A stress analysis method and system for bridge jacking construction
By combining a multi-source heterogeneous sensor network and a physical information neural network model, the shortcomings of multi-source data fusion analysis in bridge jacking construction are solved, achieving accurate reflection of the dynamic behavior of the bridge structure and improving the physical reliability of the model.
Patent Information
- Application Number
- CN202611016752.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-09
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2046-07-09
AI Technical Summary
The lack of multi-source data fusion analysis in existing bridge launching construction methods leads to misjudgments of bridge mechanical behavior and unreasonable parameter identification. Existing parameter identification algorithms are black boxes and cannot guarantee the physical credibility of the model and the accuracy of simulation results.
A multi-source heterogeneous sensor network is used for data acquisition to establish a three-dimensional digital twin model of bridge jacking construction. A physical information neural network model is used for dynamic correction and state synchronization. A loss function is constructed to modify the model in a way that combines data-driven and physical-driven approaches to ensure that the prediction results conform to physical laws.
It achieves an accurate reflection of the dynamic behavior of the structure during the bridge jacking construction process, improves the physical credibility of the model and the reliability of the simulation results, and avoids misjudgment and unreasonable parameters.
Smart Images

Figure CN122597731B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of data processing technology, and in particular to a stress analysis method and system for bridge jacking construction. Background Technology
[0002] Bridge jacking construction is an advanced construction technique that involves setting up a prefabrication yard behind the abutments, casting or assembling the bridge beams in sections, and then using jacking equipment such as horizontal jacks to gradually push them into place along the bridge axis. This method has advantages such as not affecting traffic under the bridge, smooth construction, and controllable construction period, and is widely used in the construction of continuous beams with constant cross-section, continuous beams with variable cross-section, and some arch bridges.
[0003] In the traditional field of bridge health monitoring and construction control, there are schemes for dynamically correcting three-dimensional digital twin models. However, existing correction methods often treat a single type of monitoring data in isolation. For example, they might only use displacement data to correct the model's stiffness, or only use strain data to correct the material's constitutive model. However, the mechanical behavior of a bridge is the result of strong coupling between multiple physical fields. For instance, the non-uniform temperature field in the beam caused by sunlight directly leads to considerable lateral and longitudinal displacements, which in turn alter the distribution of support reactions and internal forces within the beam. Correcting stiffness solely based on displacement data is tantamount to incorrectly attributing temperature effects to stiffness issues, leading to a misjudgment of the structure's true mechanical state. Current technologies lack the ability to fuse and analyze multi-source data and identify dominant physical effects. At the same time, many existing parameter identification algorithms are a black box for engineers. In order to fit the data, the algorithm may produce a set of physically unreasonable or inexplicable parameters. For example, it may calculate a negative stiffness value or an elastic modulus that far exceeds the theoretical limit of the material. Engineers cannot determine whether this result is due to improper model simplification, sensor failure, or the limitations of the algorithm itself. As a result, they have serious doubts about the physical credibility of the corrected model and the guiding significance of subsequent simulation results, and dare not easily adopt it. Summary of the Invention
[0004] To address the aforementioned technical problems, this invention provides a stress analysis method and system for bridge jacking construction, which solves the problems existing in the prior art.
[0005] This invention provides a stress analysis method for bridge jacking construction, comprising the following steps: S1: Establish a three-dimensional digital twin model of the bridge jacking construction; S2: Deploy a multi-source heterogeneous sensor network and collect data; S3: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the collected data; Specifically, S3 is: S3.1: Perform data preprocessing on the collected data; S3.2: Construct a physical information neural network model; S3.3: Define the loss function of the physical information neural network model and train the loss function of the physical information neural network model; The expression for the loss function of the physical information neural network model is as follows: ; In the formula, The loss function of the physical information neural network model is... , These are the weighting coefficients. This refers to the data loss term in the loss function of the physical information neural network model. The physical loss term is the loss function of the physical information neural network model. S3.4: Input the preprocessed collected data into the physical information neural network model to obtain the model prediction results; S3.5: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the prediction results of the model; S4: Perform stress analysis on bridge jacking construction based on the three-dimensional digital twin model after dynamic correction and state synchronization.
[0006] Preferably, in step S3.3, for the i-th sensor measurement point, the expression for the data loss term of the loss function of the physical information neural network model is: ; In the formula, This refers to the data loss term in the loss function of the physical information neural network model. This represents the predicted value of the physical information neural network model for the measurement point of the i-th sensor. This represents the actual measurement value collected by the i-th sensor; The physical loss term includes the residuals of the force balance equations, boundary conditions, and temperature-displacement coupling; its expression is: ; In the formula, For the physical loss term, For the residuals of the force balance equation, For boundary condition residuals, This represents the temperature-displacement coupling residual.
[0007] Preferably, the expression for the residual loss of the equilibrium equation is: ; In the formula, The residuals of the force balance equations; The expression for the boundary condition residual is: ; In the formula, For boundary condition residuals, The boundary displacement predicted by the physical information neural network model for the j-th measurement point. This represents the displacement measurement value at the j-th measuring point; The expression for the temperature-displacement coupling residual is: ; In the formula, The residual is the temperature-displacement coupling, and α is the coefficient of thermal expansion of the construction material. Let k be the temperature change at the k-th measuring point. The thermal strain predicted by the physical information neural network model for the k-th measurement point.
[0008] Preferably, in step S1, the three-dimensional digital twin model is established using parametric BIM modeling, and the three-dimensional digital twin model includes the main structure, support system, jacking system, and auxiliary mechanisms.
[0009] Preferably, the main structure includes a main beam, a front guide beam, and a rear guide beam; the support system includes all permanent piers, temporary piers, temporary supports, ground supports, and a jacking platform; and the jacking system includes jacking jacks, an oil pump, a main control console, a slide rail, a slider, and a connection device between the guide beam and the main beam.
[0010] Preferably, in step S2, the multi-source heterogeneous sensor network includes: fiber optic strain sensor, magnetostrictive displacement sensor, spoke-type pressure sensor, laser rangefinder, or wire-type displacement sensor.
[0011] Preferably, the data preprocessing of the collected data specifically includes cleaning, aligning, and fusing the collected data.
[0012] Preferably, in step S3.2, the physical information neural network model is based on a deep neural network model and includes an input layer, a hidden layer, and an output layer.
[0013] Preferably, in S3.5, the dynamic correction and state synchronization of the three-dimensional digital twin model based on the model prediction results specifically includes geometric state synchronization and internal force state synchronization of the three-dimensional digital twin model; the geometric state synchronization specifically involves applying the full-field displacement field output by the physical information neural network model as a forced boundary condition or initial displacement field to the three-dimensional digital twin model, causing the three-dimensional digital twin model to deform; the internal force state synchronization specifically involves replacing the stress field calculated by the three-dimensional digital twin model with the full-field stress field output by the physical information neural network model.
[0014] According to another aspect of the present invention, a stress analysis system for bridge jacking construction is provided, the system employing the aforementioned stress analysis method for bridge jacking construction, the system comprising: The 3D digital twin model creation module is used to create a 3D digital twin model of bridge jacking construction. The data acquisition module is used to deploy a multi-source heterogeneous sensor network and acquire data. The 3D digital twin model update module is used to dynamically correct and synchronize the state of the 3D digital twin model based on the collected data. The stress analysis module is used to perform stress analysis on bridge jacking construction based on a dynamically corrected and state-synchronized three-dimensional digital twin model.
[0015] The embodiments of the present invention have the following technical effects: This invention, when using a neural network model to dynamically correct and synchronize the state of a 3D digital twin model, sets the loss function of the neural network model to have two parameters: data-driven and physics-driven. This allows the black-box neural network to meet the requirements of providing prediction results that are both like real measurements and consistent with physical principles in the engineering field. Furthermore, it performs dual synchronization of geometry and internal forces on the 3D digital twin model. Therefore, the 3D digital twin model is no longer a model based on pure theory and ideal assumptions, but rather a model whose current state is informed by real-time data and whose evolution is governed by physical laws. This allows for a more accurate reflection of the dynamic behavior of the actual structure during bridge jacking construction. Attached Figure Description
[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0017] Figure 1This is a flowchart of a stress analysis method for bridge jacking construction provided by an embodiment of the present invention; Figure 2 This is a flowchart of the dynamic correction and state synchronization of the three-dimensional digital twin model based on the collected data provided in the embodiments of the present invention. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0019] Example 1, Figure 1 A flowchart of a stress analysis method for bridge jacking construction is shown, such as... Figure 1 As shown, a stress analysis method for bridge jacking construction includes the following steps: S1: Establish a three-dimensional digital twin model of the bridge jacking construction; In this step, parametric BIM modeling is used to establish the three-dimensional digital twin model. The three-dimensional digital twin model includes the main structure, support system, jacking system, and auxiliary mechanisms. The main structure includes the main beam, front guide beam, and rear guide beam. The three-dimensional digital twin model needs to accurately represent the cross-sectional shape, beam height variation, cross slope, and pre-camber setting of the beams. The support system includes all permanent piers, temporary piers, temporary supports, ground supports, jacking platforms, etc. The support system not only needs to establish the geometric model of the piers and foundations, but also needs to accurately simulate the type, size, and spatial position of the supports. The jacking system includes jacking jacks, oil pumps, main control console, slides, sliders, and connection devices between the guide beams and the main beams. The auxiliary structures include anti-collision facilities, railings, inspection vehicle tracks, etc., to ensure the integrity of the model and the scalability for future operation and maintenance.
[0020] Each component in the three-dimensional digital twin model is assigned realistic physical property parameters, including: material density, elastic modulus, and Poisson's ratio of the beam and guide beam; constitutive relation model of concrete and steel; stiffness coefficient of each support and temporary pier; and preset friction coefficient between the slide and the slider.
[0021] Based on the actual construction conditions, the boundary conditions of the three-dimensional digital twin model are defined. For example, the bottom of the permanent pier is set as a fixed constraint, and the bottom of the temporary pier is set as a hinged or elastic constraint according to the geological conditions. The initial state of construction is set, such as the initial position of the prefabricated platform and the number of beam segments already installed.
[0022] In the three-dimensional digital twin model, the virtual positions and numbers of all planned sensors are pre-planned and calibrated, establishing a one-to-one correspondence between physical sensors and digital model data points. Thus, a three-dimensional digital twin model with complete physical information and sensor interfaces is constructed.
[0023] S2: Deploy a multi-source heterogeneous sensor network and collect data; This step involves constructing a comprehensive, multi-source, heterogeneous sensor network capable of fully, accurately, and in real-time sensing various key physical parameters of the bridge structure throughout the entire construction process. Through the optimized configuration and integration of various types and principles of sensors, this network transforms the stress state of the bridge structure into quantifiable data, providing a solid foundation of raw data for subsequent 3D digital twin model correction and dynamic simulation analysis.
[0024] Based on the analysis of the structural mechanical properties of the bridge, monitoring targets that play a decisive role in construction safety were identified. These targets mainly include: the internal strain state of key sections, the overall displacement and deflection of the beam, the real-time reaction force at each support, the output force and displacement of the jacking equipment, and the ambient temperature field. For each type of monitoring target, the optimal sensor in terms of principle, accuracy, and adaptability was selected.
[0025] To accurately monitor the internal strain of key control sections of the beam, fiber optic grating (FBG) strain sensors were selected. These sensors offer significant advantages, including resistance to electromagnetic interference, quasi-distributed measurement capabilities, high long-term stability, and intrinsic safety, making them particularly suitable for complex and harsh construction site environments. During sensor placement, it is crucial to ensure the sensor's axis aligns with the direction of the principal stress being measured. Based on the stress distribution gradient of the section, multiple points are densely arranged on the inner and outer sides of the top, bottom, and web plates to obtain a complete cross-sectional stress state. To acquire the overall vertical deflection and horizontal displacement of the beam, magnetostrictive displacement sensors are used for data acquisition. These sensors are installed on the tops of all permanent and temporary piers to directly and accurately measure the relative displacement between the beam and the pier tops. Their advantages include being unaffected by weather and providing purely relative measurement with micrometer-level accuracy. Differential positioning with a distant fixed reference station is achieved via a global navigation satellite system to calculate the absolute coordinates of the beam in the geodetic coordinate system, thereby monitoring its overall performance. The beam's alignment and axis are misaligned. To directly measure the vertical support reaction force borne by each temporary pier and key permanent pier, a spoke-type pressure sensor is used. This spoke-type pressure sensor can be customized into a standard-sized pad and placed directly between the beam and the pier top. This achieves high-precision, dynamic measurement of concentrated reaction force without changing the original structural force transmission path. To achieve real-time monitoring of the output force of the jacking jack, a hydraulic pressure sensor is connected in series in its hydraulic circuit. Based on Pascal's principle, the hydraulic pressure sensor can accurately calculate the jacking force by measuring the hydraulic circuit pressure in real time and combining it with the effective area of the jack's piston. At the same time, a laser rangefinder or a wire-type displacement sensor is installed on the jacking platform or reaction pier to directly measure the jack's stroke. This stroke data directly corresponds to the jacking distance of the beam. To assess the influence of ambient temperature on structural deformation, platinum resistance (PT100) temperature sensors are deployed at the construction site and key parts of the beam to capture the ambient temperature and the temperature gradient formed within the beam cross-section.
[0026] The physical installation quality of the sensors in the multi-source heterogeneous sensor network directly determines the validity of the data. Therefore, strict installation procedures must be developed for different types of sensors. The fiber optic strain gauge sensors must be firmly adhered to the surface of the reinforcing bars or structural steel at predetermined locations using specialized adhesives or welding before concrete pouring, and then covered with concrete only after protective sleeves and waterproof sealing. The spoke-type pressure sensors require pre-pressure calibration before installation to ensure that their bearing surfaces are in flat and uniform contact with the beam and pier tops, avoiding the introduction of eccentric loads. The deployment plan of the entire sensor network must be closely coordinated with key processes such as civil engineering and steel structure installation in bridge construction, and detailed construction coordination plans must be developed to ensure that all installation work is completed without interfering with the normal construction progress.
[0027] All sensor nodes are connected to one or more distributed Data Acquisition Boxes (DAQs) in the field via shielded cables or wireless transmission modules, following the principle of proximity aggregation, forming a star or hybrid topology. Each DAQ integrates an embedded processor, a high-precision data acquisition card, a signal conditioning module, and necessary network equipment. The power supply system primarily uses mains power, supplemented by solar energy and batteries, and all critical equipment is equipped with uninterruptible power supplies (UPS) to ensure continuous operation in the event of grid fluctuations or unexpected power outages.
[0028] S3: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the collected data; In the traditional field of bridge health monitoring and construction control, there are schemes for dynamically correcting three-dimensional digital twin models. However, existing correction methods often treat a single type of monitoring data in isolation. For example, they might only use displacement data to correct the model's stiffness, or only use strain data to correct the material's constitutive model. However, the mechanical behavior of a bridge is the result of strong coupling between multiple physical fields. For instance, the non-uniform temperature field in the beam caused by sunlight directly leads to considerable lateral and longitudinal displacements, which in turn alter the distribution of support reactions and internal forces within the beam. Correcting stiffness solely based on displacement data is tantamount to incorrectly attributing temperature effects to stiffness issues, leading to a misjudgment of the structure's true mechanical state. Current technologies lack the ability to fuse and analyze multi-source data and identify dominant physical effects. At the same time, many existing parameter identification algorithms are a black box for engineers. In order to fit the data, the algorithm may produce a set of physically unreasonable or inexplicable parameters. For example, it may calculate a negative stiffness value or an elastic modulus that far exceeds the theoretical limit of the material. Engineers cannot determine whether this result is due to improper model simplification, sensor failure, or the limitations of the algorithm itself. As a result, they have serious doubts about the physical credibility of the corrected model and the guiding significance of subsequent simulation results, and dare not easily adopt it.
[0029] Based on this, this embodiment provides a dynamic fusion correction and state synchronization method based on a physical information neural network. The physical control equations describing the mechanical behavior of the bridge are used as prior knowledge and soft constraints and embedded into the training process of the neural network. In this way, the collected data guides and corrects the learning direction of the neural network, while the physical equations act as regularizers and constraints to ensure that the solutions learned by the neural network are physically reasonable, continuous and in accordance with the laws of mechanics.
[0030] Specifically, such as Figure 2 As shown, S3 specifically includes: S3.1: Perform data preprocessing on the collected data; The collected data is cleaned, aligned, and fused. Specifically, the collected data undergoes timestamp synchronization and data quality checks to remove outliers caused by communication interference. Secondly, the data for different physical quantities are normalized.
[0031] S3.2: Construct a physical information neural network model; The physical information neural network model uses a deep neural network framework and includes an input layer, a hidden layer, and an output layer. The input to the input layer is a four-dimensional spatiotemporal coordinate vector generated from preprocessed collected data, expressed as: ; In the formula, Let (x, y, z) be a four-dimensional spatial coordinate vector, where (x, y, z) are spatial rectangular coordinates and t is the time coordinate. The hidden layer consists of L fully connected layers, where the connections between layers are determined by a trainable weight matrix. and bias vector Define, where, from the lth The transformation from layer 1 to layer l is represented as: ; In the formula, σ is the weighted input to the neurons in layer l; σ() is the activation function. Let be the weight matrix of the l-th layer. For the l The output vector of a single layer of neurons. Let l be the bias vector of the l-th layer; Meanwhile, in order to introduce nonlinearity and stabilize training, the Swish function is selected as the activation function in this step; The output layer is used to output the physical quantities of the bridge jacking, including three-dimensional displacement components and strain components.
[0032] S3.3: Define the loss function of the physical information neural network model and train the loss function of the physical information neural network model; In this step, the training of the physical information neural network model is constrained by both data loss and physical loss, ensuring that the model's predictions match real measurements and obey physical laws, thereby improving the model's reliability in the jacking construction scenario. Specifically, the expression for the loss function of the physical information neural network model is: ; In the formula, The loss function of the physical information neural network model is... , These are the weighting coefficients. This refers to the data loss term in the loss function of the physical information neural network model. The physical loss term is the loss function of the physical information neural network model. The data loss term is used to measure the difference between the predicted value of the physical information neural network model and the measured value of the sensor at the sensor measurement point location, so that the model outputs a prediction result that fits the real world; for the i-th sensor measurement point, the expression for the data loss term of the loss function of the physical information neural network model is: ; In the formula, This refers to the data loss term in the loss function of the physical information neural network model. This represents the predicted value of the physical information neural network model for the measurement point of the i-th sensor. This represents the actual measurement value collected by the i-th sensor; The physical loss term of the loss function of the physical information neural network model is derived from the three-dimensional digital twin model, defining several sub-loss terms to force the prediction results of the physical information neural network model to satisfy physical laws such as mechanics and thermodynamics. Specifically, the physical loss term includes the residuals of the force balance equations, boundary conditions, and temperature-displacement coupling; its expression is: ; In the formula, For the physical loss term, For the residuals of the force balance equation, For boundary condition residuals, This represents the temperature-displacement coupling residual.
[0033] In geotechnical engineering, structural engineering, and other fields, force equilibrium is one of the most fundamental physical laws. Neural networks predict displacement fields. Using geometric equations and constitutive models, the displacement field can be converted into a stress field. Then, the stress field is substituted into the force equilibrium equation. For use in the loss function, the deviation of the left side of the force equilibrium equation from 0 is taken as the residual, squared, and summed to obtain the residual loss of the equilibrium equation, expressed as: ; In the formula, This represents the residual of the force balance equation.
[0034] The boundaries of structures or soil / rock masses often have displacement or force constraints. These boundary conditions are inherent constraints of the physical problem. Assuming that at known boundaries such as supports, the measured displacement values are... The boundary displacement predicted by the physical information neural network model is The boundary condition residual is defined as the sum of the squared differences between the predicted displacement and the measured / constrained displacement, expressed as: ; In the formula, For boundary condition residuals, The boundary displacement predicted by the physical information neural network model for the j-th measurement point. This represents the displacement measurement value of the j-th measuring point.
[0035] In large-volume concrete pouring, there is coupling between the temperature field and the displacement field. Temperature changes cause thermal expansion and contraction of materials, leading to deformation; conversely, displacement can also affect the temperature distribution. This embodiment focuses on the unidirectional coupling of temperature and displacement, using thermoelasticity to describe this relationship and establishing a temperature-displacement coupling residual, the expression of which is: ; In the formula, The residual is the temperature-displacement coupling, and α is the coefficient of thermal expansion of the construction material. Let k be the temperature change at the k-th measuring point. The thermal strain predicted by the physical information neural network model for the k-th measurement point.
[0036] During the training of the physical information neural network model, the loss function is minimized through the backpropagation algorithm, allowing the model to learn the data characteristics of real measurements while strictly adhering to physical laws such as force balance, boundary conditions, and temperature-displacement coupling. This results in higher reliability and generalization ability in the prediction during the construction process. This embodiment combines data-driven and physical constraints to enable the black-box neural network to meet the requirements of both real measurement and physical common sense in the engineering field.
[0037] S3.4: Input the preprocessed collected data into the physical information neural network model to obtain the model prediction results; In this step, the model prediction results are full-field displacement and stress fields that integrate the collected data and physical mechanisms, which drive and update the three-dimensional digital twin model.
[0038] S3.5: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the prediction results of the model; Specifically, in S3.5, the dynamic correction and state synchronization of the three-dimensional digital twin model based on the model prediction results includes geometric state synchronization and internal force state synchronization of the three-dimensional digital twin model. The geometric state synchronization specifically involves applying the full-field displacement field output by the physical information neural network model as a forced boundary condition or initial displacement field to the three-dimensional digital twin model, causing the three-dimensional digital twin model to deform. The internal force state synchronization specifically involves replacing the stress field calculated by the three-dimensional digital twin model with the full-field stress field output by the physical information neural network model.
[0039] Through the dual synchronization of geometry and internal forces, the three-dimensional digital twin model is no longer a model based on pure theory and ideal assumptions, but a model whose current status is informed by real-time data and whose evolution is governed by physical laws. This allows for a more accurate reflection of the dynamic behavior of the actual structure during the bridge jacking construction process.
[0040] S4: Perform stress analysis on bridge jacking construction based on the three-dimensional digital twin model after dynamic correction and state synchronization.
[0041] Example 2: The present invention also provides a stress analysis system for bridge jacking construction. The system employs a stress analysis method for bridge jacking construction as described in Example 1. The system includes: The 3D digital twin model creation module is used to create a 3D digital twin model of bridge jacking construction. The data acquisition module is used to deploy a multi-source heterogeneous sensor network and acquire data. The 3D digital twin model update module is used to dynamically correct and synchronize the state of the 3D digital twin model based on the collected data. The stress analysis module is used to perform stress analysis on bridge jacking construction based on a dynamically corrected and state-synchronized three-dimensional digital twin model.
[0042] Example 3: The present invention also provides an electronic device, including one or more processors and a memory.
[0043] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.
[0044] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and a processor may execute the program instructions to implement a stress analysis method for bridge jacking construction as described above in any embodiment of this application, and / or other desired functions. Various contents such as initial extrinsic parameters and thresholds may also be stored in the computer-readable storage medium.
[0045] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.
Claims
1. A stress analysis method for bridge jacking construction, characterized in that, Includes the following steps: S1: Establish a three-dimensional digital twin model of the bridge jacking construction; S2: Deploy a multi-source heterogeneous sensor network and collect data; S3: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the collected data; Specifically, S3 is: S3.1: Perform data preprocessing on the collected data; S3.2: Construct a physical information neural network model; S3.3: Define the loss function of the physical information neural network model and train the loss function of the physical information neural network model; The expression for the loss function of the physical information neural network model is as follows: ; In the formula, The loss function of the physical information neural network model is... , These are the weighting coefficients. This refers to the data loss term in the loss function of the physical information neural network model. This refers to the physical loss term of the loss function of the physical information neural network model; in S3.3, for the i-th sensor measurement point, the expression for the data loss term of the loss function of the physical information neural network model is: ; In the formula, This refers to the data loss term in the loss function of the physical information neural network model. This represents the predicted value of the physical information neural network model for the measurement point of the i-th sensor. This represents the actual measurement value collected by the i-th sensor; The physical loss term includes the residuals of the force balance equations, boundary conditions, and temperature-displacement coupling; its expression is: ; In the formula, For the physical loss term, For the residuals of the force balance equation, For boundary condition residuals, The residual is the temperature-displacement coupling residual; the expression for the residual loss of the equilibrium equation is: ; In the formula, The residuals of the force balance equations; The expression for the boundary condition residual is: ; In the formula, For boundary condition residuals, The boundary displacement predicted by the physical information neural network model for the j-th measurement point. This represents the displacement measurement value at the j-th measuring point; The expression for the temperature-displacement coupling residual is: ; In the formula, The residual is the temperature-displacement coupling, and α is the coefficient of thermal expansion of the construction material. Let k be the temperature change at the k-th measuring point. The thermal strain predicted by the physical information neural network model for the k-th measurement point; S3.4: Input the preprocessed collected data into the physical information neural network model to obtain the model prediction results; S3.5: Dynamically correct and synchronize the state of the three-dimensional digital twin model based on the prediction results of the model; S4: Perform stress analysis on bridge jacking construction based on the three-dimensional digital twin model after dynamic correction and state synchronization.
2. The stress analysis method for bridge jacking construction according to claim 1, characterized in that: In step S1, the three-dimensional digital twin model is established using parametric BIM modeling. The three-dimensional digital twin model includes the main structure, support system, jacking system, and auxiliary mechanisms.
3. The stress analysis method for bridge jacking construction according to claim 2, characterized in that: The main structure includes a main beam, a front guide beam, and a rear guide beam. The support system includes all permanent piers, temporary piers, temporary supports, ground supports, and a jacking platform. The jacking system includes jacking jacks, oil pumps, a main control console, slide rails, sliders, and a connection device between the guide beam and the main beam.
4. The stress analysis method for bridge jacking construction according to claim 2, characterized in that: In S2, the multi-source heterogeneous sensor network includes: fiber optic strain sensor, magnetostrictive displacement sensor, spoke-type pressure sensor, laser rangefinder, or wire-type displacement sensor.
5. The stress analysis method for bridge jacking construction according to claim 1, characterized in that: The data preprocessing of the collected data specifically includes cleaning, aligning, and fusing the collected data.
6. The stress analysis method for bridge jacking construction according to claim 1, characterized in that: In S3.2, the physical information neural network model is based on a deep neural network model and includes an input layer, a hidden layer, and an output layer.
7. The stress analysis method for bridge jacking construction according to claim 1, characterized in that: In step S3.5, the dynamic correction and state synchronization of the three-dimensional digital twin model based on the model prediction results specifically includes geometric state synchronization and internal force state synchronization of the three-dimensional digital twin model. The geometric state synchronization specifically involves applying the full-field displacement field output by the physical information neural network model as a forced boundary condition or initial displacement field to the three-dimensional digital twin model, causing the three-dimensional digital twin model to deform. The internal force state synchronization specifically involves replacing the stress field calculated by the three-dimensional digital twin model with the full-field stress field output by the physical information neural network model.
8. A stress analysis system for bridge jacking construction, characterized in that, The system employs a stress analysis method for bridge jacking construction as described in any one of claims 1-7, and the system comprises: The 3D digital twin model creation module is used to create a 3D digital twin model of bridge jacking construction. The data acquisition module is used to deploy a multi-source heterogeneous sensor network and acquire data. The 3D digital twin model update module is used to dynamically correct and synchronize the state of the 3D digital twin model based on the collected data. The stress analysis module is used to perform stress analysis on bridge jacking construction based on a dynamically corrected and state-synchronized three-dimensional digital twin model.
Citation Information
Patent Citations
Bridge early warning method and system based on physical information neural network and machine vision
CN119964334A
Construction method of bridge digital twinborn health monitoring system
CN122286910A
