Safety protection method and system for super-large bridge crossing separated overpass

Abstract

This invention discloses a safety protection method and system for overpass grade-separated interchanges of extra-large bridges, relating to the field of safety protection technology for extra-large bridges. The method includes: acquiring bridge features of a first target bridge; acquiring a first matching bridge; sending a first request to the safety accident information management platform terminal corresponding to the first matching bridge; storing a safety monitoring model after the permission request is granted, and introducing a first loss function to perform loss analysis on the bridge features, outputting loss data; acquiring a safety monitoring model after transfer learning when the loss data region converges; and using the safety monitoring model after transfer learning to perform safety monitoring on the first target bridge. This invention solves the technical problems of long bridge protection monitoring model construction cycles and low safety protection quality in existing technologies, achieving the technical effect of improving safety protection quality and monitoring accuracy.

Description

Technical Field

[0001] This invention relates to the field of safety protection technology for mega-bridges, specifically to a safety protection method and system for mega-bridges with separated overpasses. Background Technology

[0002] With rapid economic development, the demands for traffic quality are increasing. In particular, the safety protection of large bridges spanning grade-separated overpasses is of paramount importance for ensuring traffic safety and preventing accidents. Numerous factors influence bridge safety to varying degrees, requiring the analysis of vast amounts of data for protection analysis. Currently, intelligent neural networks are used for data analysis; however, the construction process for these networks is too lengthy, requiring extensive training data and multiple tests, which cannot meet current needs. Existing technologies suffer from the technical problems of long construction cycles and low safety protection quality in bridge protection monitoring models. Summary of the Invention

[0003] This application provides a safety protection method and system for super-large bridges with separated overpasses, which is used to address the technical problems of long construction cycle of bridge protection monitoring models and low safety protection quality in the prior art.

[0004] In view of the above problems, this application provides a safety protection method and system for super-large bridges with separated overpasses.

[0005] The first aspect of this application provides a safety protection method for a grade-separated overpass on a large-scale bridge, the method comprising:

[0006] Obtain the bridge features of the first target bridge, wherein the bridge features include bridge architectural features, geographical features, and bridge environmental features;

[0007] Connect to the bridge protection IoT management platform, search for similar bridges based on the bridge characteristics, and obtain the first matching bridge;

[0008] Send a first request message to the terminal of the safety accident information management platform corresponding to the first matched bridge to obtain permission to call the safety monitoring model corresponding to the first matched bridge;

[0009] Once the permission request is approved, the safety monitoring model is stored, and a first loss function is introduced to perform loss analysis on the bridge features, outputting loss data, which includes building loss data, regional loss data, and environmental loss data.

[0010] The security monitoring model is transferred to the loss data, and when the loss data region converges, the security monitoring model after transfer learning is obtained.

[0011] The safety monitoring model after transfer learning is used to monitor the safety of the first target bridge.

[0012] A second aspect of this application provides a safety protection system for a grade-separated overpass on a large-scale bridge, the system comprising:

[0013] A bridge feature acquisition module is used to acquire the bridge features of a first target bridge, wherein the bridge features include bridge architectural features, geographical features, and bridge environmental features.

[0014] The first matching bridge acquisition module is used to connect to the bridge protection IoT management platform, and search for similar bridges based on the bridge characteristics to acquire the first matching bridge.

[0015] The request information acquisition module is used to send a first request information to the safety accident information management platform terminal corresponding to the first matched bridge, in order to obtain the permission to call the safety monitoring model corresponding to the first matched bridge.

[0016] The loss data output module is used to store the safety monitoring model after the permission request is approved, and to introduce a first loss function to perform loss analysis on the bridge features and output loss data, wherein the loss data includes building loss data, regional loss data and environmental loss data.

[0017] A security monitoring model acquisition module is used to perform transfer learning on the security monitoring model based on the loss data, and to acquire the security monitoring model after transfer learning when the loss data region converges.

[0018] A safety monitoring module is used to perform safety monitoring on the first target bridge using a safety monitoring model derived from transfer learning.

[0019] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0020] This application acquires the bridge features of a first target bridge, including bridge architectural features, geographical features, and environmental features. It then connects to a bridge protection IoT management platform to search for similar bridges based on these features, identifying a first matching bridge. A first request is then sent to the safety accident information management platform terminal corresponding to the first matching bridge to obtain permission to access the safety monitoring model associated with that bridge. Once the permission request is granted, the safety monitoring model is stored, and a first loss function is used to analyze the bridge features, outputting loss data. This loss data includes architectural loss data, geographical loss data, and environmental loss data. The safety monitoring model is then transferred and learned based on this loss data. When the loss data converges, the transferred-learned safety monitoring model is obtained, and this model is used to monitor the first target bridge. This achieves the technical effect of improving the quality of safety protection and enhancing the accuracy of safety monitoring. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A schematic diagram of a safety protection method for an overpass grade-separated bridge of an extra-large bridge, provided in an embodiment of this application;

[0023] Figure 2 A flowchart illustrating the output of the first matching bridge in a safety protection method for an overpass grade-separated bridge provided in this application embodiment;

[0024] Figure 3 This is a flowchart illustrating the adjustment of the corresponding early warning configuration parameters in the safety monitoring model called in a safety protection method for an overpass grade-separated bridge provided in this application embodiment.

[0025] Figure 4 This is a schematic diagram of a safety protection system for a grade-separated overpass on a super-large bridge, provided as an embodiment of this application.

[0026] Explanation of reference numerals in the attached figures: Bridge feature acquisition module 11, First matching bridge acquisition module 12, Request information acquisition module 13, Loss data output module 14, Safety monitoring model acquisition module 15, Safety monitoring module 16. Detailed Implementation

[0027] This application provides a safety protection method and system for super-large bridges with separated overpasses, which addresses the technical problems of long construction cycles and low safety protection quality in existing technologies.

[0028] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0029] It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to such process, method, product, or device.

[0030] Example 1

[0031] like Figure 1 As shown, this application provides a safety protection method for a grade-separated overpass of a super-large bridge, wherein the method includes:

[0032] Step S100: Obtain the bridge features of the first target bridge, wherein the bridge features include bridge architectural features, geographical features, and bridge environmental features.

[0033] In one embodiment of this application, the first target bridge is any bridge requiring safety protection. By acquiring the bridge characteristics of the first target bridge, a basis is provided for subsequent safety protection analysis based on these characteristics. The bridge characteristics include bridge architectural features, geographical location features, and environmental features. Specifically, the bridge architectural features describe the technological features used in the bridge construction design, including total weld length, steel consumption, hoisting weight, structural load, and span. The geographical features describe the geographical location of the first target bridge, including geographical location, water distribution, and climate characteristics. The environmental features describe the application environment of the first target bridge, including traffic flow, humidity, temperature, traffic load, wind speed, and wind direction. By obtaining the bridge characteristics of the first target bridge, a basis is provided for subsequent analysis of hazardous factors affecting the operational safety of the overpass, thereby constructing a targeted protective structure and improving the safety and comprehensiveness of the protection.

[0034] Step S200: Connect to the bridge protection IoT management platform, search for similar bridges based on the bridge characteristics, and obtain the first matching bridge;

[0035] Furthermore, such as Figure 2 As shown, step S200 in this embodiment further includes:

[0036] Step S210: Build a bridge matching model, wherein the bridge matching model is obtained through training with multiple sets of data, including bridge architectural features, set regional features, and bridge environmental features, as well as identification information that identifies the matching degree of the bridge architectural features, set regional features, and bridge environmental features;

[0037] Step S220: According to the bridge matching model, input the bridge features of the first target bridge into the bridge matching model, and output the building matching degree, regional matching degree and environmental matching degree according to the bridge matching model;

[0038] Step S230: Calculate the bridge matching degree based on the building matching degree, regional matching degree, and environmental matching degree, and output the first matching bridge;

[0039] The first matched bridge is the bridge with the highest matching degree among all bridges included in the bridge protection IoT management platform.

[0040] In one possible embodiment, the bridge protection IoT management platform is an intelligent platform for managing bridges, possessing functions such as protection monitoring and management for various types of bridges. By using bridge features as an index, the platform searches for similar bridges to obtain the first matching bridge. The first matching bridge is the bridge with the highest matching score among all bridges included in the bridge protection IoT management platform. By searching for similar bridges, the efficiency of safety protection construction can be improved, and analysis time can be shortened.

[0041] Specifically, the bridge matching model is an intelligent model embedded in the bridge protection IoT management platform that performs similar matching on the first target bridge. Preferably, the bridge matching model is acquired through training on multiple sets of data, wherein the multiple sets of data include bridge architectural features, regional features, bridge environmental features, and identification information of the matching degree of the regional features and bridge environmental features. The framework based on a BP neural network is trained using the multiple sets of data, and the training process is supervised using the identification information of the matching degree of the bridge architectural features, regional features, and bridge environmental features until the model output converges, thus obtaining the trained bridge matching model. Then, the bridge features of the first target bridge are input into the bridge matching model, and the model outputs architectural matching degree, regional matching degree, and environmental matching degree.

[0042] Specifically, the bridge matching degree is calculated by weighting the building matching degree, regional matching degree, and environmental matching degree according to weights preset by those skilled in the art, and multiple bridge matching degrees are obtained based on the calculation results. The bridge with the highest matching degree among all bridges included in the bridge protection IoT management platform is selected as the first matching bridge from these multiple bridge matching degrees.

[0043] Step S300: Send a first request message to the terminal of the safety accident information management platform corresponding to the first matched bridge to obtain the permission to call the safety monitoring model corresponding to the first matched bridge;

[0044] Specifically, after obtaining the first matching bridge, a first request is sent to the safety incident information management platform terminal corresponding to the first matching bridge. This first request is used to obtain permission to invoke the safety monitoring model corresponding to the first matching bridge. This obtains the safety monitoring model corresponding to the first matching bridge with the highest matching degree to the first target bridge, laying the groundwork for subsequent adjustments to obtain a safety monitoring model adapted to the first target bridge.

[0045] Step S400: After the permission request is approved, the security monitoring model is stored, and a first loss function is introduced to perform loss analysis on the bridge features, and loss data is output, wherein the loss data includes building loss data, regional loss data and environmental loss data;

[0046] Furthermore, step S400 in this embodiment of the application also includes:

[0047] Step S410: Determine whether the first matching degree is greater than or equal to the first preset matching degree. If the first matching degree is greater than or equal to the first preset matching degree, obtain the model calling instruction, wherein the model calling instruction is used to call the security monitoring model corresponding to the first matching bridge.

[0048] The security monitoring model corresponding to the first matching bridge includes a security model architecture, a model training dataset, and model convergence information.

[0049] Step S420: By extracting features from the model training dataset, output bridge features based on the first matched bridge, including bridge architectural features, regional features, and bridge environmental features.

[0050] Step S430: Introduce the first loss function to perform loss analysis on the bridge features of the first target bridge and the first matched bridge, and output loss data.

[0051] Furthermore, in determining whether the first matching degree is greater than a preset matching degree, step S410 of this application embodiment further includes:

[0052] Step S411: If the first matching degree is less than the first preset matching degree, the building matching degree, the region matching degree and the environment matching degree are judged respectively, and feature types that are greater than or equal to the second preset matching degree are obtained, wherein the second preset matching degree is less than the first matching degree;

[0053] Step S412: Label and classify features with a matching degree less than the second preset degree, and output the recognition results, which include similar data and dissimilar data;

[0054] Step S413: Use the feature types with a matching degree greater than or equal to the second preset degree as the first source data, and use the similar data in the recognition results as the second source data for transfer learning.

[0055] In one possible embodiment, after the permission request is granted, the safety monitoring model of the first matched bridge is stored, and a first loss function is introduced to perform loss analysis on the bridge features in the safety monitoring model to obtain loss data. This loss data reflects the degree of deviation loss between the first target bridge and the first matched bridge, including architectural loss data, geographical loss data, and environmental loss data. This provides reliable data for subsequently determining the extent of transfer learning on the safety monitoring model.

[0056] In one possible embodiment, by determining whether the first matching degree is greater than or equal to the first preset matching degree, if the first matching degree is greater than or equal to the first preset matching degree, it indicates that the similarity between the first target bridge and the first matching bridge is high, and a model invocation instruction can be obtained. The model invocation instruction is used to invoke the safety monitoring model corresponding to the first matching bridge. Furthermore, the model invocation instruction is used to invoke the safety monitoring model corresponding to the first matching bridge.

[0057] Specifically, by extracting features from the model training dataset, bridge features based on the first matched bridge are output, including bridge architectural features, geographical features, and environmental features. Then, a first loss function is used to perform loss analysis on the bridge features of the first target bridge and the first matched bridge, outputting loss data. Preferably, the first loss function is: Where L(Y / f(x)) represents the loss data, n is the number of bridge features, and Y... i Let f(x) be the i-th bridge feature value of the first target bridge. i ) represents the i-th bridge feature value of the first matched bridge, where i is an integer greater than or equal to 1.

[0058] Specifically, if the first matching degree is less than the first preset matching degree, the building matching degree, the regional matching degree, and the environmental matching degree are judged respectively to obtain feature types with a matching degree greater than or equal to a second preset matching degree, wherein the second preset matching degree is less than the first matching degree. The second preset matching degree is the lowest matching degree that the features of the first matching bridge and the features of the first target bridge do not match well, but can be used for reference, and is set by those skilled in the art. Feature types with matching degrees less than the second preset matching degree are labeled and classified, and recognition results are output. The recognition results include similar data and dissimilar data. The similar data are data with a high degree of similarity between the features of the first matching bridge and the first target bridge. By labeling and classifying the feature types with matching degrees less than the second preset matching degree that are dissimilar to the first target bridge, they are identified as dissimilar data. Thus, feature types with matching degrees greater than or equal to the second preset matching degree are used as the first source data, and the similar data in the recognition results are used as the second source data for transfer learning. The transfer learning refers to the process of relearning and developing a safety monitoring model for the first target bridge based on the safety monitoring model of the first matching bridge.

[0059] Step S500: Perform transfer learning on the security monitoring model based on the loss data. When the loss data region converges, obtain the security monitoring model after transfer learning.

[0060] Furthermore, based on the loss data, the security monitoring model is transferred to learn. In this embodiment, step S500 further includes:

[0061] Step S510: Use the bridge features of the first matched bridge as source data and the bridge features of the first target bridge as target data to output loss data, and perform transfer learning based on the loss data to build a transfer learning channel.

[0062] The transfer learning channel includes a building feature learning branch, a regional feature learning branch, and an environmental feature learning branch, and the weight network layer embedded in the transfer learning channel is obtained by the size of the loss data in each feature learning branch.

[0063] In the embodiments of this application, the process of transferring learning the safety monitoring model using the loss data is supervised. When the loss data region converges, that is, when the deviation between the bridge features of the first target bridge and the bridge features of the first matching bridge decreases, the safety monitoring model after transfer learning is obtained.

[0064] Specifically, by using the bridge features of the first matched bridge as source data and the bridge features of the first target bridge as target data, loss data is obtained. Transfer learning is then performed based on this loss data to build a transfer learning channel. This transfer learning channel includes building feature learning branches, regional feature learning branches, and environmental feature learning branches. The weighted network layer embedded in the transfer learning channel is obtained based on the magnitude of the loss data in each feature learning branch. Preferably, the weight values ​​of the building feature learning branch, regional feature learning branch, and environmental feature learning branch in the weighted network layer are allocated according to the magnitude of the loss data, using the ratio of the loss data of each branch to the total loss data of the three learning branches as the weight value.

[0065] Step S600: Perform safety monitoring on the first target bridge using the safety monitoring model after transfer learning.

[0066] Furthermore, such as Figure 3 As shown, step S600 in this embodiment further includes:

[0067] Step S610: Obtain the bridge life index of the first target bridge and the bridge life index of the first matching bridge.

[0068] Step S620: When the bridge life index of the first matched bridge is less than the bridge life index of the first target bridge, a first hidden danger coefficient is generated.

[0069] Step S630: Adjust the corresponding early warning configuration parameters in the invoked safety monitoring model according to the first hidden danger coefficient.

[0070] Furthermore, step S600 in this embodiment of the application also includes:

[0071] Step S640: When the bridge life index of the first matched bridge is greater than or equal to the bridge life index of the first target bridge, obtain the real-time life node of the first target bridge.

[0072] Step S650: Obtain the bridge features of the first matched bridge at the real-time lifetime node, and update the source dataset for transfer learning based on the bridge features.

[0073] In one possible embodiment, the safety monitoring model after transfer learning is more adapted to the bridge characteristics of the first target bridge. Therefore, using it to monitor the safety of the first target bridge can improve the accuracy and efficiency of monitoring, as well as the quality of safety protection.

[0074] Specifically, the bridge life index of the first target bridge and the life index of the first matching bridge are obtained as basic data for analyzing the impact of bridge life on safety protection. Preferably, the bridge life index is calculated using nonlinear cumulative damage. When the bridge life index of the first matching bridge is less than that of the first target bridge, it indicates that the design life of the first matching bridge is less than that of the first target bridge. In this case, there is a potential hazard in safety monitoring of the first target bridge based on the safety monitoring model of the first matching bridge. The difference between the bridge life index of the first target bridge and the bridge life index of the first matching bridge, divided by the bridge life index of the first target bridge, is used as the first hazard coefficient. Furthermore, the corresponding early warning configuration parameters in the invoked safety monitoring model are adjusted according to the first hazard coefficient. Preferably, the larger the first hazard coefficient, the greater the range of adjustment for the corresponding early warning configuration parameters in the invoked safety monitoring model.

[0075] In one possible embodiment, when the bridge life index of the first matched bridge is greater than or equal to the bridge life index of the first target bridge, it indicates that the design life of the first matched bridge is greater than the bridge life index of the first target bridge, thereby obtaining the real-time life node of the first target bridge. The real-time life node is the bridge life time point corresponding to the real-time state of the first target bridge. Furthermore, the bridge features of the first matched bridge at the real-time life node are obtained, and the source data for transfer learning is updated using these bridge features, thereby improving the monitoring accuracy of the safety monitoring model after transfer learning.

[0076] In summary, the embodiments of this application have at least the following technical effects:

[0077] This application acquires the bridge features of a first target bridge, then uses a bridge protection IoT management platform to perform a similar search to obtain the first matching bridge with the highest matching degree. Based on the safety monitoring model corresponding to the first matching bridge, and combined with a first loss function, it determines the loss data for transfer learning, thereby improving model building efficiency, shortening the building cycle, and increasing monitoring accuracy. The safety monitoring model is then transferred to the loss data. When the loss data region converges, a transferred-learned safety monitoring model that conforms to the first target bridge is obtained, which is then used for safety monitoring of the first target bridge. This achieves the technical effects of improving safety monitoring accuracy, shortening the monitoring model building cycle, and improving the efficiency and quality of safety protection.

[0078] Example 2

[0079] Based on the same inventive concept as the safety protection method for a super-large bridge over a grade-separated overpass in the foregoing embodiments, such as Figure 4 As shown, this application provides a safety protection system for a grade-separated overpass on a large bridge. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0080] Bridge feature acquisition module 11, the bridge feature acquisition module 11 is used to acquire the bridge features of the first target bridge, wherein the bridge features include bridge architectural features, the set geographical features and bridge environmental features.

[0081] The first matching bridge acquisition module 12 is used to connect to the bridge protection IoT management platform, and search for similar bridges according to the bridge characteristics to acquire the first matching bridge.

[0082] The request information obtaining module 13 is used to send a first request information to the safety accident information management platform terminal corresponding to the first matched bridge, in order to obtain the permission to call the safety monitoring model corresponding to the first matched bridge.

[0083] The loss data output module 14 is used to store the safety monitoring model after the permission request is approved, and to introduce a first loss function to perform loss analysis on the bridge features and output loss data, wherein the loss data includes building loss data, regional loss data and environmental loss data.

[0084] The security monitoring model acquisition module 15 is used to perform transfer learning on the security monitoring model based on the loss data, and to acquire the security monitoring model after transfer learning when the loss data region converges.

[0085] Safety monitoring module 16 is used to perform safety monitoring on the first target bridge using a safety monitoring model after transfer learning.

[0086] Furthermore, the first matching bridge acquisition module 12 is used to perform the following method:

[0087] A bridge matching model is constructed, wherein the bridge matching model is obtained through training with multiple sets of data, including bridge architectural features, set regional features, and bridge environmental features, as well as identification information that identifies the matching degree of the bridge architectural features, set regional features, and bridge environmental features;

[0088] According to the bridge matching model, the bridge features of the first target bridge are input into the bridge matching model, and the building matching degree, regional matching degree and environmental matching degree are output according to the bridge matching model.

[0089] The bridge matching degree is calculated based on the building matching degree, regional matching degree, and environmental matching degree, and the first matching bridge is output.

[0090] The first matched bridge is the bridge with the highest matching degree among all bridges included in the bridge protection IoT management platform.

[0091] Furthermore, the loss data output module 14 is used to perform the following method:

[0092] Determine whether the first matching degree is greater than or equal to the first preset matching degree. If the first matching degree is greater than or equal to the first preset matching degree, obtain a model invocation instruction, wherein the model invocation instruction is used to invoke the security monitoring model corresponding to the first matching bridge.

[0093] The security monitoring model corresponding to the first matching bridge includes a security model architecture, a model training dataset, and model convergence information.

[0094] By extracting features from the model training dataset, bridge features based on the first matched bridge are output, including bridge architectural features, geographical features, and bridge environmental features.

[0095] The first loss function is introduced to perform loss analysis on the bridge features of the first target bridge and the first matched bridge, and the loss data is output.

[0096] Furthermore, the loss data output module 14 is used to perform the following method:

[0097] The bridge features of the first matched bridge are used as source data, and the bridge features of the first target bridge are used as target data to output loss data. Transfer learning is then performed based on the loss data to build a transfer learning channel.

[0098] The transfer learning channel includes a building feature learning branch, a regional feature learning branch, and an environmental feature learning branch, and the weight network layer embedded in the transfer learning channel is obtained by the size of the loss data in each feature learning branch.

[0099] Furthermore, the safety monitoring model acquisition module 15 is used to perform the following method:

[0100] If the first matching degree is less than the first preset matching degree, the building matching degree, the region matching degree and the environment matching degree are judged respectively, and feature types greater than or equal to the second preset matching degree are obtained, wherein the second preset matching degree is less than the first matching degree;

[0101] The feature types with a matching degree less than the second preset degree are labeled and classified, and the recognition results are output. The recognition results include similar data and dissimilar data.

[0102] The feature types with a matching degree greater than or equal to the second preset degree are used as the first source data, and the similar data in the recognition results are used as the second source data for transfer learning.

[0103] Furthermore, the security monitoring module 16 is used to perform the following methods:

[0104] Obtain the bridge life index of the first target bridge, and obtain the bridge life index of the first matched bridge.

[0105] When the bridge life index of the first matched bridge is less than the bridge life index of the first target bridge, a first hidden danger coefficient is generated.

[0106] Adjust the corresponding early warning configuration parameters in the invoked safety monitoring model according to the first hidden danger coefficient.

[0107] Furthermore, the security monitoring module 16 is used to perform the following methods:

[0108] When the bridge life index of the first matched bridge is greater than or equal to the bridge life index of the first target bridge, the real-time life node of the first target bridge is obtained.

[0109] Obtain the bridge features of the first matched bridge at the real-time lifetime node, and update the source dataset for transfer learning based on the bridge features.

[0110] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in a different order than that shown in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0112] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. A safety protection method for a grade-separated overpass of a super-large bridge, characterized in that, The method includes: Obtain the bridge features of the first target bridge, wherein the bridge features include bridge architectural features, geographical features, and bridge environmental features; Connect to the bridge protection IoT management platform, search for similar bridges based on the bridge characteristics, and obtain the first matching bridge; Send a first request message to the terminal of the safety accident information management platform corresponding to the first matched bridge to obtain permission to call the safety monitoring model corresponding to the first matched bridge; Once the permission request is approved, the safety monitoring model is stored, and a first loss function is introduced to perform loss analysis on the bridge features, outputting loss data, which includes building loss data, regional loss data, and environmental loss data. The security monitoring model is transferred to the loss data, and when the loss data region converges, the security monitoring model after transfer learning is obtained. The safety monitoring model after transfer learning is used to monitor the safety of the first target bridge.

2. The method as described in claim 1, characterized in that, The method further includes: A bridge matching model is constructed, wherein the bridge matching model is obtained through training with multiple sets of data, including bridge architectural features, set regional features, and bridge environmental features, as well as identification information that identifies the matching degree of the bridge architectural features, set regional features, and bridge environmental features; According to the bridge matching model, the bridge features of the first target bridge are input into the bridge matching model, and the building matching degree, regional matching degree and environmental matching degree are output according to the bridge matching model. The bridge matching degree is calculated based on the building matching degree, regional matching degree, and environmental matching degree, and the first matching bridge is output. The first matched bridge is the bridge with the highest matching degree among all bridges included in the bridge protection IoT management platform.

3. The method as described in claim 2, characterized in that, The method further includes: Determine whether the first matching degree is greater than or equal to the first preset matching degree. If the first matching degree is greater than or equal to the first preset matching degree, obtain the model invocation instruction, wherein the model invocation instruction is used to invoke the safety monitoring model corresponding to the first matching bridge. The security monitoring model corresponding to the first matching bridge includes a security model architecture, a model training dataset, and model convergence information. By extracting features from the model training dataset, bridge features based on the first matched bridge are output, including bridge architectural features, geographical features, and bridge environmental features. The first loss function is introduced to perform loss analysis on the bridge features of the first target bridge and the first matched bridge, and the loss data is output.

4. The method as described in claim 2, characterized in that, The method for performing transfer learning on the security monitoring model based on the loss data includes: The bridge features of the first matched bridge are used as source data, and the bridge features of the first target bridge are used as target data to output loss data. Transfer learning is then performed based on the loss data to build a transfer learning channel. The transfer learning channel includes a building feature learning branch, a regional feature learning branch, and an environmental feature learning branch, and the weight network layer embedded in the transfer learning channel is obtained by the size of the loss data in each feature learning branch.

5. The method as described in claim 3, characterized in that, The method for determining whether the first matching degree is greater than the preset matching degree also includes: If the first matching degree is less than the first preset matching degree, the building matching degree, the region matching degree and the environment matching degree are judged respectively, and feature types greater than or equal to the second preset matching degree are obtained, wherein the second preset matching degree is less than the first matching degree; The feature types with a matching degree less than the second preset degree are labeled and classified, and the recognition results are output. The recognition results include similar data and dissimilar data. The feature types with a matching degree greater than or equal to the second preset degree are used as the first source data, and the similar data in the recognition results are used as the second source data for transfer learning.

6. The method as described in claim 2, characterized in that, The method further includes: Obtain the bridge life index of the first target bridge, and obtain the bridge life index of the first matched bridge. When the bridge life index of the first matched bridge is less than the bridge life index of the first target bridge, a first hidden danger coefficient is generated. Adjust the corresponding early warning configuration parameters in the invoked safety monitoring model according to the first hidden danger coefficient.

7. The method as described in claim 6, characterized in that, The method further includes: When the bridge life index of the first matched bridge is greater than or equal to the bridge life index of the first target bridge, the real-time life node of the first target bridge is obtained. Obtain the bridge features of the first matched bridge at the real-time lifetime node, and update the source dataset for transfer learning based on the bridge features.

8. A safety protection system for a grade-separated overpass of a super-large bridge, characterized in that, The system includes: A bridge feature acquisition module is used to acquire the bridge features of a first target bridge, wherein the bridge features include bridge architectural features, geographical features, and bridge environmental features. The first matching bridge acquisition module is used to connect to the bridge protection IoT management platform, and search for similar bridges based on the bridge characteristics to acquire the first matching bridge. The request information acquisition module is used to send a first request information to the safety accident information management platform terminal corresponding to the first matched bridge, in order to obtain the permission to call the safety monitoring model corresponding to the first matched bridge. The loss data output module is used to store the safety monitoring model after the permission request is approved, and to introduce a first loss function to perform loss analysis on the bridge features and output loss data, wherein the loss data includes building loss data, regional loss data and environmental loss data. A security monitoring model acquisition module is used to perform transfer learning on the security monitoring model based on the loss data, and to acquire the security monitoring model after transfer learning when the loss data region converges. A safety monitoring module is used to perform safety monitoring on the first target bridge using a safety monitoring model derived from transfer learning.

Images