Intelligent bridge guardrail state monitoring method and system

Abstract

The invention discloses an intelligent bridge guardrail state monitoring method and system, and the method comprises the steps: S1, intercepting a plurality of images of a to-be-monitored bridge guardrail as original images, carrying out the appearance detection of the original images, and obtaining the first information of the guardrail; s2, determining the structure of the bridge guardrail based on the original image, and obtaining second information of the guardrail; s3, acquiring environment information of the to-be-monitored bridge as third information of the guardrail, and performing finite element analysis of the bridge guardrail based on the second information, the second information and the third information to obtain an analysis result; and S4, carrying out life detection and safety analysis based on the analysis result, and completing intelligent state monitoring of the bridge guardrail.

Description

Technical Field

[0001] The present invention relates to the technical field of guardrail monitoring, and in particular to an intelligent bridge guardrail status monitoring method and system. Background Art

[0002] With the rapid development of infrastructure construction in my country, guardrails are installed on many bridges, especially some cross-sea bridges, highway bridges, railway bridges, etc. The installed guardrails play a key role in reducing the losses caused by road safety accidents.

[0003] However, obtaining the status of the guardrail and determining the availability of the bridge guardrail is currently done manually, which is time-consuming and laborious. Especially for long bridge guardrails, maintenance and inspection take a long time, which is easy to cause property losses and casualties. Summary of the invention

[0004] In view of the above problems, the present invention provides a method for monitoring the state of an intelligent bridge guardrail, the method comprising:

[0005] Step S1, intercepting a plurality of images of the guardrail of the bridge to be monitored as original images, detecting the appearance of the original images, and obtaining first information of the guardrail;

[0006] Step S2: determining the structure of the bridge guardrail based on the original image, and obtaining second information of the guardrail;

[0007] Step S3, obtaining environmental information of the bridge to be monitored as the third information of the guardrail, performing finite element analysis of the bridge guardrail based on the second information, the second information and the third information, and obtaining an analysis result;

[0008] Step S4: Perform life detection and safety analysis based on the analysis results to complete intelligent status monitoring of the bridge guardrail.

[0009] Optionally, in step S1, a plurality of images of the guardrail of the bridge to be monitored are intercepted as original images, and the process of detecting the appearance of the original images specifically includes:

[0010] Preprocessing the original image to obtain a preprocessed image;

[0011] Adopting the SPP structure to expand the feature information of the preprocessed image;

[0012] Use the channel attention mechanism to extract image features of the preprocessed image after expanding feature information;

[0013] The appearance detection of the original image is performed based on the image features.

[0014] Optionally, in the process of using the SPP structure to expand the feature information of the preprocessed image, the MaxPool in the SPP structure is replaced by the SoftPool, and the pixel values ​​in the image area R are weighted and summed in proportion to obtain a pooling result expressed as:

[0015]

[0016] in,

[0017] Optionally, in step S3, the process of performing finite element analysis of the bridge guardrail based on the second information, the second information and the third information to obtain the analysis result specifically includes:

[0018] The first information and the second information are fused and then data enhancement is performed, and the Kriging interpolation enhancement algorithm is used on the fused information after data enhancement to obtain a grid space lattice;

[0019] Generate bridge guardrail simulation model based on grid space lattice;

[0020] generating an environment model based on the third information;

[0021] Finite element analysis is performed based on the bridge guardrail simulation model and the environmental module to obtain analysis results.

[0022] The present invention also discloses an intelligent bridge guardrail state monitoring system, the system comprising: an appearance detection module, a basic information acquisition module, a finite element analysis module and a state monitoring module;

[0023] The appearance detection module is used to intercept several images of the guardrail of the bridge to be monitored as original images, perform appearance detection on the original images, and obtain first information of the guardrail;

[0024] The basic information acquisition module is used to determine the structure of the bridge guardrail based on the original image and acquire the second information of the guardrail;

[0025] The finite element analysis module is used to obtain environmental information of the bridge to be monitored as the third information of the guardrail, and perform finite element analysis of the bridge guardrail based on the second information, the second information and the third information to obtain an analysis result;

[0026] The condition monitoring module is used to perform life detection and safety analysis based on the analysis results to complete intelligent condition monitoring of the bridge guardrail.

[0027] Optionally, the workflow of the appearance detection module specifically includes:

[0028] Preprocessing the original image to obtain a preprocessed image;

[0029] Adopting the SPP structure to expand the feature information of the preprocessed image;

[0030] Use the channel attention mechanism to extract image features of the preprocessed image after expanding feature information;

[0031] The appearance detection of the original image is performed based on the image features.

[0032] Optionally, in the process of using the SPP structure to expand the feature information of the preprocessed image, the MaxPool in the SPP structure is replaced by the SoftPool, and the pixel values ​​in the image area R are weighted and summed in proportion to obtain a pooling result expressed as:

[0033]

[0034] in,

[0035] Optionally, the workflow of the finite element analysis module specifically includes:

[0036] The first information and the second information are fused and then data enhancement is performed, and the Kriging interpolation enhancement algorithm is used on the fused information after data enhancement to obtain a grid space lattice;

[0037] Generate bridge guardrail simulation model based on grid space lattice;

[0038] generating an environment model based on the third information;

[0039] Finite element analysis is performed based on the bridge guardrail simulation model and the environmental module to obtain analysis results.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] By processing and analyzing the images, the structural features of the guardrail can be accurately identified, providing basic data for subsequent finite element analysis. Using computer vision technology, the structural type and key components of the guardrail can be automatically identified, reducing errors in manual operation. Through finite element analysis, the stress distribution and potential damage of the guardrail under different environmental conditions can be predicted, and corresponding measures can be taken based on the damage to extend the service life of the guardrail. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0043] Figure 1 It is a flow chart of a method for monitoring the status of an intelligent bridge guardrail according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] Embodiment 1

[0046] The present invention provides a method for monitoring the state of an intelligent bridge guardrail. Figure 1 As shown, the method includes:

[0047] Step S1, intercepting several images of the guardrail of the bridge to be monitored as original images, detecting the appearance of the original images, and obtaining the first information of the guardrail, specifically includes:

[0048] The original image is preprocessed to obtain a preprocessed image; the SPP structure is used to expand the feature information of the preprocessed image.

[0049] Preprocessing mainly includes denoising and filtering. Wavelet denoising and mean filtering can be used to obtain a preprocessed image.

[0050] The present invention is based on the YOLOX algorithm. In the YOLOX detection process, the feature extraction part uses the SPP structure. However, the current SPP structure has less pixel information and the feature volume to be extracted is small, which makes it easy to miss the detection. Therefore, the present invention replaces the MaxPool in SPP with SoftPool pooling to extract features in YOLOX.

[0051] In the process of using the SPP structure to expand the feature information of the preprocessed image, SoftPool is used to replace the MaxPool in the SPP structure, and the pixel values ​​in the image area R are weighted and summed proportionally to obtain the pooling result expressed as:

[0052]

[0053] in,

[0054] Use the channel attention mechanism to extract the image features of the preprocessed image after expanding the feature information; complete the appearance detection of the original image based on the image features. Use efficient channel attention to assign weight values ​​to feature channels through compression and excitation operations. Use FC to reduce the dimension and extract channel attention information.

[0055] At the end of the YOLOX algorithm, the loss function is used to optimize feature extraction, and finally the GIoU loss function is used for optimization to achieve feature positioning.

[0056]

[0057] Among them, C represents the minimum external bounding box covering the predicted box and the real box. However, when the predicted box and the real box have an inclusion relationship, GIoU will degenerate into IoU, B gt and B represent the true box and predicted box.

[0058] Step S2: Based on the original image, determine the structure of the bridge guardrail and obtain the second information of the guardrail. Because the structure of the guardrail is simple and the structures of almost all guardrails are the same or similar, the structure of the guardrail can be obtained mainly by cutting out the edges of the guardrail in the original image and connecting the edges.

[0059] In addition to obtaining the structure of the guardrail, it is also necessary to obtain the material of the guardrail to facilitate the calculation of the weight of the guardrail. The degree of embedment of the guardrail also needs to be recorded.

[0060] Step S3, obtaining environmental information of the bridge to be monitored as the third information of the guardrail, performing finite element analysis of the bridge guardrail based on the second information, the second information and the third information, and obtaining analysis results, specifically including:

[0061] After fusing the first information and the second information, data enhancement is performed, and the Kriging interpolation enhancement algorithm is used on the fused information after data enhancement to obtain a grid space lattice; a bridge guardrail simulation model is generated based on the grid space lattice; an environmental model is generated based on the third information; and a finite element analysis is performed based on the bridge guardrail simulation model and the environmental module to obtain an analysis result.

[0062] The third information specifically includes: the area where the guardrail to be tested is located, the basic conditions of the area (natural factors such as temperature, humidity, wind force, earthquake conditions, etc.), and cultural factors such as the flow of people and vehicles on the bridge. In the process of building the environmental model, it is necessary to first shoot the environment of the guardrail of the bridge to be tested, obtain the image, extract features from the image, and use normalization to process it before feature extraction. After the feature extraction is completed, the minimum redundancy maximum correlation algorithm is used in the training set to remove redundant and irrelevant features. For human factors, it is necessary to conduct periodic monitoring, use sensors and visual detection devices to calculate the flow of people and vehicles, and finally build an environmental simulation model based on the environmental characteristics and cultural data.

[0063] Step S4: Perform life detection and safety analysis based on the analysis results to complete intelligent status monitoring of the bridge guardrail.

[0064] The life detection process includes: training the traditional SVM classifier based on the finite element analysis data, obtaining the support vector in the initial stage, and constructing the decision classification surface; calculating the center vector corresponding to the sample set; based on the center vector, fuzzy membership function; constructing a fuzzy SVM classifier based on the fuzzy membership function, inputting the real-time finite element data of the guardrail to be detected into the SVM classifier, and obtaining the life prediction result. The membership function is:

[0065]

[0066] Where l is the radius, For class information.

[0067] Safety analysis: The DEMATEL method and the fuzzy analytic hierarchy process are used to calculate the relevance weight and importance weight of the common performance conditions respectively. Then, the combined weighting method is used to linearly combine the two weights to obtain the comprehensive weight, and the safety is calculated based on the weight.

[0068] Embodiment 2

[0069] An intelligent bridge guardrail state monitoring system, the system comprising: an appearance detection module, a basic information acquisition module, a finite element analysis module and a state monitoring module;

[0070] The appearance detection module is used to intercept several images of the guardrail of the bridge to be monitored as original images, detect the appearance of the original images, and obtain the first information of the guardrail. Specifically, it includes: preprocessing the original images to obtain preprocessed images; using the SPP structure to expand the feature information of the preprocessed images.

[0071] The present invention is based on the YOLOX algorithm. In the YOLOX detection process, the feature extraction part uses the SPP structure. However, the current SPP structure has less pixel information and the feature volume to be extracted is small, which makes it easy to miss the detection. Therefore, the present invention replaces the MaxPool in SPP with SoftPool pooling to extract features in YOLOX.

[0072] In the process of using the SPP structure to expand the feature information of the preprocessed image, SoftPool is used to replace the MaxPool in the SPP structure, and the pixel values ​​in the image area R are weighted and summed proportionally to obtain the pooling result expressed as:

[0073]

[0074] in,

[0075] Use the channel attention mechanism to extract the image features of the preprocessed image after expanding the feature information; complete the appearance detection of the original image based on the image features. Use efficient channel attention to assign weight values ​​to feature channels through compression and excitation operations. Use FC to reduce the dimension and extract channel attention information.

[0076] At the end of the YOLOX algorithm, the loss function is used to optimize feature extraction, and finally the GIoU loss function is used for optimization to achieve feature positioning.

[0077]

[0078] Among them, C represents the minimum external bounding box covering the predicted box and the real box. However, when the predicted box and the real box have an inclusion relationship, GIoU will degenerate into IoU, B gt and B represent the true box and prediction.

[0079] The basic information acquisition module is used to determine the structure of the bridge guardrail based on the original image and obtain the second information of the guardrail.

[0080] In addition to obtaining the structure of the guardrail, it is also necessary to obtain the material of the guardrail to facilitate the calculation of the weight of the guardrail. The degree of embedding of the guardrail also needs to be recorded.

[0081] The finite element analysis module is used to obtain the environmental information of the bridge to be monitored as the third information of the guardrail, and perform finite element analysis of the bridge guardrail based on the second information, the second information and the third information to obtain the analysis results. Specifically, it includes: fusing the first information and the second information to perform data enhancement, using the Kriging interpolation enhancement algorithm on the fused information after data enhancement to obtain a grid space lattice; generating a bridge guardrail simulation model based on the grid space lattice; generating an environmental model based on the third information; performing finite element analysis based on the bridge guardrail simulation model and the environmental module to obtain the analysis results.

[0082] The third information specifically includes: the area where the guardrail to be tested is located, the basic conditions of the area (natural factors such as temperature, humidity, wind, earthquake conditions, etc.), and human factors such as the flow of people and vehicles on the bridge. In the process of building the environmental model, it is necessary to first shoot the environment of the guardrail of the bridge to be tested, obtain the image, extract features from the image, and use normalization to process it before feature extraction. After the feature extraction is completed, the minimum redundancy maximum correlation algorithm is used in the training set to remove redundant and irrelevant features. For human factors, it is necessary to conduct periodic monitoring, use sensors and visual detection devices to calculate the flow of people and vehicles, and finally build an environmental simulation model based on the environmental characteristics and human data.

[0083] The condition monitoring module is used to perform life detection and safety analysis based on the analysis results to complete intelligent condition monitoring of the bridge guardrail.

[0084] The life detection process includes: training the traditional SVM classifier based on the finite element analysis data, obtaining the support vector in the initial stage, and constructing the decision classification surface; calculating the center vector corresponding to the sample set; based on the center vector, fuzzy membership function; constructing a fuzzy SVM classifier based on the fuzzy membership function, inputting the real-time finite element data of the guardrail to be detected into the SVM classifier, and obtaining the life prediction result. The membership function is:

[0085]

[0086] Where l is the radius, For class information.

[0087] Safety analysis: The DEMATEL method and the fuzzy analytic hierarchy process are used to calculate the relevance weight and importance weight of the common performance conditions respectively. Then, the combined weighting method is used to linearly combine the two weights to obtain the comprehensive weight, and the safety is calculated based on the weight.

[0088] The embodiments described above are only descriptions of the preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for monitoring the state of an intelligent bridge guardrail, characterized in that: The method comprises: Step S1, intercepting a plurality of images of the guardrail of the bridge to be monitored as original images, detecting the appearance of the original images, and obtaining first information of the guardrail; Step S2: determining the structure of the bridge guardrail based on the original image, and obtaining second information of the guardrail; Step S3, obtaining environmental information of the bridge to be monitored as the third information of the guardrail, performing finite element analysis of the bridge guardrail based on the second information, the second information and the third information, and obtaining an analysis result; Step S4: Perform life detection and safety analysis based on the analysis results to complete intelligent status monitoring of the bridge guardrail.

2. The intelligent bridge guardrail status monitoring method according to claim 1 is characterized in that: In step S1, a plurality of images of the bridge guardrail to be monitored are intercepted as original images, and the process of detecting the appearance of the original images specifically includes: Preprocessing the original image to obtain a preprocessed image; Adopting the SPP structure to expand the feature information of the preprocessed image; Use the channel attention mechanism to extract image features of the preprocessed image after expanding feature information; The appearance detection of the original image is performed based on the image features.

3. The intelligent bridge guardrail status monitoring method according to claim 2 is characterized in that: In the process of using the SPP structure to expand the feature information of the preprocessed image, SoftPool is used to replace the MaxPool in the SPP structure, and the pixel values ​​in the image area R are weighted and summed proportionally to obtain the pooling result expressed as: in, 4. The intelligent bridge guardrail status monitoring method according to claim 1 is characterized in that: In step S3, the process of performing finite element analysis on the bridge guardrail based on the second information, the second information and the third information to obtain the analysis result specifically includes: The first information and the second information are fused and then data enhancement is performed, and the Kriging interpolation enhancement algorithm is used on the fused information after data enhancement to obtain a grid space lattice; Generate bridge guardrail simulation model based on grid space lattice; generating an environment model based on the third information; Finite element analysis is performed based on the bridge guardrail simulation model and the environmental module to obtain analysis results.

5. An intelligent bridge guardrail status monitoring system, the system is used to implement the intelligent bridge guardrail status monitoring method according to any one of claims 1 to 4, characterized in that: The system includes: an appearance detection module, a basic information acquisition module, a finite element analysis module and a state monitoring module; The appearance detection module is used to intercept several images of the guardrail of the bridge to be monitored as original images, perform appearance detection on the original images, and obtain first information of the guardrail; The basic information acquisition module is used to determine the structure of the bridge guardrail based on the original image and acquire the second information of the guardrail; The finite element analysis module is used to obtain environmental information of the bridge to be monitored as the third information of the guardrail, and perform finite element analysis of the bridge guardrail based on the second information, the second information and the third information to obtain an analysis result; The condition monitoring module is used to perform life detection and safety analysis based on the analysis results to complete intelligent condition monitoring of the bridge guardrail.

6. The intelligent bridge guardrail status monitoring system according to claim 5 is characterized in that: The workflow of the appearance detection module specifically includes: Preprocessing the original image to obtain a preprocessed image; Adopting the SPP structure to expand the feature information of the preprocessed image; Use the channel attention mechanism to extract image features of the preprocessed image after expanding feature information; The appearance detection of the original image is performed based on the image features.

7. The intelligent bridge guardrail status monitoring system according to claim 6 is characterized in that: In the process of using the SPP structure to expand the feature information of the preprocessed image, SoftPool is used to replace the MaxPool in the SPP structure, and the pixel values ​​in the image area R are weighted and summed proportionally to obtain the pooling result expressed as: in, 8. The intelligent bridge guardrail status monitoring system according to claim 5 is characterized in that: The workflow of the finite element analysis module specifically includes: The first information and the second information are fused and then data enhancement is performed, and the Kriging interpolation enhancement algorithm is used on the fused information after data enhancement to obtain a grid space lattice; Generate bridge guardrail simulation model based on grid space lattice; generating an environment model based on the third information; Finite element analysis is performed based on the bridge guardrail simulation model and the environmental module to obtain analysis results.

Images