A method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins

Abstract

The invention discloses a fatigue performance evaluation method for steel box girders of cable-stayed bridges based on digital twins. The method is as follows: monitoring and transmission equipment is installed on the bridge, and a physical information model library of the cable-stayed bridge structure is constructed based on the dynamic monitoring data after cleaning; a digital model of the cable-stayed bridge is established based on the spatial discretization method, and twin points are arranged at the monitoring positions of the bridge; the key structural characteristics of the cable-stayed bridge are analyzed based on the monitoring data of the physical information model to obtain the probability interval distribution of the eigenvalues; the parameters of the digital model are controlled and adjusted based on the eigenvalues ​​of the physical information model to realize the construction of a numerical twin model; finally, the physical information model and the numerical twin model are integrated to perform real-time analysis of the stress level and variation amplitude of the steel box girder. The method of the present invention can be used for rapid fatigue performance evaluation of steel box girders of cable-stayed bridges, realize the integration and unification of big data and digital twins, and provide a theoretical basis for tracking and evaluating the service status of cable-stayed bridges throughout their life cycle.

Description

Technical Field

[0001] This invention belongs to the field of transportation technology, specifically relating to a method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins. Background Technology

[0002] Cable-stayed bridges' steel box girders directly bear wheel loads. Under the long-term, repeated action of these wheels, the bridge deck experiences continuous stress fluctuations. These stress amplitude variations can cause structural performance degradation and even fatigue failure. Bridge deck fatigue has become a major challenge in bridge engineering, leading to frequent repairs, reinforcements, and modifications, resulting in significant economic losses. Accurately assessing the fatigue performance of cable-stayed bridge steel box girders is crucial for ensuring the service performance and durability of cable-stayed bridges. For long-span cable-stayed bridges, health monitoring systems are typically installed. Therefore, in traditional management and maintenance, massive amounts of monitoring data are often used to analyze the fatigue stress of the steel box girders, obtaining stress spectra through long-term monitoring to assess the structure's fatigue performance. However, due to the limited number of monitoring points, this approach can only evaluate specific, localized locations and cannot provide a comprehensive assessment of the entire bridge's structural performance. Numerical models such as the finite element method are frequently used to analyze the fatigue performance of bridges. Numerical simulation can evaluate the mechanical properties of the entire bridge. However, there are significant differences between numerical models and actual physical models. The load modes, boundary conditions, and material properties used in the calculations do not perfectly match those in actual engineering projects, resulting in an inaccurate reflection of the structure's service performance. Therefore, there is an urgent need to develop an efficient method for calculating and evaluating the fatigue performance of steel box girders, accurately assessing the mechanical properties of all fatigue-related details of the entire bridge, and providing technical support for the service status assessment of long-span cable-stayed bridges. Summary of the Invention

[0003] Purpose of the invention: To address the above problems, this invention proposes a method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins, so as to quickly and accurately obtain the stress of steel box girders in cable-stayed bridges and evaluate the fatigue performance of all structural details of the entire bridge.

[0004] Technical Solution: To achieve the objectives of this invention, the technical solution adopted is: a method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins, specifically including the following steps:

[0005] Step 1: Acquire and process the dynamic monitoring data of the cable-stayed bridge, and use the processed dynamic monitoring data to construct a physical information model of the cable-stayed bridge structure; the dynamic monitoring data includes input load information and output response information.

[0006] Step 2: Based on the input load information and output response information of the physical information model described in Step 1, the structural characteristics of the cable-stayed bridge are identified, and signal tracking analysis is performed on the physical information model to obtain the structural characteristic feature values ​​and probability interval distribution of the cable-stayed bridge in the physical information model.

[0007] Step 3: Establish a digital model of the cable-stayed bridge using the spatial discretization method, and deploy twin points at the response monitoring locations of the digital model of the cable-stayed bridge;

[0008] Step 4: Based on the digital model described in Step 3, obtain the output response information, structural characteristics, and structural characteristic feature values ​​of the cable-stayed bridge from the digital model through modal analysis and static analysis.

[0009] Step 5: Based on the output response information and structural characteristic values ​​of the physical information model and the output response information and structural characteristic values ​​of the digital model, construct a mapping between the physical information model and the digital model. Use a multi-objective optimization method to adjust and correct the parameters of the digital model described in Step 3 to realize the construction of the numerical twin model.

[0010] Step 6: Analyze the stress time history of the cable-stayed bridge steel box girder using the input load information of the physical information model described in Step 1 and the numerical twin model described in Step 5. Evaluate the stress spectrum and fatigue performance of the cable-stayed bridge steel box girder using the dynamic monitoring data described in Step 1, and obtain the evaluation results.

[0011] Furthermore, the method for acquiring and processing the dynamic monitoring data of the cable-stayed bridge in step 1 is as follows:

[0012] Several cameras are installed at certain locations on the cable-stayed bridge. These cameras can capture images of all lanes on the cable-stayed bridge deck. A dynamic weighing system is installed on all lanes of the cable-stayed bridge. The outputs of the cameras and the dynamic weighing system are identified using image recognition and machine learning methods to obtain the input load information of all lanes of the cable-stayed bridge, including vehicle type and axle load. The input load information is then used to construct a vehicle load model.

[0013] Accelerometers are installed at the bottom plate inside the steel box girder of the cable-stayed bridge, and strain sensors are installed at the top plate inside the steel box girder. The output response information of the cable-stayed bridge, including acceleration and stress, is obtained through dynamic sampling, and the output response information is cleaned. The frequency of dynamic sampling is more than 5 times the upper limit of the structural analysis frequency.

[0014] Furthermore, the structural characteristics of the cable-stayed bridge include, but are not limited to, structural frequency, mode shape, and structural stiffness.

[0015] Furthermore, the formula for the probability interval distribution described in step 2 is as follows:

[0016] P(X=x i ) = p i i = 1, 2, 3…n

[0017] In the formula, P() is the probability distribution function, X is the structural characteristic value of the cable-stayed bridge, and x i For the values ​​of the eigenvalue variable, p i denoted as , where i is the probability value corresponding to the value of the eigenvalue variable, and n is the total number of values ​​of the eigenvalue variable.

[0018] Furthermore, step 3, which involves establishing a digital model of the cable-stayed bridge using a spatial discretization method, includes:

[0019] The cable-stayed bridge's cable structure was simulated using truss elements, the steel box girder structure using spatial shell elements, the concrete bridge tower structure using solid elements, and the steel bridge tower structure using spatial shell elements.

[0020] Furthermore, the parameters of the digital model described in step 5 are determined through parameter sensitivity analysis.

[0021] Furthermore, the method in step 6 is as follows:

[0022] The vehicle load model is added to the numerical twin model described in step 5. The dynamic structural response under vehicle action is obtained by transient dynamic time history analysis of moving load, and the stress time history of the steel box girder of the cable-stayed bridge is analyzed.

[0023] Based on the stress time history calculation results of the steel box girder, the stress spectrum at different structural detail locations of the steel box girder is analyzed using the rainflow counting method. The fatigue performance evaluation results of the steel box girder are obtained based on the stress-life curves at the structural detail locations.

[0024] Beneficial effects: Compared with the prior art, the technical solution of the present invention has the following beneficial technical effects:

[0025] This invention proposes a method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on monitoring data from physical models and digital models. Compared with traditional health monitoring methods, the integration of a numerical twin model allows for the acquisition of detailed fatigue stress spectra at all locations across the entire bridge. This enables a comprehensive assessment of the structure's service performance from both overall structural stiffness and local performance perspectives, avoiding the problem of failing to detect performance degradation locations due to insufficient measurement points. It improves the accuracy and efficiency of stress analysis for steel box girders, providing strong technical support for rapid performance evaluation and defect identification of steel box girders in cable-stayed bridges. Compared with traditional finite element numerical analysis methods, this invention allows for more accurate numerical simulation analysis by combining real data from physical models. By accurately acquiring vehicle load input information and dynamically adjusting the numerical model based on monitored bridge responses, a digital twin model system is formed. This makes the material, boundary, and stiffness parameters of the numerical model closer to the actual bridge condition. By reconstructing the evaluation mathematical model of the cable-stayed bridge through the numerical twin model, the correlation between numerical analysis results and actual structural performance is improved, providing effective data support for the comprehensive and systematic evaluation of long-span cable-stayed bridges. Attached Figure Description

[0026] Figure 1 This is a flowchart of a method for evaluating the fatigue performance of a cable-stayed bridge steel box girder based on digital twins, according to one embodiment.

[0027] Figure 2 It is a digital model diagram constructed under one embodiment;

[0028] Figure 3 This is a structural array diagram of a digital model of a bridge in one embodiment;

[0029] Figure 4 This is a stress loading history diagram of a steel box girder under one embodiment. Detailed Implementation

[0030] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0031] The present invention describes a method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins, with reference to... Figure 1 Specifically, it includes the following steps:

[0032] (1) Install cameras and dynamic weighing systems on the bridge to obtain the bridge's input information, including vehicle type and axle load, and build a load model library. Install acceleration sensors and strain sensors on the bridge to obtain the bridge's output information. Remove noise interference through data cleaning and build a physical information model library of the cable-stayed bridge structure based on the dynamic monitoring data.

[0033] (2) A digital model of the cable-stayed bridge is established based on the spatial discretization method. The key components such as bridge towers, cables, and main beams are modeled using different methods according to different structural forms. Twin points are set up at the same locations as the bridge response monitoring points in step (1) as the output positions for later calculation and analysis of the digital model.

[0034] (3) Based on the input load information and output response information of the physical information model in step (1), the key structural characteristics of the cable-stayed bridge, such as the structural mode shape, frequency, and structural stiffness, are identified to provide a basis for updating the digital model in step (2). The probability interval distribution of the characteristic values ​​of the above key structural characteristics is obtained through long-term signal tracking analysis.

[0035] P(X=x i ) = p i i = 1, 2, 3…n

[0036] In the formula, P is the probability distribution function, X is the structural characteristic value, and x is the structural feature value. i For the values ​​of the eigenvalue variable, p i This represents the corresponding probability value.

[0037] (4) Based on the digital model in step (2), key parameters such as the frequency, mode shape, and structural stiffness of the structure are obtained through modal analysis and static analysis. Then, the digital model is adjusted and corrected by combining the feature values ​​identified by the physical information model in step (3). A multi-objective optimization method is used to correct the digital model. The adjustment parameters include the elastic modulus of the material and boundary constraints. This ensures that the difference between the physical model response at the twin point in step (2) and the digital model response in step (3), as well as the difference between the structural characteristic feature values ​​of the physical model and the digital model, are within the allowable range, thereby realizing the construction of the numerical twin model.

[0038] (5) The stress time history of the steel box girder of the cable-stayed bridge is analyzed based on the input load information of the physical information model in step (1) and the numerical twin model in step (4), and the stress monitoring data in the physical information model is integrated to evaluate the stress spectrum and fatigue performance of the steel box girder.

[0039] As a preferred embodiment of the present invention, in step (1), the camera is installed above the bridge deck, and the camera angle is adjusted so that it can cover all lanes of the bridge deck. The dynamic weighing system is installed in all lanes of the bridge deck and identifies the vehicle type and axle load of all lanes through image recognition and machine learning methods.

[0040] As a preferred embodiment of the present invention, in step (1), the acceleration sensor is installed at the bottom plate inside the steel box girder, the strain sensor is installed at the top plate inside the steel box girder, the dynamic data sampling frequency needs to be greater than 5 times the upper limit of the structural analysis frequency, and the influence of temperature effect is eliminated by wavelet transform.

[0041] As a preferred embodiment of the present invention, in step (2), the cables are simulated using truss elements, the steel box girder is simulated using space shell elements, the concrete bridge tower is simulated using solid elements, and the steel bridge tower is simulated using space shell elements.

[0042] As a preferred embodiment of the present invention, in step (3), the mode shape and natural frequency of the bridge are identified by Hilbert transform based on the bridge acceleration monitoring signal in step (1), and the structural stiffness of the bridge is identified based on the bridge stress monitoring signal and vehicle load information in step (1).

[0043] As a preferred embodiment of the present invention, in step (4), the natural frequency and stiffness of the structure are obtained through numerical simulation dynamic analysis and static analysis, the parameters for model correction are determined through parameter sensitivity analysis, and the mapping between the physical information model and the digital model is realized by using a neural network, thereby adjusting the parameters of the digital model.

[0044] As a preferred embodiment of the present invention, in step (5), the vehicle load model identified in step (1) is applied to the numerical twin model in step (4), and the dynamic structural response under vehicle action is obtained by transient dynamic time history analysis of moving load. Based on the stress time history calculation results of the steel box girder, the stress spectrum of different structural detail locations of the steel box girder is analyzed by rainflow counting method, and the fatigue performance of the steel box girder is evaluated according to the stress-life curve of the structural detail.

[0045] The following example, using a single-cable-stayed bridge with a central single-cable-plane tower, illustrates the specific process of fatigue performance evaluation of steel box girders for cable-stayed bridges based on digital twins, as described in this invention.

[0046] (1) Installation of monitoring equipment

[0047] Two high-definition cameras are installed at a certain height above the bridge deck using the cable-stayed structure. The camera angles are adjusted to ensure coverage of all lanes, and the camera height is adjusted to ensure the captured images can be used to identify vehicle types. A dynamic weighing system is installed directly beneath each lane of the bridge deck, located at the same longitudinal section of the bridge and visible to the cameras. Accelerometers are installed at 10 sections along the longitudinal direction of the bridge, with one sensor installed on the left and right sides of each section. The sensors are installed inside the bottom plate of the steel box girder. Strain sensors are installed at two sections along the longitudinal direction of the bridge, with four strain sensors installed at each section, located at fatigue detail locations such as the top plate, U-ribs, and diaphragms inside the steel box girder. Data acquisition and transmission equipment is installed inside the box girder.

[0048] (2) Physical Information Model Construction

[0049] The acceleration sampling frequency was set to 30Hz, and the strain sampling frequency was set to 20Hz. The collected data was transmitted to the cloud platform via a 4G network. The monitoring data underwent initial cleaning by identifying outliers and removing trend terms. For the acceleration signal, wavelet transform was used to eliminate the influence of temperature effects, and ensemble empirical mode decomposition was used to eliminate live load effects. Noise interference was removed through data cleaning. For the strain signal, wavelet transform was used to eliminate the influence of temperature effects. Based on images captured by video cameras, image recognition and machine learning methods were used to identify vehicle types in all lanes. Vehicle speed was calculated based on distance, and a matching relationship between vehicle type and axle load was established using a dynamic weighing system to construct a physical information model of the load input and response output of the cable-stayed bridge.

[0050] (3) Digital Model Construction

[0051] The main tower and main girder of the central single-plane cable-stayed bridge are steel structures, simulated using spatial shell elements. The concrete foundation is simulated using solid elements. The studs and connectors of the steel-concrete composite section are simulated using spatial beam elements. The contact interface is simulated using contact relationships. The cables are simulated using truss elements, and the elastic modulus of the cables is corrected using the Ernst formula. The tower base is fixedly constrained. Boolean operations are used to cut the geometric model to ensure that there are corresponding nodes and elements at the bridge response monitoring points, thus realizing the layout of twin points. Finally, element division is performed to establish a digital model of the cable-stayed bridge. (Reference) Figure 2 This is a digital model diagram.

[0052] (4) Bridge feature value identification

[0053] Based on the input load and output response information in the physical information model, key structural characteristics of cable-stayed bridges, such as mode shapes, frequencies, and structural stiffness, are identified. The probability interval distribution of key structural characteristic feature values ​​is obtained through long-term signal tracking analysis. For bridge acceleration monitoring signals, after cleaning and removing load effects, the eigenvalues ​​of each order of natural mode components are obtained using ensemble empirical mode decomposition. Then, the 10 layers of eigenvalues ​​obtained from the decomposition are subjected to Hilbert transform, and combined with intrinsic time-scale decomposition, the bridge's natural frequencies and damping ratios are obtained. The range and probability distribution interval of the natural frequencies are obtained based on a large amount of monitoring data. For bridge stress monitoring signals and vehicle load information, a database relating axle load, vehicle type information, and peak stress is established.

[0054] (5) Construction of numerical twin model

[0055] The subspace method was used to calculate and analyze the digital model, obtaining simulation results of the bridge's array configuration and natural frequencies. Figure 3This paper presents the structural alignment diagram of the bridge's digital model. Parameters such as elastic modulus, component thickness, area, and boundary constraint stiffness are selected. Through parameter sensitivity analysis, parameters with significant impact on the bridge's natural frequencies are identified and chosen as parameters for model correction. A five-layer neural network structure is established to frame the model correction parameters and frequency alignment parameters from the physical information model library. Deep learning is used to adjust and optimize the values ​​of parameters such as the elastic modulus, ultimately completing the construction of a numerical twin model and reducing the dynamic response error between the digital and physical models. Then, based on stress data from monitoring signals and simulation results of twin points in the numerical model, the neural network model is used to optimize and adjust the local stiffness parameters of the steel box girder, reducing the static response error between the digital and physical models.

[0056] (6) Stress assessment of steel box girder

[0057] The load information model obtained from the physical information model is used as input to the parameter-adjusted numerical twin model. By applying load steps, the stress-time history of the cable-stayed bridge steel box girder when vehicles pass through the points of interest is obtained, with reference to... Figure 4 This paper presents the longitudinal and transverse stress variations of the top plate of a steel box girder when a heavy vehicle passes over it. The stress amplitude and frequency of the steel box girder are statistically obtained using the rainflow counting method, and then its equivalent stress amplitude is obtained using the Miner criterion. The numerical twin model is continuously adjusted by fusing the numerical calculation results with monitoring data in the physical information model to ensure the accuracy of its twin point results. Statistical analysis is performed on the stress levels of fatigue details throughout the bridge. Finally, the fatigue performance of different detailed locations of the steel box girder is evaluated based on the stress-life curves specified in the standards.

[0058] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A method for evaluating the fatigue performance of steel box girders in cable-stayed bridges based on digital twins, characterized in that, Specifically, the steps include the following: Step 1: Acquire and process the dynamic monitoring data of the cable-stayed bridge, and use the processed dynamic monitoring data to construct a physical information model of the cable-stayed bridge structure; the dynamic monitoring data includes input load information and output response information. Step 2: Based on the input load information and output response information of the physical information model described in Step 1, the structural characteristics of the cable-stayed bridge are identified, and signal tracking analysis is performed on the physical information model to obtain the structural characteristic feature values ​​and probability interval distribution of the cable-stayed bridge in the physical information model. Step 3: Establish a digital model of the cable-stayed bridge using the spatial discretization method, and deploy twin points at the response monitoring locations of the digital model of the cable-stayed bridge; Step 4: Based on the digital model described in Step 3, obtain the output response information, structural characteristics, and structural characteristic feature values ​​of the cable-stayed bridge from the digital model through modal analysis and static analysis. Step 5: Based on the output response information and structural characteristic values ​​of the physical information model and the output response information and structural characteristic values ​​of the digital model, construct a mapping between the physical information model and the digital model. Use a multi-objective optimization method to adjust and correct the parameters of the digital model described in Step 3 to realize the construction of the digital twin model. Step 6: Analyze the stress time history of the cable-stayed bridge steel box girder using the input load information of the physical information model described in Step 1 and the digital twin model described in Step 5; evaluate the stress spectrum and fatigue performance of the cable-stayed bridge steel box girder using the dynamic monitoring data described in Step 1, and obtain the evaluation results. The structural characteristics of the cable-stayed bridge include structural frequency, mode shape, and structural stiffness.

2. The fatigue performance evaluation method for cable-stayed bridge steel box girders based on digital twins according to claim 1, characterized in that, Step 1 describes the acquisition and processing of dynamic monitoring data of the cable-stayed bridge. The specific method is as follows: Several cameras are set up at preset locations on the cable-stayed bridge. The cameras are used to capture images of all lanes on the cable-stayed bridge deck. A dynamic weighing system is set up in all lanes of the cable-stayed bridge. The outputs of the cameras and the dynamic weighing system are identified using image recognition and machine learning methods to obtain the input load information of all lanes of the cable-stayed bridge, including vehicle type and axle load. The input load information is then used to construct a vehicle load model. Accelerometers are installed at the bottom plate inside the steel box girder of the cable-stayed bridge, and strain sensors are installed at the top plate inside the steel box girder. The output response information of the cable-stayed bridge, including acceleration and stress, is obtained through dynamic sampling, and the output response information is cleaned. The frequency of dynamic sampling is more than 5 times the upper limit of the structural analysis frequency.

3. The fatigue performance evaluation method for cable-stayed bridge steel box girders based on digital twins according to claim 1, characterized in that, The formula for the probability interval distribution mentioned in step 2 is as follows: P(X=x i )=p i ,i=1,2,3…n In the formula, P() is the probability distribution function, X is the structural characteristic value of the cable-stayed bridge, and x i For the values ​​of the eigenvalue variable, p i denoted as , where i is the probability value corresponding to the value of the eigenvalue variable, and n is the total number of values ​​of the eigenvalue variable.

4. The fatigue performance evaluation method for steel box girders of cable-stayed bridges based on digital twins according to claim 1, characterized in that, Step 3 describes the establishment of a digital model of a cable-stayed bridge using a spatial discretization method, which includes: simulating the cable structure of the cable-stayed bridge using truss elements, simulating the steel box girder structure using spatial shell elements, simulating the concrete bridge tower structure using solid elements, and simulating the steel bridge tower structure using spatial shell elements.

5. The fatigue performance evaluation method for steel box girders of cable-stayed bridges based on digital twins according to claim 1, characterized in that, The parameters of the digital model described in step 5 are determined through parameter sensitivity analysis, including elastic modulus, component thickness, area, and boundary constraint stiffness.

6. The fatigue performance evaluation method for steel box girders of cable-stayed bridges based on digital twins according to claim 2, characterized in that, The method in step 6 is as follows: the vehicle load model is added to the digital twin model described in step 5, the dynamic structural response under vehicle action is obtained by transient dynamic time history analysis of moving load, and the stress time history of the cable-stayed bridge steel box girder is analyzed; based on the stress time history calculation results of the steel box girder, the stress spectrum of different structural detail locations of the steel box girder is analyzed by rainflow counting method, and the fatigue performance evaluation results of the steel box girder are obtained according to the stress-life curve of the structural detail location.

Images