0054] Since the vibration data is one-dimensional time-series data with high complexity, it is difficult for a simple neural network to comprehensively extract the features of this time-series data, and important features such as time correlation are often ignored, affecting the accuracy of the final output result. Correspondingly, if a complex model is used for learning and training, the training and time costs will increase. Therefore, directly diagnosing and evaluating anomalies based on one-dimensional time-series vibration data has an unsatisfactory effect. For this reason, in this embodiment, converting the one-dimensional time-series vibration data into a two-dimensional image can effectively strengthen the time correlation. The neural network can evaluate and diagnose based on the distribution of each pixel block of the image. Compared with one-dimensional time-series data, the two-dimensional image can highlight the internal features of the data, and the distribution rules of the pixel blocks of various data images are clear, which is more conducive to the neural network to identify and improve the accuracy of the final judgment.

[0055] In this embodiment, the Gramian Summation Angular Field (GAF) method is used to convert the one-dimensional time-series data into a two-dimensional grayscale image, so as to better reflect the potential features between the time-series data and extract richer feature representations. As another implementation method, methods such as Markov Transition Field (MTF) and Recurrence Plot (RP) can also be used to convert the one-dimensional time-series data into a two-dimensional image.

[0056] Specifically, the principle of the Gramian Angular Field is to transform the scaled one-dimensional sequence data from the rectangular coordinate system to the polar coordinate system, and then identify the correlations at different time points by considering the sum or difference of the angles between different points. In this embodiment, the Gramian Difference Angular Field (GADF) method is adopted. Through angle difference calculation, the one-dimensional time series data is converted into a two-dimensional grayscale image, which specifically includes the following steps:

[0057] Step S2.2.1: For each denoised time series vibration data, such as the i-th time series vibration data x i ={x i1 , x i2 ,..., x ij ,...x im}, where i = 1, 2,..., n, j = 1, 2,..., m, and m represents the number of time series data. The data is normalized using the following formula to scale the sequence to the range [-1, 1]. The normalization formula is:

[0058]

[0059] Step S2.2.2: Encode the normalized data as the cosine of the vector angle The corresponding timestamp is encoded as the radius r j , then the time series in the Cartesian coordinate system can be transformed into a time series in the polar coordinate system, as follows:

[0060]

[0061]

[0062] In the above formula, t j is the timestamp, j represents the j-th data, and m is the sample size of the data.

[0063] Under the above encoding method, as time increases, the corresponding values are distorted between different angular points on the spanning circle. By transforming to the polar coordinate system, the polar coordinate maintains the time dependence of the sequence through the radius r j , and each sequence generates a unique polar coordinate mapping diagram.

[0064] Step S2.2.3: For the time series data with a length of n, the size of the converted GADF is an [n, n] matrix. Obtain the vector angle between two values using the following formula, calculate the sine value of the difference between the two vector angles to obtain the Gramian matrix, and draw an image based on this matrix to obtain a two-dimensional grayscale image.

[0065]

[0066] In the above formula, represents the cosine of the vector included angle.

[0067] Step S2.2.4: Input the two-dimensional grayscale image into the trained state evaluation model based on the BP network, output the confidence of the safety state of the highway bridge, and output this confidence as the evaluation value, which is the value obtained through the preliminary state evaluation. Among them, in this embodiment, the state evaluation model is built using a BP neural network. Using the time-series vibration data of several pre-labeled state values as the training set, the time-series vibration data is converted into a two-dimensional grayscale image, and the BP neural network is trained using this image training set to obtain the trained state evaluation model.

[0068] Step S3: Identify all cracks in the surface image data, divide the cracks into regions, calculate the secondary evaluation value according to the cracks in each region, and use the secondary evaluation value to assist in adjusting the initial evaluation value.

[0069] Step S3.1: Input the surface image of the overall outer surface of the highway bridge into the target detection model to identify the cracks in the image and detect the length and width of the cracks.

[0070] In this embodiment, a target detection model is built. This target detection model includes a crack recognition module and a crack extraction module. Use this model to identify the cracks in the surface image and extract the crack targets with a single-pixel width.

[0071] As Figure 4 shown, first, for the crack recognition module: Since the crack targets to be detected have the characteristics of small-size structures, it is easy to lose their detailed information during the feature extraction process, resulting in difficult crack recognition and low recognition accuracy. For this reason, in this embodiment, the VGG16 network is used as the feature extraction network to extract the global feature map of the input image using this network. This network only uses a 3×3 convolutional kernel. Under the condition of the same receptive field, it can increase the network depth while reducing the number of parameters, and reduce the probability of losing the details of the crack targets caused by convolutional operations. The feature map extracted by the VGG16 feature extraction network is input into the RPN network (Region Proposal Network), and a series of candidate boxes for identifying cracks are output through this network. The output candidate boxes are combined with the feature map and input into the RoIAlign layer to clip the part of the feature map corresponding to the candidate boxes, that is, the candidate box feature map. Finally, the candidate box feature map is classified and regressed through the fully connected layer to output the identified cracks.

[0072] Secondly, for the crack extraction module: The identified crack region images are successively subjected to binarization, denoising, and thinning operations to extract the crack targets from the crack regions. Specifically, using the adaptive threshold method, the crack targets are segmented from the background of the crack region images to generate binary images of the crack targets. Considering that the binarization process usually generates isolated foreground noises and target fractures, therefore, through fracture connection and denoising, a clean and complete binary image of the crack targets is obtained. Among them, the closing operation is used to bridge the fractured parts of the crack targets, smooth the boundaries of the crack targets under the condition of keeping the original area of the crack targets unchanged, and the connected component labeling method is used to remove the background noises; the thinning operation is performed on the binary image of the crack targets to obtain crack targets with a single-pixel width.

[0073] Furthermore, based on the proposed crack targets, the length and width values of the crack targets are obtained by calculating the number of pixel points of the crack targets.

[0074] Step S3.2: Divide all the identified cracks into regions. Specifically, according to the layout positions of n vibration sensors on the highway bridge, the entire highway bridge is correspondingly divided into n regions, and each region includes at least one component; then, based on the divided regions, all the identified cracks are divided into regions to determine all the cracks in each region.

[0075] Step S3.3: For all the cracks in each region, calculate the secondary evaluation value of each region of the highway bridge according to the crack length and width, and among them, corresponding to the region order from left to right (or from right to left), obtain the corresponding sequence of secondary evaluation values

[0076] Specifically, according to the identified cracks and their length and width in the current region, calculate the secondary evaluation value of this region Including:

[0077] Step S3.3.1: Calculate the crack area s of all the cracks in each region of the highway bridge q , and the crack area of the i-th region is:

[0078]

[0079] Among them, l q represents the crack length, h q represents the crack width, q represents the q-th crack identified in this region, and i = 1, 2,... n.

[0080] Step S3.3.2: Based on the crack area of each region, determine the secondary evaluation value of each region according to the ratio of the crack area to the total crack area The formula is:

[0081]

[0082] Step S3.4. Based on the data consistency analysis method, use the secondary evaluation value to assist in adjusting the initial evaluation value, specifically as follows:

[0083] Step S3.4.1. Normalize the initial evaluation value sequence and the secondary evaluation value sequence for multiple regions respectively.

[0084] Step S3.4.2. Based on the normalized initial evaluation value sequence and secondary evaluation value sequence, perform data consistency analysis on each pair of evaluation values, calculate the weight coefficient of each pair of evaluation values, and obtain the adjusted evaluation value sequence through weighted fusion, denoted as y1, y2,..., y i ,..., y n .

[0085] Specifically, based on the initial evaluation value sequence and the secondary evaluation value sequence analyze the consistency between each pair of evaluation values using the support function. The formula is as follows:

[0086]

[0087] In the above formula, represents the support degree between the two evaluation values at the i-th node (or the i-th region), and this support degree is denoted as the consistency coefficient; G[*] represents the support function, K ∈ [0, 1] is the amplitude of the support function, and β ≥ 0 is the attenuation rate of the support function.

[0088] Through the above method, obtain the consistency coefficient sequence, denoted as z1, z2,..., z i ,..., z n .

[0089] Furthermore, set the consistency coefficient setting threshold Z. In this embodiment, set the threshold Z = 0.8. When the consistency coefficient is greater than or equal to this threshold, it means that the initial evaluation value and the secondary evaluation value are close at this time, and the results of the two evaluation methods are not much different. At this time, the initial evaluation value can more accurately reflect the actual situation. Use the consistency coefficient as the weight coefficient of the two evaluation values for weighted fusion to obtain the adjusted evaluation value y n , and its calculation formula is:

[0090]

[0091] Correspondingly, when the consistency coefficient is less than the set threshold Z, it indicates that the difference between the initial evaluation value and the secondary evaluation value is large at this time, and the results of the two evaluation methods differ greatly. The initial evaluation value at this time may be difficult to accurately reflect the actual state of the highway bridge due to problems such as unreasonable sensor layout. Therefore, at this time, the secondary evaluation value needs to be used for auxiliary adjustment: on the one hand, considering that the initial evaluation value reflects the safety status of different aspects of the highway bridge, directly replacing it with the secondary evaluation value is likely to result in the loss of consideration of some factors affecting the safety status. On the other hand, at this time, the secondary evaluation value can better reflect the real safety status. Therefore, directly using the above weighted fusion method will lead to bias in the final result. For this reason, considering that each region and component of the highway bridge are continuous and related, and their consistency also has a certain degree of continuity and relevance, in this embodiment, the current consistency coefficient is used as the weight coefficient of the initial evaluation value, and the maximum consistency coefficient within the set scale before and after the current consistency coefficient (in this embodiment, the set scale is 1, that is, the current consistency coefficient and the two coefficients before and after it) is used as the weight coefficient of the secondary evaluation value, and weighted fusion is performed in this way to obtain the adjusted evaluation value y n , and its calculation formula is:

[0092]

[0093] Through the above method, the adjusted evaluation value sequence y1, y2,..., y i ,..., y n is obtained.

[0094] Step S4: For each adjusted evaluation value, integrate based on the support degree between data to obtain the overall evaluation result of the safety status of the highway bridge. Specifically, based on the adjusted evaluation value sequence y1, y2,..., y i ,..., y n , analyze the support degree between data, and integrate according to this support degree to obtain the overall evaluation value y of the safety status of the highway bridge. Through this method, the problem that the artificially set weight value in the traditional weighted fusion method affects the accuracy of the final result can be effectively avoided.

[0095] Step S4.1: Calculate the support degree r between every two evaluation values based on the support degree function calculation formula ij , and construct a support degree matrix, which can be expressed as:

[0096]

[0097] The values in the above matrix represent the mutual support degree between every two evaluation values. Since it cannot explain the total support degree of a certain value relative to the whole, the weight value of each evaluation value is calculated through the following method to complete the integration calculation.

[0098] Step S4.2: Based on the support degree matrix, calculate the weight coefficient of each evaluation value among all evaluation values, specifically as follows:

[0099] Randomly set a group of non - negative numbers (a1, a2,..., a i ,..., a n ) such that:

[0100] w i = a1r i1 + a2r i2 + … + a n r in , i = 1, 2,..., n;

[0101] Then it can be expressed in matrix form as:

[0102] W = R * A;

[0103] where, W = [w1, w2,..., w i ,..., w n T , A = [a1, a2,..., a i ,..., a n T .

[0104] Since the support degree matrix is a non - negative matrix, there exists a maximum - modulus eigenvalue λ≥0. By solving 1 * A = R * A, the eigenvector A of the maximum - modulus eigenvalue can be obtained, and thus the weight of each data can be solved, which can be expressed as:

[0105]

[0106] Step S4.3: According to the weight coefficient, perform integration calculation to obtain the overall evaluation result of the highway bridge safety status, that is: y = w1y1 + w2y2 + … + w n y n .

[0107] As another implementation manner, based on the overall evaluation result obtained from the above calculation, combined with the preset evaluation levels (different evaluation values correspond to different levels), accurately classify the safety level of the highway bridge, providing data support for subsequent maintenance measures of the highway bridge.

[0108] Example Two

[0109] This example provides a safety status evaluation system for highway bridges, including:

[0110] A data acquisition module, used to collect the time - series vibration data and the overall surface image data at multiple key nodes of the highway bridge;

[0111] ​​An initial evaluation module is configured to input each piece of sequential vibration data into a trained improved generative adversarial network to generate denoised sequential vibration data, and then perform a preliminary state evaluation based on the denoised data to obtain an initial evaluation value. In the improved generative adversarial network, the generator adopts an encoder-decoder structure, and a channel attention mechanism is added between feature encoding and decoding.

[0112] A secondary adjustment module is configured to identify all cracks in the surface image data, divide the cracks into regions, calculate a secondary evaluation value based on the cracks in each region, and use the secondary evaluation value to assist in adjusting the initial evaluation value.

[0113] An overall evaluation module is configured to integrate each adjusted evaluation value based on the support degree between data to obtain an overall evaluation result of the safety state of the highway bridge.

[0114] Embodiment III

[0115] This embodiment provides an electronic device, including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the steps in the above-mentioned safety state evaluation method of the highway bridge are completed.

[0116] Embodiment IV

[0117] This embodiment also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by the processor, the steps in the above-mentioned safety state evaluation method of the highway bridge are completed.

[0118] The steps involved in Embodiments II to IV above correspond to those in Method Embodiment I. For specific implementation manners, reference may be made to the relevant description part of Embodiment I. The term "computer-readable storage medium" should be understood to include a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode, or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0119] Those skilled in the art should understand that the above-mentioned modules or steps of the present invention can be implemented by a general-purpose computer device. Optionally, they can be implemented by program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0120] The above are only the preferred embodiments of the present invention. Although the specific implementation manners of the present invention are described in conjunction with the accompanying drawings, it is not a limitation on the protection scope of the present invention. Those skilled in the art should understand that various modifications or deformations that can be made without creative efforts on the basis of the technical solution of the present invention are still within the protection scope of the present invention.

Claims

1. A safety state assessment method for highway bridges, characterized in that, Including: Collecting time-series vibration data and overall surface image data at multiple key nodes of highway bridges; Inputting each time-series vibration data into the trained improved generative adversarial network to generate denoised time-series vibration data, and then performing a preliminary state assessment based on the denoised data to obtain an initial assessment value; In the improved generative adversarial network, the generator adopts an encoder-decoder structure, and a channel attention mechanism is added between feature encoding and decoding; Identifying all cracks in the surface image data and dividing the cracks into regions, calculating a secondary assessment value according to the cracks in each region, and using the secondary assessment value to assist in adjusting the initial assessment value; For each adjusted assessment value, integrating based on the support degree between data to obtain an overall assessment result of the safety state of the highway bridge.

2. The safety status assessment method for a highway bridge according to claim 1, characterized in that, The improved generative adversarial network includes a generator and a discriminator. Taking the time-series vibration data as a noisy signal, the generator is used to remove the noise component in the noisy signal and establish a mapping relationship between the noisy signal and the denoised signal; the discriminator is used to classify and judge the denoised signal generated by the generator; The generator adopts an encoder-decoder structure, including a feature encoding unit and a feature decoding unit; The feature encoding unit includes multiple encoding layers, and each encoding layer includes a downsampling unit. The input noisy signal undergoes multiple one-dimensional convolutions through multiple downsampling units to extract deep features; The feature decoding unit includes multiple decoding layers, and each decoding layer includes an upsampling unit. The extracted deep features are gradually restored into a feature vector with the same scale as the input signal through multiple one-dimensional transposed convolutions by multiple upsampling units; among them, the encoded feature map output by each encoding layer and the decoded feature map output by the corresponding decoding layer are jump-stitched through the channel attention mechanism.

3. The safety status assessment method of a highway bridge according to claim 2, characterized in that Adding a channel attention mechanism between the decoded feature map and the encoded feature map, including: The encoded feature map x output by the encoding layer e and the decoded feature map x output by the decoding layer d are respectively convolved through the same convolutional kernel to obtain two weight vectors W e and W d ; Add the weight vectors W e and W d and then activate the result using the LeakyRelu activation function to obtain the activated result W a1 ; Activation result W a1 After convolution and activation, the attention weight layer W is obtained a2 , and this attention weight layer W a2 After resampling, it is multiplied by the decoded feature map x output by the current decoding layer d to complete the attention activation of the decoded feature map x d ​ 4. A safety status evaluation method for a highway bridge according to claim 1, characterized in that, Performing a preliminary state assessment based on the denoised data to obtain an initial assessment value, including: Adopting the Gram angular field method to convert the denoised time-series vibration data into a two-dimensional grayscale image, and inputting the two-dimensional grayscale image into the state assessment model based on the BP neural network to output the initial assessment value.

5. A safety state assessment method for a highway bridge as described in claim 1, characterized in that, Identifying all cracks in the surface image data and dividing the cracks into regions, calculating a secondary assessment value according to the cracks in each region, including: Inputting the surface image of the overall outer surface of the highway bridge into the target detection model, identifying all cracks in the image and extracting each crack target, and obtaining the length and width of each crack target by calculating the number of pixel points of the crack target; Dividing all the identified cracks into regions; For all the cracks in each region, calculating the secondary assessment value of each region of the highway bridge according to the crack length and width.

6. The safety status assessment method of a highway bridge according to claim 1, characterized in that, Based on the data consistency analysis method, using the secondary assessment value to assist in adjusting the initial assessment value, including: Performing normalization processing on the initial assessment value sequence and the secondary assessment value sequence of multiple regions respectively; Based on the initial evaluation value sequence and the secondary evaluation value sequence after normalization, data consistency analysis is performed for each pair of evaluation values, the weight coefficient of each pair of evaluation values is calculated, and through weighted fusion, an adjusted evaluation value sequence is obtained; Among them, for each pair of evaluation values, if the calculated consistency coefficient is greater than or equal to the set threshold, the consistency coefficient is used as the weight coefficient of the two evaluation values; if the calculated consistency coefficient is less than the set threshold, the current consistency coefficient is used as the weight coefficient of the initial evaluation value, and the maximum consistency coefficient within the set scale before and after the current consistency coefficient is used as the weight coefficient of the secondary evaluation value.

7. The safety state assessment method for a highway bridge according to claim 1, characterized in that, For each adjusted evaluation value, integration is performed based on the support degree between data to obtain an overall evaluation result of the safety state of the highway bridge, including: Based on the support degree function calculation formula, the support degree between two evaluation values is calculated to construct a support degree matrix; Based on the support degree matrix, the weight coefficient of each evaluation value among all evaluation values is calculated; According to the weight coefficient, integration calculation is performed to obtain an overall evaluation result of the safety state of the highway bridge.

8. A safety state assessment system for highway bridges, characterized in that, Including: A data acquisition module for acquiring time-series vibration data at multiple key nodes of the highway bridge and overall surface image data; An initial evaluation module for inputting each time-series vibration data into a trained improved generative adversarial network to generate denoised time-series vibration data, and then performing a preliminary state evaluation based on the denoised data to obtain an initial evaluation value; In the improved generative adversarial network, the generator adopts an encoder-decoder structure, and a channel attention mechanism is added between feature encoding and decoding; A secondary adjustment module for identifying all cracks in the surface image data and dividing the cracks into regions, calculating secondary evaluation values according to the cracks in each region, and using the secondary evaluation values to assist in adjusting the initial evaluation values; An overall evaluation module for, for each adjusted evaluation value, performing integration based on the support degree between data to obtain an overall evaluation result of the safety state of the highway bridge.

9. An electronic device, characterized in that, Including a memory, a processor, and computer instructions stored on the memory and running on the processor. When the computer instructions are run by the processor, the steps of a method for evaluating the safety state of a highway bridge as described in any one of claims 1-7 are completed.

10. A computer-readable storage medium, characterized in that, For storing computer instructions, when the computer instructions are executed by the processor, the steps of a method for evaluating the safety state of a highway bridge as described in any one of claims 1-7 are completed.

Images