A steel truss bridge construction linear self-adaptive closed-loop monitoring method and system

Abstract

The application provides a steel truss bridge construction linear self-adaptive closed-loop monitoring method and system, relates to the field of bridge construction, and comprises the following steps: establishing a parameterized BIM model of a steel truss bridge and a finite element mechanical analysis model for simulation in a construction stage; establishing a Kriging surrogate model with a pre-lifting amount as an output and construction parameters as inputs and completing model training; comparing elevation information collected on a construction site with a design elevation to obtain a linear state; inputting real-time construction parameters of the current construction stage into the trained Kriging surrogate model to obtain optimal pre-lifting amounts of the next construction stage; and based on the linear state and the optimal pre-lifting amounts, calculating pre-lifting adjustment values from the BIM model and generating standardized construction monitoring instructions containing the pre-lifting adjustment values and construction precautions. The application realizes intelligent prediction and accurate regulation and control of the construction linear state of the steel truss bridge, and improves construction precision and intelligent level.

Description

Technical Field

[0001] This invention relates to the field of bridge construction monitoring technology, and more specifically, to an adaptive closed-loop monitoring method and system for the construction alignment of steel truss bridges. Background Technology

[0002] Controlling the alignment of steel truss bridges during construction presents unique engineering challenges: First, the structure is assembled from a large number of discrete members, and manufacturing errors and assembly deviations of individual members are easily transmitted and accumulated section by section during construction, seriously affecting the alignment and stress state of the completed bridge; Second, the determination of the pre-lifting amount is affected by multiple factors such as temperature, humidity, and load, and there are strong nonlinear and time-varying effects among these factors. Traditional empirical formulas or simple linear extrapolation methods cannot accurately characterize this complex relationship.

[0003] To address the aforementioned issues, existing technologies primarily employ the following solutions: First, construction monitoring systems based on Building Information Modeling (BIM) are used, but their functions are largely limited to static visualization and post-construction recording. Monitoring data, analysis models, and construction instructions are disconnected, and adjustments rely on human experience, making real-time dynamic response difficult. Second, purely data-driven methods (such as neural networks) are introduced for pre-lift prediction. However, these models lack physical constraints, resulting in low prediction reliability when extrapolating construction conditions or when conditions change. Furthermore, they cannot quantify prediction uncertainty, failing to meet the decision-making needs of high-risk construction scenarios. Third, while traditional physical models such as finite element methods can accurately simulate structural behavior, they struggle to integrate real-time monitoring data to dynamically correct deviations, leading to a lag between analysis results and actual construction conditions. Therefore, existing technologies have not yet solved the core requirement of high-precision real-time prediction and adaptive closed-loop control in steel truss bridge construction. An integrated intelligent linear monitoring system capable of deeply integrating physical laws and real-time data, applicable to the entire construction process of this type of structure, is urgently needed. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art or related technologies, and discloses an adaptive closed-loop monitoring method and system for the construction alignment of steel truss bridges, which realizes high-precision real-time prediction and adaptive closed-loop control in the construction of steel truss bridges.

[0005] The first aspect of this invention discloses an adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge, comprising: S1, model establishment: establishing a parametric BIM model integrating structural geometric attributes and information on the construction stage, and simultaneously establishing a finite element mechanical analysis model for simulating the time-varying structural system at each construction stage; S2, surrogate model training: based on the finite element mechanical analysis model, performing multivariate parametric mechanical simulation to obtain an initial high-fidelity training sample set of input parameters and output pre-lift; based on a Gaussian process regression framework, establishing a Kriging surrogate model with pre-lift as output and construction parameters as input, optimizing the relevant function parameters through maximum likelihood estimation, and obtaining a trained Kriging surrogate model that meets engineering accuracy requirements; S3, data acquisition: collecting real-time construction parameters at the construction site and writing them into a database, wherein the construction parameters include: component ID, measured elevation information of control points. S4, Alignment Deviation Analysis: Compare the measured elevation information of control points with the design elevation, calculate the absolute alignment error, and automatically determine the alignment status of each control point as normal, warning, or exceeding limits based on the preset first and second standard thresholds, and perform differentiated color-coded visual markings in the parametric BIM model; S5, Intelligent Prediction of Pre-lift Amount: Input the real-time construction parameters of the current construction stage into the trained Kriging proxy model, use its built-in related kernel functions for nonlinear mapping, output the predicted pre-lift amount value for the next construction stage, and simultaneously output the prediction variance; S6, Adaptive Generation of Monitoring Instructions: Based on the alignment status, the optimal predicted pre-lift amount value, and the prediction variance, the BIM model dynamically calculates the pre-lift adjustment value, and calls the document automation component to dynamically generate construction monitoring instructions containing the final assembly elevation control parameter table.

[0006] This technical solution uses Building Information Modeling (BIM) as a digital twin carrier, deeply integrating the Kriging proxy model and finite element analysis to achieve real-time perception, intelligent prediction, and adaptive closed-loop control of the structural alignment during construction. By integrating design theoretical values, real-time monitoring data, and pre-lifting optimization results driven by the Kriging proxy model into a parametric BIM model, intelligent calculation, dynamic adjustment, and precise positioning of the structural alignment are achieved in a 3D visualization environment. Based on a closed-loop process of "monitoring-prediction-decision-execution-feedback," construction adjustment instructions can be automatically generated and issued, ensuring precise construction execution. Furthermore, the proxy model is continuously optimized through data feedback, thereby achieving dynamic optimization and autonomous control of the construction alignment, significantly improving the accuracy, efficiency, and intelligence level of large-span steel truss bridge construction.

[0007] According to the adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge disclosed in this invention, preferably, it further includes: S7, adaptive closed-loop feedback and optimization: the construction monitoring command is issued and executed, and after the construction is completed, the monitoring data of the new round of construction stage is collected and fed back to the BIM model to update the alignment status. At the same time, the new round of monitoring data is converted into input-output sample pairs and added to the training sample library of the Kriging proxy model to optimize the model, so as to perform closed-loop adaptive control of the construction alignment.

[0008] According to the adaptive closed-loop monitoring method for the construction alignment of steel truss bridge disclosed in this invention, preferably, in step S1, the BIM model is established using Autodesk Revit software to integrate the geometric shape of bridge components and information on the construction stage to which they belong; the finite element mechanical analysis model is established using Midas / Civil software to simulate the structural system, time-varying material properties, loads and boundary conditions of each construction stage.

[0009] According to the adaptive closed-loop monitoring method for construction alignment of steel truss bridges disclosed in this invention, preferably, step S2 specifically includes: Training sample generation: Based on the finite element mechanical analysis model, parametric analysis is performed. Orthogonal experimental design method is adopted to sample in the parameter space including temperature, humidity and self-weight coefficient to obtain the training sample set of input parameters and output preload, and the data is normalized and preprocessed. Constructing a Kriging surrogate model: Using the Gaussian process regression framework, a constant regression function and a Gaussian correlation function are selected to establish a Kriging surrogate model, and the correlation function parameters are optimized through maximum likelihood estimation; Model validation: The leave-one-out cross-validation method is used to evaluate the model, and the coefficient of determination and root mean square error are calculated to ensure that the prediction accuracy meets the engineering requirements.

[0010] According to the adaptive closed-loop monitoring method for the construction alignment of steel truss bridges disclosed in this invention, preferably, the alignment state determination process in step S4 specifically includes: If the elevation error value of a certain measuring point is less than the first standard threshold, the linear state of that point is determined to be "normal". If the elevation error value is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it is judged as a "warning"; If the elevation error value is greater than the second standard threshold, it is judged as "exceeding the limit"; Linear states are visualized and marked with different colors in both the BIM model and the exported standardized EXCEL file.

[0011] According to the adaptive closed-loop monitoring method for the construction alignment of steel truss bridge disclosed in this invention, preferably, in step S6, the construction monitoring instruction is a standardized Word document containing construction conditions, an assembly elevation control parameter table, and construction precautions; the assembly elevation control parameter table is dynamically calculated by substituting the design elevation, the predicted optimal pre-lift amount, and the error adjustment value based on the alignment state fine-tuning into the assembly elevation calculation formula, and the assembly elevation calculation formula is: assembly elevation = design elevation + pre-lift amount + error adjustment value.

[0012] According to the adaptive closed-loop monitoring method for the construction alignment of steel truss bridge disclosed in this invention, preferably, in step S7, the process of optimizing the model includes: constructing newly collected field measured data into input-output sample pairs and expanding them into the training sample library of the Kriging surrogate model so that the model can be retrained or its parameters tuned so that the model can continuously approximate the actual structural response.

[0013] The second aspect of the present invention discloses an adaptive closed-loop monitoring system for the construction alignment of a steel truss bridge, comprising: a memory for storing program instructions; and a processor for calling the program instructions stored in the memory to implement the adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge as described in any of the above technical solutions.

[0014] The beneficial effects of the present invention include at least the following: (1) Deep integration of physical mechanism and data-driven approach. Physical constraint samples are generated by finite element method, and Kriging model is used to achieve real-time prediction, which overcomes the shortcomings of pure data-driven model that lack physical constraints and has low extrapolation reliability.

[0015] (2) Quantification of prediction uncertainty. The Kriging model is based on Gaussian process regression. It outputs the prediction variance while providing the prediction amount, thus providing a clear risk boundary for construction decisions.

[0016] (3) BIM-driven intelligent closed-loop control. BIM is elevated to a dynamic decision-making center to realize automatic graded early warning of deviations, intelligent prediction of pre-lifting quantities, and automatic generation of construction instructions.

[0017] (4) The model continuously evolves itself. By dynamically replenishing the training sample library with newly collected data, the model is driven to continuously approximate the actual structural response, realizing the transformation from passive correction to active intelligent pre-control.

[0018] This invention applies the Kriging surrogate model to the intelligent prediction of pre-lift during steel truss bridge construction. By constructing a technical route of "finite element sample training—Kriging model building—real-time prediction—dynamic updating," it achieves a fundamental shift from empirical estimation to data-driven intelligent decision-making for pre-lift. Based on Gaussian process theory, this model can establish a high-precision surrogate model using samples generated from finite element analysis, accurately capturing the nonlinear coupling effects of multiple factors such as temperature, humidity, and self-weight on pre-lift. It effectively solves the technical problems of traditional pre-lift determination methods, which rely on empirical formulas, cannot adapt to complex time-varying effects, have insufficient prediction accuracy, and lack quantitative risk assessment. This provides reliable and efficient predictive support for adaptive control of construction alignment. Attached Figure Description

[0019] Figure 1 A flowchart illustrating an adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to an embodiment of the present invention is shown.

[0020] Figure 2 A schematic block diagram of an adaptive closed-loop monitoring system for the construction alignment of a steel truss bridge according to an embodiment of the present invention is shown.

[0021] Figure 3 A schematic diagram of the application process of a software platform according to an embodiment of the present invention is shown.

[0022] Figure 4 A schematic diagram of the system architecture of a software platform according to an embodiment of the present invention is shown.

[0023] Figure 5 A schematic diagram of the functional interface of a software platform according to an embodiment of the present invention is shown. Detailed Implementation

[0024] To better understand the above-described objects, features, and advantages of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. Many specific details are set forth in the following description to provide a thorough understanding of the invention; however, the invention may be practiced in other ways different from those described herein, and therefore, the invention is not limited to the specific embodiments disclosed below.

[0025] like Figure 1 As shown, an adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge is disclosed according to an embodiment of the present invention, comprising: Step S1, Model Building: Build a parametric BIM model that integrates structural geometric properties and information on the construction stage, and simultaneously build a finite element mechanical analysis model for simulating the time-varying structural system at each construction stage. In this step, based on the bridge design parameters, a parametric BIM model with geometric shape and key attributes is created using Autodesk Revit software. In the BIM model, information such as the location of the beam segment to which each component belongs, the beam segment number, and the construction stage are defined. Based on the same bridge design parameters and construction plan, a finite element mechanical analysis model for structural calculation is established using Midas / Civil finite element analysis software to define the structural system, material time-varying properties, stage loads, and boundary conditions corresponding to each construction stage. Step S2, surrogate model training: Based on the finite element mechanical analysis model, perform multivariate parametric mechanical simulation to obtain an initial high-fidelity training sample set of input parameters and output pre-lifting amount; based on the Gaussian process regression framework, establish a Kriging surrogate model with pre-lifting amount as output and construction parameters as input, optimize the relevant function parameters through maximum likelihood estimation, and obtain a trained Kriging surrogate model that meets the engineering accuracy requirements. (1) A training sample set is generated through parametric analysis of the finite element mechanical analysis model. The input parameters include temperature, humidity, component information and other parameters, and the output parameter is the prelift amount. (2) A Kriging model is constructed using Gaussian process regression, and the relevant function parameters θ are automatically optimized; (3) Establish an independent pre-lifting prediction model for each key control point; (4) Evaluate the accuracy of the model through cross-validation to ensure that the prediction error of the pre-lifting amount is within the allowable range of the project.

[0026] Step S3, Data Acquisition: Collect real-time construction parameters at the construction site and write them into the database. The construction parameters include: component ID, measured elevation information of control points, ambient temperature and ambient humidity. Step S4, Linear Deviation Analysis: Compare the measured elevation information of the control points with the design elevation, calculate the absolute error of the line, and automatically determine the line status of each control point as: normal, warning or exceeding the limit based on the preset first standard threshold and second standard threshold, and perform differentiated color coding visualization marking in the parametric BIM model. Step S5, Intelligent Prediction of Pre-lift Amount: Input the real-time construction parameters of the current construction stage into the trained Kriging proxy model, use its built-in related kernel function to perform nonlinear mapping, output the predicted value of the pre-lift amount for the next construction stage, and output the prediction variance simultaneously. Step S6, adaptive generation of monitoring instructions: Based on the linear state, the optimal pre-lifting amount prediction value and the prediction variance, the pre-lifting amount adjustment value is dynamically calculated by the BIM model, and the document automation component is called to dynamically generate construction monitoring instructions containing the final assembly elevation control parameter table. Step S7, Adaptive Closed-Loop Feedback and Optimization: The construction monitoring command is issued and executed. After the construction is completed, the monitoring data of the new construction phase is collected and fed back to the BIM model to update the alignment status. At the same time, the new round of monitoring data is converted into input-output sample pairs and added to the training sample library of the Kriging proxy model to optimize the model and perform closed-loop adaptive control of the construction alignment.

[0027] According to yet another embodiment of the present invention, the specific implementation of step S2 in the above embodiments is also disclosed: Training data generation and preprocessing: Based on the finite element analysis model of the bridge construction stage, an orthogonal experimental design method is used to systematically generate training samples within the construction parameter combination space. The defined input parameters include specifications, temperature, humidity, structural self-weight coefficient, etc.; the defined output is the pre-lift (or displacement response) of the key control points of the structure. Input and output data are preprocessed by normalization to eliminate dimensional differences and improve the numerical stability of the model. The formula is as follows: , , In the formula: X Indicates input data, , These are the minimum and maximum values ​​in the input training dataset, respectively; Y Indicates the output data. , This outputs the minimum and maximum values ​​in the training dataset.

[0028] After normalization, both input and output data are mapped to the [0,1] interval, thereby improving the numerical stability and convergence efficiency of the Kriging surrogate model during training and prediction.

[0029] Data cleaning is performed, including checking and handling missing values ​​and removing completely duplicate sample points, to ensure the quality of training samples.

[0030] Kriging proxy model construction: Using a Gaussian process regression framework, an independent Kriging surrogate model is established for the key output response (pre-lifting quantity); a constant regression function and a Gaussian correlation function are selected as the basic functions of the model, with the following expressions: Constant regression function: In the formula: The vector of regression basis functions; This is the transpose operator for matrices or vectors. , which is the vector of regression coefficients; is the intercept coefficient in the constant regression term.

[0031] Gaussian function: In the formula: The correlation function between sample points; For relevant parameters, control the first k The correlation decay rate in each input variable direction; m Dimensions of the input variables; and Sample points and The k 3D coordinates.

[0032] The model hyperparameters (correlation length θ) are automatically optimized through maximum likelihood estimation, and reasonable boundaries are set to constrain the optimization process; For new sample points The expressions for the predicted response and the predicted variance are as follows: , In the formula: For point The predicted response value at the location; For the relevant vector; The response column vector for the training sample points; This is the regression matrix; The maximum likelihood estimate of the regression coefficient vector; , In the formula: For point The predicted variance at the location; R This is the correlation matrix.

[0033] Model training and accuracy validation: Model training was completed using MATLAB's built-in DACE toolbox and equivalent algorithms; Leave-one-out cross-validation was used to evaluate the model’s generalization ability and prediction accuracy. Calculate the accuracy metrics of the output model, including the coefficient of determination (R²). 2 The expression for the root mean square error (RMSE) is: , , In the formula: For the first The actual response value of each sample; For the model to the first The predicted response value for each sample; This is the sample mean of the actual response values; n The total number of test samples. Coefficient of determination. The root mean square error (RMSE) is used to measure a model's ability to explain the variability of the response variable; a value closer to 1 indicates a better model fit. This value reflects the average deviation between the predicted and actual values; the smaller the value, the higher the model's prediction accuracy.

[0034] According to yet another embodiment of the present invention, the specific implementation of step S3 in the above embodiments is also disclosed: Sensor deployment and numbering: A sensor network is pre-deployed at key components of the bridge structure (such as main beams, supports, and sliding beams), and each sensor is uniquely numbered or an initial binding relationship is established with the component ID.

[0035] The sensor network collects the following data in real time: 3D spatial coordinates: Precise 3D coordinates of control points are obtained through machine vision intelligent measuring instruments. Elevation and deflection: Elevation changes and vertical displacement (deflection) of key points are measured using a level or tilt sensor. Horizontal displacement and tilt: Lateral displacement and rotation of components are monitored using displacement gauges or tilt sensors. Temperature: The surface temperature of the structure and the ambient temperature are monitored to correct for temperature-induced deformation. Humidity: Ambient humidity is monitored to aid in the analysis of changes in material properties.

[0036] Data identification and transmission: All data includes time, component ID, measurement points, and construction stage information, and is transmitted in real time to the construction monitoring database or BIM platform via wireless network, providing a real-time data foundation for subsequent alignment deviation analysis.

[0037] like Figure 5 As shown, according to another embodiment of the present invention, the specific implementation of step S4 in the above embodiment is also disclosed: the linear deviation under each construction condition is obtained through data comparison and analysis, and the linear status is divided into normal, warning, and over-limit according to the magnitude of the deviation; and the "over-limit", "warning", or "normal" status is visually identified in the interface using color coding (red, yellow, green); specifically: The system sets permissible thresholds for alignment deviation, including a first threshold (warning threshold) and a second threshold (over-limit threshold), with the first threshold being less than the second threshold. The calculated actual alignment deviation is compared to these two thresholds. If the alignment deviation exceeds the second threshold, the alignment status is determined to be "over-limit," and the corrective control process is immediately initiated, with the construction plan reviewed. If the alignment deviation is greater than or equal to the first threshold but less than or equal to the second threshold, the alignment status is determined to be "warning," and the system increases the monitoring frequency and marks this condition, prompting on-site attention and preparation for adjustments. If the alignment deviation is less than the first threshold, the alignment status is determined to be "normal," indicating that the current alignment meets the control objectives, and the monitoring system will continue with the next stage of prediction and construction according to the established plan. By classifying alignment deviations, quantitative assessment and differentiated responses to alignment status are achieved, effectively improving the intelligence level and data reliability of the BIM model in bridge construction monitoring.

[0038] According to another embodiment of the present invention, the specific implementation of steps S5 and S6 in the above embodiments is also disclosed: Construction parameters (ambient temperature, humidity, component information, etc.) are input into the trained Kriging proxy model. This model, through a built-in Gaussian process regression relationship, quickly maps the nonlinear relationship between input features and output response, and directly outputs the predicted pre-lift values ​​of each key control point in the next construction stage. Based on the linear deviation and the pre-lift prediction results of the proxy model, the BIM platform automatically generates construction monitoring instructions for subsequent construction stages. The instructions include at least the pre-lift adjustment value and construction precautions. The specific process of generating construction monitoring instructions includes: based on the selected beam segment location, beam segment number, and construction conditions, automatically querying and loading the corresponding design elevation and pre-lift value output by the Kriging model through the SQL database interface.

[0039] Based on the formula: Assembly Elevation = Design Elevation + Pre-lift Amount + Error Adjustment Value, the final control elevation of each measuring point is calculated in real time. The "Error Adjustment Value" can be manually entered based on real-time alignment deviations to allow for fine-tuning of the final assembly elevation.

[0040] By calling the document automation component, the above parameterized data is combined with the standard template to dynamically generate a standardized Word monitoring instruction document containing the following core contents: header information, construction conditions and process descriptions, assembly elevation control parameter table, and construction precautions.

[0041] According to another embodiment of the present invention, the specific implementation of step S7 in the above embodiments is also disclosed: Instruction output, issuance, and tracking: The generated instruction file is pushed to the project and issued to relevant parties such as construction and supervision through various means; the instruction is accompanied by a unique number, providing an index for subsequent execution feedback and data traceability. After a new construction phase is completed according to the monitoring instructions, the on-site sensor network collects the latest alignment data, including the three-dimensional coordinates, elevation, displacement, and ambient temperature of each control point. The collected data is encapsulated and transmitted back to the construction monitoring database for associated storage. After receiving the new monitoring data, the BIM model automatically calculates the deviation between the measured values ​​of each measuring point and the target alignment of the current phase. The alignment status is determined according to preset thresholds (first threshold, second threshold): if the deviation is less than the first threshold, the status is "normal"; if it is between the first and second thresholds, the status is "warning"; if it is greater than the second threshold, the status is "out of limit". The analysis results are visualized in the BIM three-dimensional model. Based on the Kriging agent model, and combining the latest alignment deviation data with the construction parameters for the next stage, the optimal pre-lift amount for the next construction phase is re-predicted and generated, automatically triggering the generation and issuance of a new round of monitoring instructions. This forms a complete closed loop of "monitoring → analysis → prediction → instruction → execution → feedback".

[0042] like Figure 2 As shown, according to another embodiment of the present invention, an adaptive closed-loop monitoring system 200 for construction alignment of a steel truss bridge is also disclosed, comprising: a memory 201 for storing program instructions; and a processor 202 for calling the program instructions stored in the memory to implement the adaptive closed-loop monitoring method for construction alignment of a steel truss bridge as described in the above embodiment.

[0043] like Figure 3As shown in the embodiments of the present invention, a specific application flowchart of the software platform developed according to the adaptive closed-loop monitoring method for steel truss bridge construction alignment disclosed in the above embodiments is also disclosed. This process integrates multiple stages, including preliminary initialization work, real-time monitoring of the construction process, risk warning and manual intervention, and generation and issuance of monitoring instructions. The process begins with the preliminary work of the project, including platform initialization settings, design compliance calculation, initial construction process simulation calculation, and platform database initialization. These foundational tasks ensure that the platform can correctly load bridge design parameters, finite element models, and construction stage information. After entering the process-oriented functions, the platform collects real-time on-site measured values ​​through the monitoring data input module and aggregates multi-source data through the monitoring data integration module. The monitoring data warning module and the visualization warning analysis module determine whether there is a risk in the current data: if the current data is risk-free, no adjustments are needed for subsequent construction, and the next segment construction can proceed directly; if the current data presents a risk, the risk level is further determined. If the risk level is not high, the platform sends the analysis results of the abnormal phenomenon to the on-site personnel, prompting them to pay attention to construction details and continue subsequent construction. If the risk level is high, the comprehensive analysis stage is entered. The platform offers comprehensive query and analysis functions, and also supports manual analysis. Based on the analysis results, it determines whether the finite element model needs modification: if modification is required, the finite element model is recalculated, and the theoretical calculation values ​​are updated with one click; if no modification is required, a decision is made on whether to hold a monitoring technology issue seminar. Finally, the platform enters the monitoring instruction generation function, and the generated instructions must be reviewed by all parties. If the instruction fails the review, it returns to the comprehensive analysis stage for correction; if it passes the review, the construction monitoring instruction is officially released to guide on-site construction.

[0044] like Figure 4 As shown in the embodiments of the present invention, the platform architecture of the software platform developed according to the adaptive closed-loop monitoring method for steel truss bridge construction alignment disclosed in the above embodiments is also disclosed. The platform architecture adopts a layered design concept, consisting of a presentation layer, an application layer, a support layer, and a data layer from top to bottom.

[0045] (1) Data Layer: The data layer is mainly responsible for the preliminary processing of bridge alignment monitoring data, including existing structural data such as design elevation, theoretical elevation after each construction stage, and on-site measured data. The implementation of various functions of the platform is based on the operation of various types of data, and the core of this process lies in the processing and analysis of raw data.

[0046] (2) Support layer: mainly combines the functional requirements of each module to carry out specific operations on the data layer, and provides underlying technical support for the implementation of various functions of the platform.

[0047] (3) Application Layer: As the core hub of the platform, this layer acts as a bridge connecting the upper and lower layers. It is used to receive user commands from the front-end interface, access and operate the underlying data through various functional modules of the application layer, perform various operations on the data layer with the support layer, and finally provide feedback to the user through the presentation layer.

[0048] (4) Presentation layer: The core responsibility of this module is to receive user instructions, initiate service requests, schedule function execution, and return processing results.

[0049] Table 1. Monitoring results of partial alignment of steel truss girders ; As shown in Table 1, for the key measuring points E1 to E5 of the side main trusses of the steel truss bridge, the absolute value of the difference between the measured elevation and the design elevation is controlled within ±1 cm, which is far smaller than the allowable deviation range of conventional construction alignment control. This result demonstrates that the construction alignment monitoring method and system based on BIM and Kriging proxy models proposed in this invention have good application effects and high practical engineering application value.

[0050] In summary, the above embodiments provide an adaptive closed-loop monitoring method for the construction alignment of steel truss bridges based on BIM and a Kriging proxy model. This method establishes a parametric BIM model and a finite element analysis model of the bridge, incorporating construction stage information and geometric attributes. Training samples are generated based on the parametric analysis data from the finite element mechanical analysis model. A Kriging proxy model is established, with construction parameters as input and pre-lifting amount as output. Real-time monitoring data from the construction site is collected, alignment deviations are calculated, and three states—normal, warning, and exceeding limits—are identified. The proxy model, combined with construction parameters and alignment status, predicts the optimal pre-lifting amount and determines the error adjustment value. The BIM platform automatically generates monitoring instructions containing the pre-lifting adjustment value and construction precautions. After the instructions are executed, new data is collected and fed back to the platform for alignment analysis. This achieves adaptive control throughout the entire process, from data collection, intelligent analysis, prediction and adjustment to closed-loop feedback, significantly improving the accuracy and intelligence level of adaptive closed-loop monitoring of the steel truss bridge construction alignment.

[0051] The methods and algorithms described in the embodiments herein can be implemented through a combination of BIM platform software, finite element analysis tools, data-driven modeling programs, and necessary sensor hardware and circuit systems. Whether these functions are implemented primarily through software or through a combination of software and hardware depends on the specific application scenario and design constraints of the technical solution. Professionals may use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention. All or part of the steps in the various methods of the above embodiments can be implemented by a program controlling the relevant hardware. The program can be stored in a readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically-erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other readable medium that can be used to carry or store data.

[0052] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for adaptive closed-loop monitoring of the construction alignment of a steel truss bridge, characterized in that, include: S1, Model Building: Establish a parametric BIM model that integrates structural geometric properties and information on the construction stage, and simultaneously establish a finite element mechanical analysis model for simulating the time-varying structural system at each construction stage. S2, Proxy Model Training: Based on the finite element mechanical analysis model, perform multivariate parametric mechanical simulation to obtain an initial high-fidelity training sample set of input parameters and output pre-lifting amount; Based on the Gaussian process regression framework, a Kriging surrogate model with pre-lifting amount as output and construction parameters as input is established. The parameters of the relevant function are optimized by maximum likelihood estimation to obtain a trained Kriging surrogate model that meets the engineering accuracy requirements. S3, Data Acquisition: Collect real-time construction parameters at the construction site and write them into the database. The construction parameters include: component ID, measured elevation information of control points, ambient temperature and ambient humidity. S4, Linear Deviation Analysis: Compare the measured elevation information of the control points with the design elevation, calculate the absolute error of the line, and automatically determine the line status of each control point as: normal, warning or exceeding the limit according to the preset first standard threshold and second standard threshold, and perform differentiated color-coded visual marking in the parametric BIM model; S5, Intelligent Prediction of Pre-lift Amount: Input the real-time construction parameters of the current construction stage into the trained Kriging proxy model, use its built-in related kernel function to perform nonlinear mapping, output the predicted value of the pre-lift amount for the next construction stage, and simultaneously output the prediction variance. S6, Adaptive generation of monitoring instructions: Based on the linear state, the optimal pre-lifting prediction value and the prediction variance, the BIM model dynamically calculates the pre-lifting adjustment value and calls the document automation component to dynamically generate construction monitoring instructions containing the final assembly elevation control parameter table.

2. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 1, characterized in that, Also includes: S7, Adaptive Closed-Loop Feedback and Optimization: The construction monitoring command is issued and executed. After the construction is completed, the monitoring data of the new construction phase is collected and fed back to the BIM model to update the alignment status. At the same time, the new round of monitoring data is converted into input-output sample pairs and added to the training sample library of the Kriging proxy model to optimize the model, so as to perform closed-loop adaptive control of the construction alignment.

3. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 1, characterized in that, In step S1, the BIM model is built using Autodesk Revit software to integrate the geometric shape of bridge components and information on the construction stage to which they belong; the finite element mechanical analysis model is built using Midas / Civil software to simulate the structural system, time-varying material properties, loads and boundary conditions of each construction stage.

4. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 1, characterized in that, Step S2 specifically includes: Training sample generation: Based on the finite element mechanical analysis model, parametric analysis is performed. Orthogonal experimental design method is adopted to sample in the parameter space including temperature, humidity and self-weight coefficient to obtain a training sample set of input parameters and output preload, and the data is normalized and preprocessed. Constructing a Kriging surrogate model: Using the Gaussian process regression framework, a constant regression function and a Gaussian correlation function are selected to establish a Kriging surrogate model, and the correlation function parameters are optimized through maximum likelihood estimation; Model validation: The leave-one-out cross-validation method is used to evaluate the model, and the coefficient of determination and root mean square error are calculated to ensure that the prediction accuracy meets the engineering requirements.

5. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 1, characterized in that, In step S4, the process of determining the linear state specifically includes: If the elevation error value of a certain measuring point is less than the first standard threshold, the linear state of that point is determined to be "normal". If the elevation error value is greater than or equal to the first standard threshold and less than or equal to the second standard threshold, it is determined as a "warning". If the elevation error value is greater than the second standard threshold, it is determined to be "out of limit"; The linear state is visualized and marked with different colors in both the BIM model and the exported standardized EXCEL file.

6. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 1, characterized in that, In step S6, the construction monitoring instruction is a standardized Word document containing construction conditions, an assembly elevation control parameter table, and construction precautions. The assembly elevation control parameter table is dynamically generated by substituting the design elevation, the predicted optimal pre-lift amount, and the error adjustment value based on the alignment state fine-tuning into the assembly elevation calculation formula. The assembly elevation calculation formula is as follows: Assembly elevation = Design elevation + Pre-lifting amount + Error adjustment value.

7. The adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge according to claim 2, characterized in that, In step S7, the process of optimizing the model includes: constructing input-output sample pairs from the newly collected field measurement data and expanding them into the training sample library of the Kriging surrogate model so that the model can be retrained or its parameters tuned so that the model can continuously approximate the actual structural response.

8. A construction alignment adaptive closed-loop monitoring system for steel truss bridges, characterized in that, include: Memory, used to store program instructions; A processor is configured to invoke the program instructions stored in the memory to implement the adaptive closed-loop monitoring method for the construction alignment of a steel truss bridge as described in any one of claims 1 to 7.

Images