Bridge structure risk safety identification method based on artificial intelligence

Abstract

The invention discloses a bridge structure risk safety identification method based on artificial intelligence, and relates to the technical field of bridge structure risk identification, and the method comprises the steps: building a topological graph model which reflects the geometric and mechanical connection relation of a bridge, taking piers, main beam sections and supports as graph nodes, and taking the physical connection between components as graph edges; structure response signals and environment load parameters in the service period of the bridge are collected, and the collected data are synchronized according to time and then mapped to corresponding nodes and edges in the topological graph model; based on the mapped data and a topological graph model, establishing a graph neural network embedded with structural dynamic constraints, and outputting the risk probability of a component corresponding to each node by jointly optimizing the consistency of a monitoring data fitting error and a physical rule; and performing causal relationship analysis on the structural response signal and the environmental load parameter, identifying a causal path between environmental interference and structural abnormality, and separating an abnormal component caused by structural degradation from the original response according to an identification result.

Description

Technical Field

[0001] This invention relates to the field of bridge structure risk identification technology, and in particular to an artificial intelligence-based method for identifying bridge structure risks and safety. Background Technology

[0002] Bridge structural risk identification technology refers to a class of technical methods that monitor, analyze, and assess the structural performance degradation or damage caused by factors such as material aging, environmental erosion, excessive loads, construction defects, or unforeseen events during the service life of a bridge, thereby determining whether there are any safety hazards and quantifying the risk level. Therefore, how to utilize advanced technologies to improve the intelligence and security of file encryption has become one of the urgent problems to be solved.

[0003] Existing data-driven artificial intelligence models often misjudge structural responses caused by environmental factors such as temperature, wind, and traffic flow as damage signals. They lack the ability to distinguish between correlation and causation, resulting in a large number of false warnings and reducing the credibility of the system. Furthermore, bridges are composed of heterogeneous components such as piers, main beams, and supports. Traditional convolutional neural networks or recurrent neural networks cannot effectively express their non-Euclidean spatial relationships, making it difficult for local damage to propagate to global risk assessment. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an artificial intelligence-based method for identifying bridge structural risks and safety. This addresses the problem that existing data-driven artificial intelligence models often misjudge structural responses caused by environmental factors such as temperature, wind, and traffic flow as damage signals, lack the ability to distinguish between correlation and causation, resulting in numerous false warnings and reduced system credibility. Furthermore, bridges are composed of heterogeneous components such as piers, main beams, and supports, and traditional convolutional neural networks or recurrent neural networks cannot effectively express their non-Euclidean spatial relationships, making it difficult for local damage to propagate to global risk assessment.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a bridge structural risk safety identification method based on artificial intelligence, comprising:

[0008] A topological graph model reflecting the geometric and mechanical connections of a bridge is constructed, with piers, main beam segments, and supports as graph nodes and physical connections between components as graph edges.

[0009] Structural response signals and environmental load parameters of the bridge during its service life are collected, and the collected data are mapped to the corresponding nodes and edges in the topological graph model after being synchronized by time.

[0010] Based on the mapped data and topological graph model, a graph neural network with embedded structural dynamic constraints is established. By jointly optimizing the consistency between the monitoring data fitting error and physical laws, the risk probability of each node corresponding to the component is output.

[0011] Causal relationship analysis is performed on the structural response signal and environmental load parameters to identify the causal path between environmental disturbance and structural anomaly. Based on the identification results, the abnormal components caused by structural degradation are separated from the original response to generate decontaminated abnormal response data.

[0012] The decontamination of abnormal response data is input into the inversion network of the integrated finite element physical model. The stiffness degradation state of each component is inverted under the condition of maintaining mechanical equilibrium using the differentiable finite element mechanism, and a physically interpretable damage index is output.

[0013] By combining the risk probability output by the graph neural network with the damage index output by the inversion network, a comprehensive dynamic risk index is calculated.

[0014] Based on the comparison between the comprehensive dynamic risk index and the preset threshold, the corresponding level of risk warning is triggered, and the safety identification of bridge structural risks is completed.

[0015] As a preferred embodiment of the bridge structural risk and safety identification method based on artificial intelligence described in this invention, the construction of a topological graph model reflecting the geometric and mechanical connection relationship of the bridge, with piers, main beam segments, and supports as graph nodes and physical connections between components as graph edges, specifically involves the following steps:

[0016] Extract all load-bearing components and their connection methods based on the bridge design data;

[0017] Each pier, main beam segment, and support is defined as a graph node, and the set of nodes is denoted as . If two components are physically directly connected, then an edge is established between the corresponding nodes, and the set of edges is denoted as . To form an undirected graph ;

[0018] Associate structural properties with each node, including the moment of inertia and elastic modulus of the cross section;

[0019] Associate connection attributes for each edge, including constraint type and damping characteristics.

[0020] As a preferred embodiment of the bridge structural risk and safety identification method based on artificial intelligence described in this invention, the steps of collecting structural response signals and environmental load parameters during the bridge's service life, and mapping the collected data to corresponding nodes and edges in the topological graph model after time synchronization, are as follows:

[0021] Accelerometers, strain gauges, and displacement sensors were installed on the bridge, and thermometers and anemometers were deployed simultaneously.

[0022] Time series data are continuously collected at a uniform sampling frequency;

[0023] Perform timestamp alignment on all channels to eliminate asynchronous transmission and generate a synchronous data matrix;

[0024] Based on the sensor installation location, acceleration data is mapped to pier nodes, strain data is mapped to main beam nodes, and support displacement data is mapped to connection edges;

[0025] This generates graph signals that correspond one-to-one with the nodes and edges of the topological graph model.

[0026] As a preferred embodiment of the bridge structure risk and safety identification method based on artificial intelligence described in this invention, the following steps are taken: A graph neural network with embedded structural dynamic constraints is established based on the mapped data and topological graph model. By jointly optimizing the consistency between the fitting error of the monitoring data and the physical laws, the risk probability of the corresponding component at each node is output.

[0027] A multi-layer graph neural network is constructed, in which the message passing process integrates the information of adjacent nodes and edge attributes;

[0028] During network training, the loss function is defined by two terms:

[0029] The first term is the mean square error between the predicted displacement and the measured displacement, and the second term is the residual of the structural dynamics equations.

[0030] The dynamic residuals are calculated based on the mass, damping, stiffness matrices, and external loads.

[0031] The total loss is expressed as:

[0032] ;

[0033] in, To predict displacement for the network, For the measured displacement, , , These are the mass, damping, and stiffness matrices, respectively. For external load vectors, , Regularization weights; minimize Train the network so that its output both fits the data and satisfies the laws of physics;

[0034] After training, a classification head is applied to the node representation, and the risk probability of each component is output.

[0035] As a preferred embodiment of the bridge structural risk and safety identification method based on artificial intelligence described in this invention, the steps of performing causal relationship analysis on the structural response signal and environmental load parameters, identifying the causal path between environmental disturbances and structural anomalies, separating the abnormal components caused by structural degradation from the original response based on the identification results, and generating decontamination-free abnormal response data are as follows:

[0036] The synchronized data is divided into a set of structural response variables and a set of environmental variables.

[0037] A constrained gene-fruit discovery algorithm is used to construct a directed causal graph on time-series data;

[0038] If a structural response variable is only causally affected by environmental variables, then its residual is the part that cannot be explained by the environment.

[0039] A regression model is built for each response variable, and the residuals are calculated as follows:

[0040] ;

[0041] in, This is the original response. The response predicted by environmental variables;

[0042] All such residuals are spliced ​​together to form a decontamination sequence of abnormal responses.

[0043] As a preferred embodiment of the bridge structure risk and safety identification method based on artificial intelligence described in this invention, the steps of inputting decontamination-free abnormal response data into the inversion network of the integrated finite element physical model, using the differentiable finite element mechanism to invert the stiffness degradation state of the component while maintaining mechanical equilibrium, and outputting physically interpretable damage indicators are as follows:

[0044] Initialize the finite element model of the bridge and assign initial stiffness parameters to each component;

[0045] Construct a differentiable finite element solver, whose forward computation solution is expressed as:

[0046] ;

[0047] Theoretical displacement is obtained ,in, Let be the stiffness parameter to be inverted, then the inversion loss is defined as:

[0048] ;

[0049] in, The resulting abnormal response is updated via automatic differentiation. Continue until convergence, and obtain the optimized stiffness parameters. The stiffness reduction rate is calculated for each component, and the expression is:

[0050] ;

[0051] in, For initial stiffness, This is a damage indicator.

[0052] As a preferred embodiment of the bridge structure risk and safety identification method based on artificial intelligence described in this invention, the specific steps for calculating the comprehensive dynamic risk index by combining the risk probability output by the graph neural network and the damage index output by the inversion network are as follows:

[0053] Obtain risk probability With damage indicators The comprehensive dynamic risk index is defined as follows:

[0054] ;

[0055] in, For adjustable weights, The value continuously reflects the overall risk level of the bridge; the higher the value, the lower the safety margin.

[0056] As a preferred embodiment of the bridge structure risk safety identification method based on artificial intelligence described in this invention, the specific steps for triggering a risk warning of the corresponding level based on the comparison result of the comprehensive dynamic risk index and the preset threshold, thereby completing the safety identification of bridge structure risks, are as follows:

[0057] Set three-level early warning thresholds Real-time calculation ;

[0058] when < The condition is determined to be safe.

[0059] when ≤ < This triggered a Level 1 warning, prompting increased patrols.

[0060] when ≤ < This triggers a Level II alert; special testing is recommended.

[0061] when ≥ This triggers a Level 3 warning, restricts traffic, and initiates an emergency assessment.

[0062] Each warning will simultaneously output a list of high-risk components.

[0063] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the artificial intelligence-based bridge structure risk safety identification method described in the first aspect of the present invention.

[0064] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the artificial intelligence-based bridge structure risk safety identification method described in the first aspect of the present invention.

[0065] The beneficial effects of this invention are as follows: By integrating graph neural networks, causal reasoning, and differentiable finite element inversion, a three-in-one intelligent bridge risk identification system integrating geometry, physics, and data is constructed. This effectively solves the problems of traditional artificial intelligence methods in bridge health monitoring, such as susceptibility to environmental interference, lack of physical consistency, ambiguous damage location, and high false alarm rate. The method can not only accurately separate structural degradation signals from multi-source heterogeneous monitoring data, but also simultaneously output risk probability and stiffness damage indicators with engineering interpretability. Furthermore, it achieves graded early warning through dynamic risk index, effectively improving the accuracy, robustness, and practicality of bridge structural safety assessment. Attached Figure Description

[0066] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0067] Figure 1 This is a flowchart of the bridge structure risk safety identification method based on artificial intelligence in Example 1. Detailed Implementation

[0068] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0069] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0070] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0071] Example 1, referring to Figure 1 This embodiment of the invention provides a bridge structure risk safety identification method based on artificial intelligence, including:

[0072] S1. Construct a topological graph model that reflects the geometric and mechanical connection relationship of the bridge, with piers, main beam segments and supports as graph nodes, and physical connections between components as graph edges;

[0073] Furthermore, all load-bearing components and their connection methods are extracted based on the bridge design data;

[0074] Each pier, main beam segment, and support is defined as a graph node, and the set of nodes is denoted as . If two components are physically directly connected, then an edge is established between the corresponding nodes, and the set of edges is denoted as . To form an undirected graph ;

[0075] Associate structural properties with each node, including the moment of inertia and elastic modulus of the cross section;

[0076] Associate connection attributes with each edge, including constraint type and damping characteristics;

[0077] It should be noted that abstracting the various components of a bridge into nodes and representing their physical connections through edges helps to create a clear framework for displaying the bridge structure and its mechanical relationships. The model not only simplifies the expression of complex structures but also provides a basic framework for subsequent analysis, making the assessment of the bridge's health status more systematic and accurate.

[0078] S2. Collect structural response signals and environmental load parameters of the bridge during its service life, and map the collected data to the corresponding nodes and edges in the topology model after synchronizing the data over time.

[0079] Furthermore, accelerometers, strain gauges, and displacement sensors are installed on the bridge, along with thermometers and anemometers.

[0080] Time series data are continuously collected at a uniform sampling frequency;

[0081] Perform timestamp alignment on all channels to eliminate asynchronous transmission and generate a synchronous data matrix;

[0082] Based on the sensor installation location, acceleration data is mapped to pier nodes, strain data is mapped to main beam nodes, and support displacement data is mapped to connection edges;

[0083] Form graph signals that correspond one-to-one with the nodes and edges of the topological graph model;

[0084] It should be noted that ensuring accurate synchronization of data from all sensors during the data acquisition phase is crucial. This step guarantees that data from different sensors can be analyzed on the same time base, thereby improving the reliability and effectiveness of monitoring data, which is essential for correctly identifying changes in the state of the bridge structure.

[0085] S3. Based on the mapped data and topological graph model, establish a graph neural network with embedded structural dynamic constraints. By jointly optimizing the consistency between the monitoring data fitting error and physical laws, output the risk probability of each node corresponding to the component.

[0086] Furthermore, a multi-layer graph neural network is constructed, in which the message passing process integrates neighbor node information and edge attributes;

[0087] During network training, the loss function is defined by two terms:

[0088] The first term is the mean square error between the predicted displacement and the measured displacement, and the second term is the residual of the structural dynamics equations.

[0089] The dynamic residuals are calculated based on the mass, damping, stiffness matrices, and external loads.

[0090] The total loss is expressed as:

[0091] ;

[0092] in, To predict displacement for the network, For the measured displacement, , , These are the mass, damping, and stiffness matrices, respectively. For external load vectors, , Regularization weights; minimize Train the network so that its output both fits the data and satisfies the laws of physics;

[0093] After training, a classification head is applied to the node representations to output the risk probability of each component;

[0094] It should be noted that by constructing a graph neural network and incorporating structural dynamic constraints, the real behavior of bridge structures can be effectively simulated. The method can not only capture the dynamic characteristics of the structure, but also maintain consistency with physical laws during the prediction process, thereby providing more accurate risk assessment results.

[0095] S4. Perform causal relationship analysis on the structural response signal and environmental load parameters, identify the causal path between environmental disturbance and structural anomaly, and separate the abnormal components caused by structural degradation from the original response based on the identification results to generate decontaminated abnormal response data.

[0096] Furthermore, the synchronized data is divided into a set of structural response variables and a set of environmental variables;

[0097] A constrained gene-fruit discovery algorithm is used to construct a directed causal graph on time-series data;

[0098] If a structural response variable is only causally affected by environmental variables, then its residual is the part that cannot be explained by the environment.

[0099] A regression model is built for each response variable, and the residuals are calculated as follows:

[0100] ;

[0101] in, This is the original response. The response predicted by environmental variables;

[0102] All such residuals are spliced ​​together to form a decontamination sequence of abnormal responses;

[0103] It should be noted that causal relationship analysis can help distinguish between normal variations caused by environmental factors and abnormalities caused by structural degradation. In this way, information that truly reflects the health status of the structure can be filtered out from the monitoring data, thereby improving the accuracy and reliability of anomaly detection.

[0104] S5. Input the decontamination of abnormal response data into the inversion network of the integrated finite element physical model, and use the differentiable finite element mechanism to invert the stiffness degradation state of each component under the condition of maintaining mechanical equilibrium, and output physically interpretable damage indicators.

[0105] Furthermore, the bridge finite element model is initialized, and initial stiffness parameters are assigned to each component;

[0106] Construct a differentiable finite element solver, whose forward computation solution is expressed as:

[0107] ;

[0108] Theoretical displacement is obtained ,in, Let be the stiffness parameter to be inverted, then the inversion loss is defined as:

[0109] ;

[0110] in, The resulting abnormal response is updated via automatic differentiation. Continue until convergence, and obtain the optimized stiffness parameters. The stiffness reduction rate is calculated for each component, and the expression is:

[0111] ;

[0112] in, For initial stiffness, As a damage indicator;

[0113] It should be noted that using inversion networks combined with the finite element method to estimate the change in component stiffness provides an effective way to extract structural damage information from actual monitoring data. The process can not only quantify the degree of structural damage, but also provide a direct basis for maintenance decisions, thus promoting the development of condition-based maintenance strategies.

[0114] S6. Combine the risk probability output by the graph neural network with the damage index output by the inversion network to calculate the comprehensive dynamic risk index.

[0115] Furthermore, obtain the probability of risk. With damage indicators The comprehensive dynamic risk index is defined as follows:

[0116] ;

[0117] in, For adjustable weights, The value continuously reflects the overall risk level of the bridge; the higher the value, the lower the safety margin.

[0118] It should be noted that the method of comprehensively considering risk probability and damage indicators to calculate the overall risk index provides a comprehensive assessment of bridge safety status. This method can simultaneously reflect potential short-term risks and long-term structural health trends, supporting more precise safety management and maintenance planning.

[0119] S7. Based on the comparison results between the comprehensive dynamic risk index and the preset threshold, trigger the corresponding level of risk warning and complete the safety identification of bridge structural risks;

[0120] Furthermore, a three-tiered early warning threshold is set. Real-time calculation ;

[0121] when < The condition is determined to be safe.

[0122] when ≤ < This triggered a Level 1 warning, prompting increased patrols.

[0123] when ≤ < This triggers a Level II alert; special testing is recommended.

[0124] when ≥ This triggers a Level 3 warning, restricts traffic, and initiates an emergency assessment.

[0125] Each warning simultaneously outputs a list of high-risk components;

[0126] It should be noted that triggering the corresponding early warning mechanism based on the calculated overall risk index can help managers understand the current safety status of the bridge in a timely manner and take appropriate measures to deal with potential risks. This hierarchical early warning system can effectively improve the efficiency of bridge safety management and protect public safety.

[0127] This embodiment also provides a computer device applicable to the bridge structure risk and safety identification method based on artificial intelligence, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the bridge structure risk and safety identification method based on artificial intelligence as proposed in the above embodiment.

[0128] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0129] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the bridge structure risk safety identification method based on artificial intelligence as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0130] In summary, this invention constructs a three-in-one intelligent bridge risk identification system integrating geometry, physics, and data by fusing graph neural networks, causal reasoning, and differentiable finite element inversion. This effectively solves the problems of traditional artificial intelligence methods in bridge health monitoring, such as susceptibility to environmental interference, lack of physical consistency, ambiguous damage location, and high false alarm rate. The method can not only accurately separate structural degradation signals from multi-source heterogeneous monitoring data, but also simultaneously output risk probability and stiffness damage indicators with engineering interpretability. Furthermore, it achieves graded early warning through dynamic risk index, effectively improving the accuracy, robustness, and practicality of bridge structural safety assessment.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for identifying bridge structural risks and safety based on artificial intelligence, characterized in that: include: A topological graph model reflecting the geometric and mechanical connections of a bridge is constructed, with piers, main beam segments, and supports as graph nodes and physical connections between components as graph edges. Structural response signals and environmental load parameters of the bridge during its service life are collected, and the collected data are mapped to the corresponding nodes and edges in the topological graph model after being synchronized by time. Based on the mapped data and topological graph model, a graph neural network with embedded structural dynamic constraints is established. By jointly optimizing the consistency between the monitoring data fitting error and physical laws, the risk probability of each node corresponding to the component is output. Causal relationship analysis is performed on the structural response signal and environmental load parameters to identify the causal path between environmental disturbance and structural anomaly. Based on the identification results, the abnormal components caused by structural degradation are separated from the original response to generate decontaminated abnormal response data. The decontamination of abnormal response data is input into the inversion network of the integrated finite element physical model. The stiffness degradation state of each component is inverted under the condition of maintaining mechanical equilibrium using the differentiable finite element mechanism, and a physically interpretable damage index is output. By combining the risk probability output by the graph neural network with the damage index output by the inversion network, a comprehensive dynamic risk index is calculated. Based on the comparison between the comprehensive dynamic risk index and the preset threshold, the corresponding level of risk warning is triggered, and the safety identification of bridge structural risks is completed.

2. The bridge structural risk and safety identification method based on artificial intelligence as described in claim 1, characterized in that: The construction of the topological graph model reflecting the geometric and mechanical connections of the bridge involves using piers, main beam segments, and supports as graph nodes, and physical connections between components as graph edges. The specific steps are as follows: Extract all load-bearing components and their connection methods based on the bridge design data; Each pier, main beam segment, and support is defined as a graph node, and the set of nodes is denoted as . If two components are physically directly connected, then an edge is established between the corresponding nodes, and the set of edges is denoted as . To form an undirected graph ; Associate structural properties with each node, including the moment of inertia and elastic modulus of the cross section; Associate connection attributes for each edge, including constraint type and damping characteristics.

3. The bridge structure risk and safety identification method based on artificial intelligence as described in claim 2, characterized in that: The process involves collecting structural response signals and environmental load parameters of the bridge during its service life, and then mapping the collected data to corresponding nodes and edges in the topological graph model after time synchronization. The specific steps are as follows: Accelerometers, strain gauges, and displacement sensors were installed on the bridge, and thermometers and anemometers were deployed simultaneously. Time series data are continuously collected at a uniform sampling frequency; Perform timestamp alignment on all channels to eliminate asynchronous transmission and generate a synchronous data matrix; Based on the sensor installation location, acceleration data is mapped to pier nodes, strain data is mapped to main beam nodes, and support displacement data is mapped to connection edges; This generates graph signals that correspond one-to-one with the nodes and edges of the topological graph model.

4. The bridge structural risk safety identification method based on artificial intelligence as described in claim 3, characterized in that: Based on the mapped data and topological graph model, a graph neural network with embedded structural dynamic constraints is established. Through joint optimization, the consistency between the monitoring data fitting error and physical laws is observed, and the risk probability of the corresponding component at each node is output. The specific steps are as follows: A multi-layer graph neural network is constructed, in which the message passing process integrates the information of adjacent nodes and edge attributes; During network training, the loss function is defined by two terms: The first term is the mean square error between the predicted displacement and the measured displacement, and the second term is the residual of the structural dynamics equations. The dynamic residuals are calculated based on the mass, damping, stiffness matrices, and external loads. The total loss is expressed as: ; in, To predict displacement for the network, For the measured displacement, , , These are the mass, damping, and stiffness matrices, respectively. For external load vectors, , Regularization weights; minimize Train the network so that its output both fits the data and satisfies the laws of physics; After training, a classification head is applied to the node representation, and the risk probability of each component is output.

5. The bridge structure risk and safety identification method based on artificial intelligence as described in claim 4, characterized in that: The specific steps for performing causal relationship analysis between structural response signals and environmental load parameters, identifying the causal path between environmental disturbances and structural anomalies, separating the anomalous components caused by structural degradation from the original response based on the identification results, and generating decontaminated anomalous response data are as follows: The synchronized data is divided into a set of structural response variables and a set of environmental variables. A constrained gene-fruit discovery algorithm is used to construct a directed causal graph on time-series data; If a structural response variable is only causally affected by environmental variables, then its residual is the part that cannot be explained by the environment. A regression model is built for each response variable, and the residuals are calculated as follows: ； in, This is the original response. The response predicted by environmental variables; All such residuals are spliced ​​together to form a decontamination sequence of abnormal responses.

6. The bridge structural risk safety identification method based on artificial intelligence as described in claim 5, characterized in that: The decontamination of abnormal response data is input into the inversion network of the integrated finite element physical model. The differentiable finite element mechanism is used to invert the stiffness degradation state of the component while maintaining mechanical equilibrium, outputting a physically interpretable damage index. The specific steps are as follows: Initialize the finite element model of the bridge and assign initial stiffness parameters to each component; Construct a differentiable finite element solver, whose forward computation solution is expressed as: ； Theoretical displacement is obtained ,in, Let be the stiffness parameter to be inverted, then the inversion loss is defined as: ; in, The resulting abnormal response is updated via automatic differentiation. Continue until convergence, and obtain the optimized stiffness parameters. The stiffness reduction rate is calculated for each component, and the expression is: ; in, For initial stiffness, This is a damage indicator.

7. The bridge structural risk safety identification method based on artificial intelligence as described in claim 6, characterized in that: The comprehensive dynamic risk index is calculated by combining the risk probability output by the graph neural network and the damage index output by the inversion network. The specific steps are as follows: Obtain risk probability With damage indicators The comprehensive dynamic risk index is defined as follows: ; in, For adjustable weights, The value continuously reflects the overall risk level of the bridge; the higher the value, the lower the safety margin.

8. The bridge structural risk safety identification method based on artificial intelligence as described in claim 7, characterized in that: The process of triggering a risk warning of the corresponding level based on the comparison result of the comprehensive dynamic risk index and the preset threshold, thereby completing the safety identification of bridge structural risks, includes the following steps: Set three-level early warning thresholds Real-time calculation ; when < The condition is determined to be safe. when ≤ < This triggered a Level 1 warning, prompting increased patrols. when ≤ < This triggers a Level II alert; special testing is recommended. when ≥ This triggers a Level 3 warning, restricts traffic, and initiates an emergency assessment. Each warning will simultaneously output a list of high-risk components.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the artificial intelligence-based bridge structure risk safety identification method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the artificial intelligence-based bridge structure risk safety identification method according to any one of claims 1 to 8.

Images