A method, system and device for monitoring data and natural language driven bridge finite element model rapid construction and a storage medium

Abstract

The application discloses a bridge finite element model rapid construction method and system driven by monitoring data and natural language, a device and a storage medium, belongs to the technical field of bridge models, and comprises the following steps: constructing a bridge initial finite element model based on Midas Civil NX and natural language; obtaining optimized and corrected bridge finite element model parameters based on the bridge initial finite element model and a PSO optimization model architecture, correcting the bridge initial finite element model, and obtaining a bridge benchmark finite element model which can reflect the true state of the structure; importing real-time bridge key monitoring data into the bridge benchmark finite element model, calculating accurate bridge structure responses and checking coefficients, and performing rapid inspection and visual display of the safety state of the bridge structure. The application realizes efficient construction of the bridge finite element model, is especially suitable for a large number of bridge groups with similar structures, and provides technical support for bridge monitoring system scheme design and safety inspection.

Description

Technical Field

[0001] This invention belongs to the field of bridge health monitoring technology, specifically relating to a method, system, device, and storage medium for rapid construction of bridge finite element models driven by monitoring data and natural language. Background Technology

[0002] Driven by a series of policies from the Ministry of Transport, various regions are accelerating the construction of health monitoring systems for highway and national / provincial road bridges. Finite element analysis plays a crucial role in the layout of monitoring points and the setting of early warning values ​​in bridge health monitoring systems. According to Article 10.4.6 of the "Technical Specification for Monitoring Highway Bridge Structures" (JTT1037-2022), a sub-database of structural finite element models needs to be constructed and saved in a standard file format. Furthermore, these models need to be periodically revised based on monitoring data and maintenance results.

[0003] However, bridge health monitoring technology and finite element model analysis technology are currently disconnected. Most domestic monitoring system platforms primarily use BIM or VR models for bridge structures, lacking structural analysis and calculation capabilities. Since both monitoring data analysis and finite element model analysis aim at structural safety analysis, it is necessary to integrate the two, breaking down data silos and enabling monitoring data-driven finite element model structural analysis, thereby further improving the timeliness and accuracy of structural analysis.

[0004] Traditional manual finite element modeling is cumbersome, requiring users to manually input various modeling parameters into the finite element software interface. This process is time-consuming, labor-intensive, and demands highly skilled technicians. In practical engineering, many bridge monitoring systems (especially lightweight systems) do not perform finite element calculations, relying solely on experience to set up monitoring points. This introduces significant uncertainty into the quality control of the monitoring scheme. Currently, some scholars have conducted research on bridge finite element modeling, but this mainly focuses on rapid bridge finite element modeling, with little attention paid to post-processing functions such as automatic calculation execution, result extraction, and automatic generation of calculation reports.

[0005] Furthermore, bridges constructed in the early stages inevitably experience resistance degradation and damage accumulation due to environmental erosion, load effects, and material aging, thus threatening their operational safety and applicability. However, existing models are mostly constructed based on design drawings and parameters, resulting in highly idealized models. This directly leads to significant discrepancies between calculated and measured values. Therefore, correcting the initial three-dimensional finite element model of the structure is essential for a more accurate and comprehensive study of the structural stress state and to provide a benchmark model for studying other mechanical properties of the structure. However, mainstream finite element model correction methods generally rely on response surface methods to construct surrogate models (such as hypersurface functions) to approximate complex implicit relationships. These methods are not only prone to simulation bias due to overfitting, but the surrogate model construction process itself also increases system complexity and computational burden. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, system, device and storage medium for rapid construction and analysis of bridge finite element models driven by monitoring data and natural language. The finite element model can be directly updated by using PSO optimization model combined with Midas API, which has high accuracy and reliability and improves modeling efficiency.

[0007] To achieve the above objectives, the present invention is implemented using the following technical solution: In a first aspect, the present invention provides a method for rapid construction and analysis of a bridge finite element model driven by monitoring data and natural language, comprising the following steps: Step S1: Construct the initial finite element model of the bridge based on Midas Civil NX and natural language processing; Step S2: Based on the initial finite element model of the bridge and the PSO optimized model architecture, obtain the optimized and corrected bridge finite element model parameters, correct the initial finite element model of the bridge, and obtain a bridge reference finite element model that can reflect the true state of the structure. Step S3: Import the acquired key bridge monitoring data into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status.

[0008] Furthermore, in step S1, an adaptation program for heterogeneous monitoring data such as bridge wind field, structural temperature, and displacement is developed using the Midas Open API. A data parsing and transmission framework is built using the Python language, and the monitoring data is accurately connected to the bridge's initial finite element model through the JSON data protocol.

[0009] Furthermore, in step S2, based on the initial finite element model of the bridge, with the first three natural frequencies of the main beam and the deflection of the key sections as the target responses, and based on the equivalent elastic modulus, Poisson's ratio, equivalent density, equivalent cross-sectional area, equivalent moment of inertia of the Z-axis and the equivalent moment of inertia of the Y-axis of the main beam, the key parameters to be corrected in the bridge model are identified through sensitivity analysis. A combined static and dynamic objective function is constructed, and the value of the combined static and dynamic objective function is calculated by calling the finite element method through the Midas API. The combined static and dynamic objective function is iteratively calculated based on the PSO model architecture to obtain the optimized and corrected bridge finite element model parameters. The optimized and corrected bridge finite element model parameters are updated and applied to the initial finite element model of the bridge based on Midas for model correction, resulting in a bridge reference finite element model that reflects the true state of the bridge structure.

[0010] Furthermore, by establishing a connection between Python and MIDAS, standardizing the JSON data-driven modeling process, and combining DeepSeek large-scale model training with automated script generation, intelligent construction of bridge finite element models is achieved. The steps also include: Enter your username and key in MIDAS CIVIL NX to establish a connection between Midas Civil and the server; Use Python to write code to build an HTTP service, input a username and key, and establish a connection between the application and the server; Standardize all kinds of information required for modeling into JSON format, including creating bridge materials, sections, nodes, elements, boundary conditions, different types of loads, load combinations, and structural response extraction; Python reads JSON sequentially and calls the MIDAS API to create objects in batches: first write the nodes, then the elements, then assign material and section values, then apply boundary and load values, and finally submit static and eigenvalue solutions; after the calculation is completed, it reads back the results database and automatically exports the displacement, internal force, and first three frequencies as CSV, completing the pre- and post-processing closed loop; In the parameterized space, the equivalent elastic modulus, density, cross-sectional dimensions and other variables of the main beam are sampled by Latin hypercube sampling. The process is repeated cyclically to obtain several sets of "input parameters - structural response" sample data, which are then uniformly stored in HDF5 files to form a training dataset for the bridge finite element model for deep learning. Supervised learning is performed on the training dataset of the bridge finite element model based on the DeepSeek large model to obtain the mapping relationship between "model parameters and structural response"; Based on the key parameters of the bridge structure of the project, Python modeling scripts and standardized JSON data files are automatically generated for automatic modeling, calculation and response extraction.

[0011] Furthermore, a joint static and dynamic objective function is constructed, which is a scalar function that quantifies the difference between the finite element model response and the measured response. The formula is expressed as: ; Where X = [x1, x2, ..., x] n ] is the parameter vector to be corrected; f si (x) is the i-th simulated natural frequency calculated using the Midas API; f mi It is the measured i-th natural frequency; d mj (X) is the simulated deflection of the j-th critical section under a specified load condition, calculated using the Midas API; d mj This corresponds to the measured deflection; w f wd It is a weighting coefficient used to balance the importance of dynamic (frequency) and static (deflection) targets in the overall objective function (it can be set according to sensitivity or engineering experience). Constraints: Define the physical feasible range (upper and lower limits) of the parameter x to be corrected.

[0012] Furthermore, in step S2, the PSO optimization model architecture employs the particle swarm optimization algorithm, assuming there is a population of n particles in the m-dimensional solution space. The position vector of the i-th particle is The velocity vector is Particles search for new solutions by continuously adjusting their positions, calculating their fitness value using a fitness function, and recording the optimal solution found. and the optimal position experienced by the entire particle swarm ; Particles pass through and To update its velocity and position representation as follows: ; ; In the formula: Indicates inertia weight, Indicates the current iteration number; Indicates the acceleration factor; This represents a random number that follows a uniform distribution in the interval [0, 1]. These are non-negative constants, used to adjust the maximum step size for flight towards the individual optimal particle and the global optimal particle, respectively; for inertial weights... Generally speaking, a larger inertia weight is beneficial to improving global search capability, while a smaller inertia weight is beneficial to improving local search capability. To better balance the global and local search capabilities of the algorithm, a linearly decreasing inertia weight is used, i.e.: ; In the formula, This represents the maximum inertial weight; This represents the minimum inertia weight; This indicates the maximum number of iterations.

[0013] Furthermore, based on the B / S architecture visualization model, an interactive analysis platform is built using the Vue.js framework, integrating functional modules such as historical analysis backtracking and structural response visualization; the D3.js rendering engine is used to realize the dynamic loading of structural response cloud maps, and the reliability of the system is verified based on actual monitoring data of the engineering bridge.

[0014] Secondly, the present invention provides a rapid construction and analysis system for bridge finite element models driven by monitoring data and natural language, characterized by comprising the following modules: The initial modeling module is used to build the initial finite element model of the bridge based on Midas Civil NX and natural language. The benchmark modeling module is used to obtain optimized and corrected bridge finite element model parameters based on the bridge initial finite element model and PSO optimized model architecture, correct the bridge initial finite element model, and obtain a bridge benchmark finite element model that can reflect the true state of the structure. The inspection visualization module is used to import the acquired key monitoring data of the bridge into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status.

[0015] Thirdly, the present invention provides an electronic device, characterized in that it comprises: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the rapid construction method of a bridge finite element model driven by monitoring data as described in any of the first aspects.

[0016] Fourthly, the present invention provides a computer-readable storage medium, characterized in that, when the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is able to execute the rapid construction method of a bridge finite element model driven by monitoring data as described in any one of the first aspects.

[0017] Compared with existing technologies, the beneficial effects achieved by this invention are as follows: The monitoring data and natural language-driven bridge finite element model rapid construction method, system, equipment, and storage medium provided by this invention allow for direct updating of the finite element model using PSO optimization combined with Midas API; by integrating monitoring data and finite element analysis, and through script modeling and pre- and post-processing, efficiency is improved, and the accuracy and reliability of the model are enhanced; thus, it solves the problems of cumbersome, inefficient, costly, and demanding software proficiency requirements of traditional manual finite element modeling processes. Through natural language input, efficient construction of bridge finite element models can be achieved, making it particularly suitable for a large number of highway and national / provincial highway bridge groups with similar structural forms and high degree of standardization, providing technical support for bridge monitoring system design and safety inspection. Attached Figure Description

[0018] Figure 1 The flowchart illustrates the rapid construction method for bridge finite element models driven by natural language based on monitoring data provided in this embodiment of the invention.

[0019] Figure 2 The Midas API workflow diagram provided for embodiments of the present invention.

[0020] Figure 3 A flowchart of bridge finite element modeling provided for an embodiment of the present invention.

[0021] Figure 4 The flowchart of the PSO optimization algorithm provided in the embodiment of the present invention is shown. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0023] like Figure 1 As shown in the figure, this invention provides a method for rapidly constructing a bridge finite element model driven by monitoring data and natural language, including the following steps: Step S1: Construct the initial finite element model of the bridge based on Midas Civil NX and natural language processing; Step S2: Based on the initial finite element model of the bridge and the PSO optimized model architecture, obtain the optimized and corrected bridge finite element model parameters, correct the initial finite element model of the bridge, and obtain a bridge reference finite element model that can reflect the true state of the structure. Step S3: Import the acquired key bridge monitoring data into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status.

[0024] Application Programming Interfaces (APIs), as interaction standards between software systems, act as "neural connectors" in the digital world. Their core value lies in achieving functional abstraction, resource reuse, and ecosystem expansion through standardized interfaces. Traditional APIs, such as operating system interfaces (file management, hardware control), while solving basic interaction needs, are limited by platform binding and protocol fragmentation. With the development of cloud computing and Web technologies, RESTful APIs, thanks to the universality of the HTTP protocol, stateless architecture, and resource-based design (such as URL location models and HTTP method definition operations), have become the industry standard for cross-platform integration—from API calls for social media sharing buttons and real-time weather data aggregation to services on bank open platforms, all are built on this foundation.

[0025] MIDAS innovatively adopts a web-based API architecture, achieving a triple breakthrough in engineering software interaction paradigms: ① In terms of communication protocols, it replaces traditional in-process calls (DLL / COM) with a combination of HTTP and WebSocket, penetrating firewalls through stateless HTTP transmission commands while utilizing full-duplex WebSocket communication to push analysis progress; ② In terms of architecture design, it abstracts finite element models into REST resources (e.g., GET / models / nodes to retrieve node sets) and uses mapi-key to implement multi-instance routing, enabling a single server to manage hundreds of engineering instances in parallel; ③ In terms of data exchange, it uses JSON format to achieve extremely lightweight transmission. Its hierarchical structure naturally fits the needs of finite element modeling (nodal coordinate arrays, nested material constitutive objects), combined with cross-platform zero-parsing-cost characteristics, constructing a complete chain from desktop modeling to cloud computing. Figure 2 As shown, its workflow is as follows: Step 1: The client obtains a REST API request from the server via the HTTP protocol; Step 2: The server retrieves the request and requests node information; Step 3: The product sends node information to the server; Step 4: After obtaining the node information, the server sends it to the client again.

[0026] In this embodiment, in step S1, the Midas Open API is used to develop an adaptation program for heterogeneous monitoring data such as bridge wind field, structural temperature, and displacement. The Python language is used to build a data parsing and transmission framework, and the monitoring data and the initial finite element model of the bridge are accurately connected through the JSON data protocol.

[0027] In this embodiment, in step S2, based on the initial finite element model of the bridge, the first three natural frequencies of the main beam and the deflection of the key section are taken as the target responses. Based on the equivalent elastic modulus, Poisson's ratio, equivalent density, equivalent cross-sectional area, equivalent moment of inertia of the Z-axis and equivalent moment of inertia of the main beam, the key parameters to be corrected in the bridge model are identified through sensitivity analysis. A combined static and dynamic objective function is constructed, and the value of the combined static and dynamic objective function is calculated by calling the finite element method through the Midas API. The combined static and dynamic objective function is iteratively calculated based on the PSO model architecture to obtain the optimized and corrected bridge finite element model parameters. The optimized and corrected bridge finite element model parameters are updated and applied to the initial finite element model of the bridge based on Midas for model correction, resulting in a bridge reference finite element model that reflects the true state of the bridge structure.

[0028] Traditional manual finite element modeling is cumbersome, requiring users to manually input various modeling parameters in the finite element software interface, which is inefficient. In contrast, the Midas API technology uses JSON (JavaScript Object Notation) natural language as the data format to build a bridge finite element modeling method, which completely changes the interaction mode of bridge finite element software.

[0029] In this embodiment, a connection is established between Python and MIDAS to standardize the JSON data-driven modeling process. Combined with DeepSeek large-scale model training and automated script generation, intelligent construction of the bridge finite element model is achieved. The steps also include: Enter your username and key in MIDAS CIVIL NX to establish a connection between Midas Civil and the server; Use Python to write code to build an HTTP service, input a username and key, and establish a connection between the application and the server; Standardize all kinds of information required for modeling into JSON format, including creating bridge materials, sections, nodes, elements, boundary conditions, different types of loads, load combinations, and structural response extraction; Python reads JSON sequentially and calls the MIDAS API to create objects in batches: first write the nodes, then the elements, then assign material and section values, then apply boundary and load values, and finally submit static and eigenvalue solutions; after the calculation is completed, it reads back the results database and automatically exports the displacement, internal force, and first three frequencies as CSV, completing the pre- and post-processing closed loop; In the parameterized space, the equivalent elastic modulus, density, cross-sectional dimensions and other variables of the main beam are sampled by Latin hypercube sampling. The process is repeated cyclically to obtain several sets of "input parameters - structural response" sample data, which are then uniformly stored in HDF5 files to form a training dataset for the bridge finite element model for deep learning. Supervised learning is performed on the training dataset of the bridge finite element model based on the DeepSeek large model to obtain the mapping relationship between "model parameters and structural response"; Based on the key parameters of the bridge structure of the project, Python modeling scripts and standardized JSON data files are automatically generated for automatic modeling, calculation and response extraction.

[0030] In this embodiment, the code is run, using a third-party JSON library in Python to parse JSON data. Then, the MIDAS API function is used to send commands to the server. The server retrieves the request and sends the request information to the MidasCivil finite element software. MidasCivil automatically performs finite element pre- and post-processing operations such as creating new projects, defining materials and sections, establishing nodes and elements, inputting boundary conditions, inputting loads, load combinations, and extracting structural response.

[0031] In this embodiment, the load input includes three parts: the design load, binary text from the bridge monitoring system database, and manually specified load. For the binary text, the structural temperature, wind speed, and other load data transmitted from the InfluxDB database of the bridge monitoring system are converted into JSON format for loading. For manually specified loads, they are directly edited according to the JSON format. For the design load, it is directly input according to the design load level.

[0032] The binary text to JSON conversion is as follows: ① Data parsing: Retrieve binary data from InfluxDB and parse it according to a predefined data structure (such as sensor ID, timestamp, temperature, wind speed, etc.). ② Data conversion: Convert the parsed data into a Python dictionary structure according to the MIDAS load input format. ③ JSON serialization: Serialize the converted data structure into a JSON string using Python's json library.

[0033] In this embodiment, sensitivity analysis studies how changes in risk factors qualitatively and quantitatively affect the structural response of a forward model, and is a necessary step in revising the finite element model. The single-factor sensitivity analysis method assumes that the influencing factors are independent of each other. By controlling variables, other risk factors are fixed, and the values ​​of the target factor are changed one by one, calculating the model output value for each. Then, the sensitivity of the target factor is analyzed based on the changing pattern of the output value with the level value of the target factor.

[0034] In this embodiment, a combined static and dynamic objective function is constructed, which is a scalar function that quantifies the difference between the finite element model response and the measured response. The formula is expressed as: ; Where X = [x1, x2, ..., x] n ] is the parameter vector to be corrected; f si (x) is the i-th simulated natural frequency calculated using the Midas API; f mi It is the measured i-th natural frequency; d mj(X) is the simulated deflection of the j-th critical section under a specified load condition, calculated using the Midas API; d mj This corresponds to the measured deflection; w f w d It is a weighting coefficient used to balance the importance of dynamic (frequency) and static (deflection) targets in the overall objective function (it can be set according to sensitivity or engineering experience). Constraints: Define the physical feasible range (upper and lower limits) of the parameter x to be corrected.

[0035] In this embodiment, in step S2, the PSO optimization model architecture employs the particle swarm optimization algorithm. It is assumed that in the m-dimensional solution space, there is a swarm consisting of n particles. The position vector of the i-th particle is The velocity vector is Particles search for new solutions by continuously adjusting their positions, calculating their fitness value using a fitness function, and recording the optimal solution found. and the optimal position experienced by the entire particle swarm ; Particles pass through and To update its velocity and position representation as follows: ; ; In the formula: Indicates inertia weight, Indicates the current iteration number; Indicates the acceleration factor; This represents a random number that follows a uniform distribution in the interval [0, 1]. These are non-negative constants, used to adjust the maximum step size for flight towards the individual optimal particle and the global optimal particle, respectively; for inertial weights... Generally speaking, a larger inertia weight is beneficial to improving global search capabilities, while a smaller inertia weight is beneficial to improving local search capabilities.

[0036] To better balance the global and local search capabilities of the algorithm, a linearly decreasing inertia weight is used, i.e.: ; In the formula, This represents the maximum inertial weight; This represents the minimum inertia weight; This indicates the maximum number of iterations.

[0037] In this embodiment, a visualization model based on a B / S architecture is used to build an interactive analysis platform using the Vue.js framework, integrating functional modules such as historical analysis backtracking and structural response visualization; the D3.js rendering engine is used to realize the dynamic loading of structural response cloud maps, and the reliability of the system is verified based on actual monitoring data of the engineering bridge.

[0038] In some embodiments, the present invention provides a rapid bridge finite element model construction system driven by monitoring data and natural language, characterized in that it includes the following modules: The initial modeling module is used to build the initial finite element model of the bridge based on Midas Civil NX and natural language. The benchmark modeling module is used to obtain optimized and corrected bridge finite element model parameters based on the bridge initial finite element model and PSO optimized model architecture, correct the bridge initial finite element model, and obtain a bridge benchmark finite element model that can reflect the true state of the structure. The inspection visualization module is used to import the acquired key monitoring data of the bridge into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status.

[0039] In some embodiments, the present invention provides an electronic device, characterized in that it comprises: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the rapid construction method of a bridge finite element model driven by monitoring data as described in any of the first aspects.

[0040] In some embodiments, the present invention provides a computer-readable storage medium, characterized in that, when the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is able to execute the rapid construction method of a bridge finite element model driven by monitoring data as described in any one of the first aspects.

[0041] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for rapid construction of bridge finite element models driven by monitoring data and natural language, characterized in that, Includes the following steps: Step S1: Construct the initial finite element model of the bridge based on Midas Civil NX and natural language processing; Step S2: Based on the initial finite element model of the bridge and the PSO optimized model architecture, obtain the optimized and corrected bridge finite element model parameters, correct the initial finite element model of the bridge, and obtain a bridge reference finite element model that can reflect the true state of the structure. Step S3: Import the acquired key bridge monitoring data into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status. In step S1, using the Midas Open API, an adaptation program for monitoring heterogeneous bridge wind field, structural temperature, and displacement is developed. A data parsing and transmission framework is built using Python, and the monitoring data is accurately connected to the bridge's initial finite element model through the JSON data protocol. In step S2, based on the initial finite element model of the bridge, the first three natural frequencies of the main beam and the deflection of the key sections are taken as the target responses. Based on the equivalent elastic modulus, Poisson's ratio, equivalent density, equivalent cross-sectional area, equivalent moment of inertia of the Z-axis and equivalent moment of inertia of the main beam, the key parameters to be corrected in the bridge model are identified through sensitivity analysis. A combined static and dynamic objective function is constructed, and the value of the combined static and dynamic objective function is calculated by calling the finite element method through the Midas API. The combined static and dynamic objective function is iteratively calculated based on the PSO model architecture to obtain the optimized and corrected bridge finite element model parameters. The optimized and corrected bridge finite element model parameters are updated to the initial bridge finite element model based on Midas for model correction, so as to obtain a bridge reference finite element model that reflects the true state of the bridge structure. By establishing a connection between Python and MIDAS, standardizing the JSON data-driven modeling process, and combining DeepSeek large-scale model training with automated script generation, intelligent construction of bridge finite element models is achieved. The steps also include: Enter your username and key in MIDAS CIVIL NX to establish a connection between Midas Civil and the server; Use Python to write code to build an HTTP service, input a username and key, and establish a connection between the application and the server; Standardize all kinds of information required for modeling into JSON format, including creating bridge materials, sections, nodes, elements, boundary conditions, different types of loads, load combinations, and structural response extraction; Python reads JSON sequentially and calls the MIDAS API to create objects in batches: first write the nodes, then the elements, then assign material and section values, then apply boundary and load values, and finally submit static and eigenvalue solutions; after the calculation is completed, it reads back the results database and automatically exports the displacement, internal force, and first three frequencies as CSV, completing the pre- and post-processing closed loop; Latin hypercube sampling is performed on the equivalent elastic modulus, density, and cross-sectional dimensions of the main beam in the parameterized space. This process is repeated cyclically to obtain several sets of "input parameters - structural response" sample data, which are then uniformly stored in an HDF5 file to form a training dataset for the bridge finite element model used for deep learning. Supervised learning is performed on the training dataset of the bridge finite element model based on the DeepSeek large model to obtain the mapping relationship between "model parameters and structural response"; Based on the key parameters of the bridge structure of the project, Python modeling scripts and standardized JSON data files are automatically generated for automatic modeling, calculation and response extraction.

2. The method for rapid construction of bridge finite element models driven by monitoring data and natural language as described in claim 1, characterized in that, A joint static and dynamic objective function is constructed, which is a scalar function that quantifies the difference between the finite element model response and the measured response. The formula is as follows: ； Where X = [x1, x2, ..., x] n ] is the parameter vector to be corrected; f si (x) is the i-th simulated natural frequency calculated using the Midas API; f mi It is the measured i-th natural frequency; d mj (X) is the simulated deflection of the j-th critical section under a specified load condition, calculated using the Midas API; d mj This corresponds to the measured deflection; w f w d It is a weighting coefficient used to balance the importance of dynamic frequency and static deflection targets in the overall objective function. Constraint: Define the physical feasible range of the parameter x to be corrected.

3. The method for rapid construction of bridge finite element models driven by monitoring data and natural language as described in claim 1, characterized in that, In step S2, the PSO optimization model architecture employs the particle swarm optimization algorithm. Assume there is a population of n particles in the m-dimensional solution space. The position vector of the i-th particle is The velocity vector is Particles search for new solutions by continuously adjusting their positions, calculating their fitness value using a fitness function, and recording the optimal solution found. and the optimal position experienced by the entire particle swarm ; Particles pass through and To update its velocity and position representation as follows: ； ； In the formula: Indicates inertia weight, Indicates the current iteration number; Indicates the acceleration factor; This represents a random number that follows a uniform distribution in the interval [0, 1]. These are non-negative constants, used to adjust the maximum step size for flight towards the individual optimal particle and the global optimal particle, respectively; for inertial weights... Generally speaking, a larger inertia weight is beneficial to improving global search capability, while a smaller inertia weight is beneficial to improving local search capability. To better balance the global and local search capabilities of the algorithm, a linearly decreasing inertia weight is used, i.e.: ； In the formula, This represents the maximum inertial weight; This represents the minimum inertia weight; This indicates the maximum number of iterations.

4. The method for rapid construction of bridge finite element models driven by monitoring data and natural language as described in claim 1, characterized in that, Based on the B / S architecture, a visualization model is built using the Vue.js framework to create an interactive analysis platform that integrates historical analysis backtracking and structural response visualization modules. The D3.js rendering engine is used to dynamically load the structural response cloud map, and the reliability of the system is verified based on the actual monitoring data of the bridge in the project.

5. A rapid construction and analysis system for bridge finite element models driven by monitoring data and natural language, characterized in that, The method for rapid construction of bridge finite element models using monitoring data as described in any one of claims 1 to 4, and natural language-driven modeling, is implemented. Includes the following modules: The initial modeling module is used to build the initial finite element model of the bridge based on Midas Civil NX and natural language. The benchmark modeling module is used to obtain optimized and corrected bridge finite element model parameters based on the bridge initial finite element model and PSO optimized model architecture, correct the bridge initial finite element model, and obtain a bridge benchmark finite element model that can reflect the true state of the structure. The inspection visualization module is used to import the acquired key monitoring data of the bridge into the bridge reference finite element model in real time, calculate the accurate bridge structural response and verification coefficients, and perform rapid inspection and visualization of the bridge structural safety status.

6. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute the instructions to implement the rapid construction method of monitoring data and natural language-driven bridge finite element model as described in any one of claims 1 to 4.

7. A computer-readable storage medium, characterized in that, When the instructions in the storage medium are executed by the processor of the electronic device, the electronic device is able to execute the rapid construction method of the bridge finite element model driven by monitoring data as described in any one of claims 1 to 4.

Images