A BIM-based method and system for monitoring the sliding installation of steel box girders

Abstract

This invention discloses a BIM-based monitoring method and system for the sliding installation of steel box girders, belonging to the field of construction monitoring technology. It addresses the problems of insufficient real-time monitoring accuracy and delayed risk warning during the sliding process of steel box girders. Through non-uniform rational B-spline surface reconstruction, high-precision matching between the BIM model and laser-scanned point clouds is achieved, generating dynamic anchor point coordinates and correcting strain measurements. Sliding stability is assessed by combining the rate of change of tilt angle and displacement increment. A bidirectional long short-term memory neural network model is used to predict the cumulative deviation of the sliding trajectory within the next ten seconds, generating hydraulic jacking correction parameters when the deviation exceeds a threshold. A stress distribution heatmap is constructed based on a radial basis function interpolation algorithm, and collision risk coordinates are marked in real time. Finally, based on the risk index, the proportional-integral coefficient is dynamically adjusted to generate hydraulic cylinder adjustment commands, achieving closed-loop real-time correction control of the sliding process and improving installation safety and accuracy.

Description

Technical Field

[0001] This invention relates to the field of construction monitoring technology, specifically to a BIM-based method and system for monitoring the sliding installation of steel box girders. Background Technology

[0002] With the rapid advancement of urban viaducts, cross-river passages, and integrated transportation hubs, steel box girder structures have become the primary form of construction for long-span bridges due to their advantages such as large span, light weight, and high degree of prefabrication. In complex construction environments (such as crossing operational lines, waterways, or densely built-up areas), traditional lifting and installation methods are limited by workspace and safety control, making it difficult to meet construction requirements. In contrast, sliding installation technology, with its advantages of continuous advancement, less interference, and space saving, is widely used in various large-scale steel structure bridge projects. However, the sliding process involves the synchronous driving and precise positioning of large-volume components. Factors such as structural displacement accuracy, stress state changes, and surrounding space risks significantly impact construction safety, necessitating the use of digital means to achieve full-process visual monitoring and dynamic control.

[0003] Existing slip monitoring methods generally employ multiple independent devices to collect single parameters such as displacement, tilt angle, and strain. Data is stored across different platforms, making information fusion and status linkage judgment difficult, resulting in insufficient timeliness in slip status identification. In actual working conditions, influenced by temperature changes and uneven support stiffness, the slip path may exhibit a gradual deviation trend. Traditional mechanisms relying on preset thresholds to trigger alarms cannot effectively identify early signs of deviation, posing a risk of response lag. Furthermore, control parameters during hydraulic jacking primarily depend on on-site manual experience for adjustment, limiting control accuracy and hindering synchronous and coordinated adjustment of large-span steel box girders, potentially leading to structural attitude deviations or track operation interference and other safety hazards. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a BIM-based method and system for monitoring the sliding installation of steel box girders, thus solving the problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a BIM-based method for monitoring the sliding installation of steel box girders, comprising the following steps: S1. By reconstructing and matching the BIM model with the laser scanning point cloud using a non-uniform rational B-spline surface, a dynamic anchor point coordinate set is generated at the key sections of the steel box girder and the sliding track control points are bound. The strain measurement value is corrected according to the nonlinear mapping relationship between temperature and strain, and the sliding stability is evaluated by combining the correlation analysis between the rate of change of tilt angle and the displacement increment; S2. The sliding time series data of the steel box girder and the stability evaluation results are input into a bidirectional long short-term memory neural network model to predict the sliding trajectory within the next ten seconds. S3. The cumulative deviation is generated as hydraulic jacking correction parameters are generated when the cumulative deviation exceeds the safety threshold; S4. The cumulative deviation is mapped to the anchor point spatial coordinate set, and the stress distribution heat map is dynamically mapped on the BIM model surface through the radial basis function interpolation algorithm. The collision risk coordinates are marked in real time according to the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector; S5. According to the hydraulic jacking correction parameters and the collision risk coordinates, a response mechanism is constructed in which the proportional coefficient increases with the risk exponent and the integral coefficient is dynamically adjusted with the prediction deviation. The hydraulic cylinder jacking pressure or speed adjustment command is generated and sent to the jacking equipment controller to perform real-time correction.

[0006] Furthermore, the specific process of generating a dynamic anchor point coordinate set and binding the sliding track control points at key sections of the steel box girder by reconstructing and matching the BIM model and laser scanning point cloud through non-uniform rational B-spline surface reconstruction is as follows: The laser scanning point cloud is subjected to curvature-based partitioned filtering to extract feature point cloud clusters of the steel box girder web and top and bottom plates; an adaptive surface control network with the same topology as the BIM model control points is constructed using a non-uniform rational B-spline algorithm, and registration is performed through iterative nearest-point optimization; a dynamic anchor point cluster is generated at equal intervals along the sliding axis at key sections of the registered BIM model, each anchor point is assigned a unique spatial identifier code, and the coordinates of the track control points measured by the total station are simultaneously fused to establish a two-way coordinate transformation mapping relationship between the anchor points and the track control points, forming a dynamic spatial reference network for the entire sliding process.

[0007] Furthermore, the specific process for evaluating slip stability by correcting the strain measurement value based on the nonlinear mapping relationship of temperature and strain, combined with the correlation analysis of the tilt angle change rate and displacement increment, is as follows: A piecewise temperature strain compensation function is constructed by calling the material thermal expansion coefficient database: the normal temperature segment is fitted using a quadratic polynomial, and the high temperature segment introduces an exponential correction term based on the material's thermal expansion characteristics; the dynamic Pearson correlation coefficient between the tilt angle change rate and displacement increment within the sliding time window is calculated in real time. When the absolute value of the correlation coefficient exceeds a preset threshold and the first derivative of the tilt angle change rate remains positive, an instability classification signal is triggered, and risk level parameters and the spatial coordinate codes of associated anchor points are output.

[0008] Furthermore, the logical process of inputting the steel box girder slip time series data and stability assessment results into a bidirectional long short-term memory neural network model to predict the cumulative deviation of the slip trajectory within the next ten seconds is as follows: One-dimensional convolutional noise reduction processing is performed on the raw data stream from the displacement sensors to extract key displacement feature waveforms, analyze the risk level parameters and spatial coordinate codes, generate regional sensitivity weighting coefficients, and fuse the track surface friction coefficient detection values ​​to construct a working condition feature vector; the weighted displacement sequence, spatial coordinate codes, and track friction coefficient vector are input into the bidirectional long short-term memory neural network, and an attention weight allocation mechanism is embedded through a gating unit to output the predicted value of the cumulative deviation within the next ten seconds.

[0009] Furthermore, the logical process for generating hydraulic jacking correction parameters when the cumulative deviation exceeds the safety threshold is as follows: the cumulative deviation is converted into normal pressure compensation and tangential speed adjustment at each jacking point based on a spatial vector decomposition algorithm; parameter generation is activated through a dual-condition triggering mechanism of static safety threshold and deviation increase dynamic threshold, and the output hydraulic jacking correction parameters include the target cylinder spatial topology code, control mode decision flag, and proportional-integral basic parameter set, wherein the proportional-integral basic parameter set includes the proportional coefficient reference value and the integral coefficient reference value.

[0010] Furthermore, the process of mapping the cumulative deviation to the anchor point spatial coordinate set and dynamically mapping the stress distribution heatmap on the BIM model surface using the radial basis function interpolation algorithm is as follows: Based on the dynamic anchor point spatial coordinate set, a physical field coupling mapping relationship between the cumulative deviation and the corrected strain data is established; a continuous stress field distribution function is constructed using the radial basis function interpolation algorithm with the anchor points as control nodes; the stress transfer effect between adjacent anchor points is dynamically weighted using the Gaussian kernel function to generate a gradient stress heatmap on the BIM model surface; the slip trajectory change trend is integrated in real time, and high-stress areas are dynamically focused and rendered.

[0011] Furthermore, the specific process of marking collision risk coordinates in real time based on the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector is as follows: extract the spatial normal vectors of key structures including protective canopies and bridge piers from the obstacle library of the BIM model, calculate the dot product value of the current sliding direction vector of the steel box girder and the obstacle normal vector, and mark the three-dimensional risk cube in the anchor point coordinate system when the dot product result is greater than the preset collision risk coefficient and the real-time distance is less than the preset threshold; color-code the cube according to the risk level and mark the collision risk coordinates.

[0012] Furthermore, based on the hydraulic jacking correction parameters and collision risk coordinates, the specific process of constructing a response mechanism in which the proportional coefficient increases with the risk index and the integral coefficient dynamically adjusts with the prediction deviation is as follows: Based on the hydraulic jacking correction parameters, the reference values ​​of the proportional coefficient and the integral coefficient are extracted, and the target cylinder topology code and control mode flag are obtained simultaneously. The collision risk coordinates are converted into a spatial risk density index. According to the control mode flag, the corresponding proportional coefficient enhancement method is selected: under pressure control mode, the proportional coefficient increases geometrically with the risk index, and under speed control mode, the proportional coefficient increases linearly with the risk index. The integral coefficient is uniformly adjusted in reverse according to the cumulative deviation. When the deviation growth rate exceeds the critical slope, the integral action is locked, and the dynamic response parameter set bound to the cylinder topology code is output.

[0013] Furthermore, the specific process of generating and sending the hydraulic cylinder pushing pressure or speed adjustment command to the pushing equipment controller for real-time correction is as follows: Parse the dynamic response parameter group and its bound cylinder topology code, and generate execution commands according to the control mode flag: in pressure control mode, convert the proportional coefficient and integral coefficient into a pressure correction amount containing integral operation; in speed control mode, convert the proportional coefficient into a speed adjustment amount; encapsulate the command data stream through the industrial real-time communication protocol, locate the target device according to the cylinder topology code, and send it to the programmable controller.

[0014] A BIM-based steel box girder slip installation monitoring system includes the following modules: slip assessment module, deviation early warning module, risk monitoring module, and closed-loop control module. The slip assessment module uses a non-uniform rational B-spline surface to reconstruct and match the BIM model with laser-scanned point clouds, generating a dynamic anchor point coordinate set at key sections of the steel box girder and binding the slip track control points. It corrects strain measurements based on the nonlinear mapping relationship between temperature and strain, and assesses slip stability by combining the correlation analysis between the rate of change of tilt angle and displacement increment. The deviation early warning module inputs the steel box girder slip time-series data and stability assessment results into a bidirectional long short-term memory neural network model to predict the slip track within the next ten seconds. The cumulative deviation of the track is used to generate hydraulic jacking correction parameters when the cumulative deviation exceeds the safety threshold. The risk monitoring module maps the cumulative deviation to the anchor point spatial coordinate set, dynamically maps the stress distribution heat map on the BIM model surface through the radial basis function interpolation algorithm, and marks the collision risk coordinates in real time according to the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector. The closed-loop control module is used to construct a response mechanism based on the hydraulic jacking correction parameters and collision risk coordinates, which dynamically adjusts the proportional coefficient with the risk exponential growth and the integral coefficient with the prediction deviation, generates adjustment commands for the hydraulic cylinder jacking pressure or speed, and sends them to the jacking equipment controller to perform real-time correction.

[0015] The present invention has the following beneficial effects:

[0016] (1) A BIM-based method for monitoring the slip installation of steel box girders. By establishing a BIM point cloud twin benchmark and a dynamic anchor point binding mechanism, the spatiotemporal benchmark deviation of multi-source sensor data is eliminated, and real-time collaborative analysis of strain, tilt angle and displacement parameters is achieved. This solves the problem of missed judgment of slip instability caused by isolated data in traditional methods. Based on a bidirectional long short-term memory neural network model, by integrating track friction characteristics and stability assessment results, the passive nature of traditional threshold alarms is broken. It can provide early warning of the gradual deviation trend under complex working conditions such as track slippage and hydraulic failure, thereby reducing the risk of sudden accidents from the root.

[0017] (2) A BIM-based steel box girder sliding installation monitoring system maps the cumulative deviation to a dynamic anchor point spatial coordinate set and uses a radial basis function interpolation algorithm to dynamically visualize the stress distribution on the BIM model surface, achieving real-time perceptible expression of the structural state. Simultaneously, by combining the spatial relationship between the sliding path direction vector and the obstacle normal vector for collision point product discrimination, potential contact areas can be identified, providing alerts for local structural risks during construction and improving the timeliness of risk prediction. Furthermore, the system constructs a control response mechanism with risk index and prediction deviation as inputs, employing a proportional-integral dynamic adjustment strategy to generate pressure or speed adjustment commands for the hydraulic cylinders, and implementing closed-loop correction control through a controller. This control mechanism can adjust response parameters in real-time according to the risk evolution trend, improving the coordination and stability of the jacking execution during the sliding process, thus providing support for precision control and safety assurance in steel box girder sliding construction.

[0018] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0019] Figure 1 This is a flowchart of a BIM-based method for monitoring the sliding installation of steel box girders according to the present invention.

[0020] Figure 2 This is a flowchart of a BIM-based steel box girder sliding installation monitoring system according to the present invention. Detailed Implementation

[0021] This application provides a BIM-based method and system for monitoring the sliding installation of steel box girders. It addresses the problems of fragmented multi-source data, delayed state assessment, insufficient risk prediction capabilities, and inaccurate hydraulic correction response in existing sliding monitoring processes, thereby improving the monitoring integration, deviation prediction accuracy, and correction control efficiency during the steel box girder sliding construction process.

[0022] The overall concept of the solution in this application embodiment is as follows:

[0023] A point cloud anchor network aligned with the BIM model is constructed, and the correlation between temperature strain correction and tilt displacement is integrated to achieve slip stability analysis; a short-term prediction of trajectory deviation is performed based on a bidirectional long short-term memory neural network, and a dynamic risk perception mechanism is formed by combining spatial stress mapping and collision risk detection; the hydraulic system is driven to perform real-time correction through a proportional-integral parameter adaptive adjustment strategy, realizing intelligent and closed-loop monitoring and control of the steel box girder slip process.

[0024] Please see Figure 1 This invention provides a technical solution: a BIM-based method for monitoring the sliding installation of steel box girders, comprising the following steps: S1. By reconstructing and matching the BIM model and laser scanning point cloud using a non-uniform rational B-spline surface, a dynamic anchor point coordinate set is generated at the key sections of the steel box girder and the sliding track control points are bound. The strain measurement value is corrected according to the nonlinear mapping relationship between temperature and strain, and the sliding stability is evaluated by combining the correlation analysis between the rate of change of tilt angle and the displacement increment; S2. The sliding time series data of the steel box girder and the stability evaluation results are input into a bidirectional long short-term memory neural network model to predict the cumulative deviation of the sliding trajectory within the next ten seconds. S3. When the cumulative deviation exceeds the safety threshold, hydraulic jacking correction parameters are generated; S4. The cumulative deviation is mapped to the anchor point spatial coordinate set, and the stress distribution heat map is dynamically mapped on the BIM model surface through the radial basis function interpolation algorithm. The collision risk coordinates are marked in real time according to the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector; S5. Based on the hydraulic jacking correction parameters and the collision risk coordinates, a response mechanism is constructed in which the proportional coefficient increases with the risk exponent and the integral coefficient is dynamically adjusted with the prediction deviation. The hydraulic cylinder jacking pressure or speed adjustment command is generated and sent to the jacking equipment controller to perform real-time correction.

[0025] In this implementation scheme, S1 precisely registers the BIM model from the design phase with the on-site laser scanning point cloud using Non-Uniform Rational B-Spline (NURBS) surface reconstruction. This generates dynamic anchor point coordinates with unique spatial identifiers at key sections of the steel box girder and establishes a mapping relationship with the sliding track control points, thereby constructing a spatial reference for the entire sliding process. Simultaneously, the nonlinear mapping relationship of temperature and strain is used to correct the strain values ​​measured by sensors in real time (the influence of temperature changes on strain is compensated by a pre-established mapping function). Furthermore, the correlation analysis between the tilt angle change rate (representing the attitude change rate) and displacement increment is combined to assess the structural stability during sliding, ensuring the reliability of subsequent monitoring data in a unified coordinate system and timely detection of potential anomalies. Non-Uniform Rational B-Spline (NURBS) surface reconstruction: a flexible surface representation and fitting method that accurately describes complex geometry through control points and weights, used to align the BIM model surface with point cloud data. Dynamic anchor point coordinate set: a set of reference points updated in real time during the sliding process, typically located at key sections, used to track component position changes and establish a mapping with track control points. Sliding track control points: Track node coordinates measured on-site or pre-designed, used to correct the sliding path and position. Temperature-strain nonlinear mapping relationship: A function or model characterizing the influence of structural temperature changes on strain measurement, used to compensate for strain sensor output. Inclination change rate: The rate of change of the structural tilt angle per unit time, reflecting attitude dynamics; Displacement increment: The difference in displacement of a component along the sliding direction or other directions between adjacent time points; Correlation analysis: Often refers to statistical methods (dynamic Pearson correlation), used to quantify the degree of correlation between two time-series signals (such as inclination change rate and displacement increment) to determine state change characteristics. S2 inputs time-series monitoring data, including displacement, inclination angle, and corrected strain, along with stability indicators derived from S1, into a bidirectional long short-term memory neural network (BiLSTM). This model performs forward and backward learning of short-time series features to predict the cumulative deviation of the sliding trajectory within the next ten seconds. When the predicted deviation exceeds the preset safety limit, the system automatically generates hydraulic jacking correction parameters (including target cylinder identification and corresponding pressure or speed adjustment parameters), thereby achieving forward-looking warning and rapid response to offset trends. Bidirectional Long Short-Term Memory Neural Network (BiLSTM): A deep learning temporal model that processes sequence information in both forward and backward directions. Compared to unidirectional LSTM, it can capture more comprehensive temporal features and is suitable for short-term trajectory prediction. Cumulative Deviation of Sliding Trajectory: Refers to the cumulative offset of the sliding component relative to the predetermined ideal path within the future prediction period, used to quantify the deviation trend. Safety Threshold: A deviation limit set based on engineering design or experience; when the predicted deviation reaches or exceeds this limit, the correction logic is triggered. Hydraulic Jacking Correction Parameters: Includes information such as the pressure or speed adjustment parameters required by the hydraulic system and the target cylinder identification, used to guide the hydraulic jacking equipment in implementing correction operations.The cumulative deviation predicted in S3 is mapped to the corresponding 3D position of the dynamic anchor point coordinate set. A continuous stress field distribution heatmap is constructed on the BIM model surface using a radial basis function (RBF) interpolation algorithm, allowing monitoring personnel to intuitively understand the local stress changes caused by the deviation. Simultaneously, the dot product of the slip path direction vector and the surface normal vector of surrounding obstacles is calculated in real time, and combined with distance information to determine potential collision risks. When risk conditions are met, the corresponding risk coordinates are marked in 3D space to promptly identify areas that need to be avoided or focused on during the slip process. Radial basis function interpolation: Based on known discrete control point (anchor point) values ​​(such as strain or deviation mapping values), a continuous field distribution is constructed using radial basis functions, suitable for visualizing field quantities in irregular point cloud data. Stress distribution heatmap: Displays the local stress value distribution on the 3D model surface using color gradients or intensity, helping to intuitively identify high-stress areas. Slip path direction vector: A unit vector representing the slip direction of the steel box girder, usually calculated from the difference between the current position and the previous moment or the ideal path direction. Obstacle Normal Vector: The normal vector of the obstacle surface at this location, used to determine the angle between the sliding direction and the obstacle surface. Spatial Dot Product Relationship: The dot product value of two vectors reflects the cosine of the angle between them. When it meets the set rules together with the distance condition, it can determine the potential collision risk. Collision Risk Coordinates: Marking the possible collision or approach positions in three-dimensional space, which can be used for subsequent avoidance or control strategies. Based on the hydraulic jacking correction parameters generated by S2 and the collision risk coordinates marked by S3, S4 drives the adaptive adjustment of the proportional-integral (PI) control gain through the risk index (derived from factors such as collision risk density and prediction deviation scale): when the risk index increases, the proportional coefficient is increased to enhance the correction force; changes in prediction deviation trigger the dynamic adjustment or freezing of the integral coefficient; the system generates corresponding hydraulic cylinder pressure or speed commands accordingly and sends them to the jacking equipment controller in real time, forming a closed-loop execution, so as to take into account both trajectory correction and risk avoidance in sliding construction, and ensure intelligent monitoring and precise control of the process. Proportional and integral coefficients: Gain parameters in classic PI control, dynamically adjusted here based on the risk index and deviation trend to achieve a more sensitive or stable correction response. Risk index: A quantified risk value, calculated comprehensively from factors such as collision risk coordinate density and predicted deviation scale, used to drive changes in control gain. Adaptive response mechanism: A strategy where control parameters (such as proportional and integral gain) automatically adjust according to changes in the input state (risk index, deviation prediction), without relying on fixed empirical values. Hydraulic cylinder jacking pressure / speed adjustment command: Defines the required pressure or jacking speed for the control signal of the hydraulic jacking device, used to implement specific correction actions. Jacking equipment controller: A field PLC or dedicated control unit that receives commands and drives the hydraulic system to execute corresponding actions.

[0026] Specifically, the process of generating a dynamic anchor point coordinate set and binding the sliding track control points at key sections of the steel box girder by reconstructing and matching the BIM model and laser scanning point cloud using non-uniform rational B-spline surfaces is as follows: The laser scanning point cloud is subjected to curvature-based partitioned filtering to extract feature point cloud clusters of the steel box girder web and top and bottom plates; an adaptive surface control network with the same topology as the BIM model control points is constructed using a non-uniform rational B-spline algorithm, and registration is performed through iterative nearest-point optimization; dynamic anchor point clusters are generated at equal intervals along the sliding axis at key sections of the registered BIM model, each anchor point is assigned a unique spatial identifier code, and the coordinates of the track control points measured by the total station are simultaneously fused to establish a two-way coordinate transformation mapping relationship between the anchor points and the track control points, forming a dynamic spatial reference network for the entire sliding process.

[0027] In this implementation scheme, point cloud partitioning filtering based on curvature features is used to extract structural feature surface point cloud clusters. (Note: This involves processing the original laser-scanned point cloud.) Curvature calculations are performed, and the regions are divided using the differences in principal curvature tensor eigenvalues. Structural boundaries such as the web, top plate, and bottom plate are used as primary targets to form a feature point cloud subset Q. web Q top Q bot This step is a preprocessing operation, establishing the initial structural geometric partitioning basis. Constructing a NURBS adaptive surface control net consistent with the BIM model topology serves to ensure that the fitted surface is consistent with the BIM model structure in parameter space, supporting subsequent registration. Formula (NURBS surface): in: Parameter description: ξ, η: two parameters in the NURBS parameter space; These are p-order and q-order B-spline basis functions, respectively; ω αβ : Weighting factor at the (α,β)th control point; D αβ : Coordinates of the (α,β)th control point (3D spatial point); u, v: Upper limit of the number of control points along the ξ and η directions. This surface control network automatically densifies the control point density in high curvature regions to match structural edge features. The iterative nearest point algorithm (ICP) is used to register and fit the surface to the BIM model, precisely aligning the fitted point cloud geometry with the BIM model and establishing a unified spatial coordinate reference. Optimizing the model representation: The objective is to minimize the following cost function: Parameter description: s κ b: The κ-th fitted surface sample point; κ : with s κ The most recent BIM model point; T: rotation matrix (3×3) for fitting to BIM; d: displacement vector (3×1), representing the translation amount; N: number of point pairs involved in the matching. Through... Minimize the solution (T,d) to ensure the fitted surface accurately covers the BIM model surface. The function of equidistantly arranged cross-sectional anchor point clusters along the slip axis is to construct a spatial anchor point system during structural slippage, providing calibration references for dynamic monitoring and control. Anchor point generation formula: X λ =U0+λ·δ·e; Parameter description: X λ : Spatial coordinates (3D point) of the λth anchor point; U0: Coordinates of the reference point of the sliding start section; δ: Anchor point spacing; e: Unit vector of the sliding direction (already set in BIM); λ: Anchor point number, value range [0, N] s -1];N s Number of anchor points. A unique spatial identifier is assigned to each anchor point. Where τ λ To generate timestamps, a two-way coordinate transformation model is established by integrating total station measured track control points. Its function is to achieve a mapping relationship between the measured coordinates of the physical track and the coordinates of the BIM model, ensuring that anchor point data can be implemented. The registration and transformation model defines the set of measured control points. Corresponding BIM anchor point set Construct the following mapping relationship: Z ρ =M ρ ·X ρ +v ρ Reverse mapping: Parameter description: Z ρ : Measured coordinates of track control points; X ρ : Corresponds to the anchor point coordinates in the BIM model; M ρ : Local affine transformation matrix (3D rotation / scaling); v ρ : Offset compensation vector; ρ: Track control point number. The function of constructing a dynamic spatial reference network for the entire sliding process is to unify the coordinate basis of various monitoring devices and control systems, serving as a spatial bridge for information transmission and control feedback. Execution logic: The unified coordinate system is set as the BIM model reference system. All displacement, attitude, strain, and other data acquired by the sensors are uniformly mapped to the anchor point mesh; this data is then used for path deviation prediction, stress heat map drawing, collision risk analysis, and hydraulic control parameter feedback.

[0028] Specifically, the process of correcting strain measurements based on the nonlinear mapping relationship of temperature and strain, and evaluating slip stability by combining the correlation analysis of the tilt angle change rate and displacement increment, is as follows: A piecewise temperature-strain compensation function is constructed by calling the material thermal expansion coefficient database: the normal temperature range is fitted using a quadratic polynomial, and the high temperature range introduces an exponential correction term based on the material's thermal expansion characteristics; the dynamic Pearson correlation coefficient between the tilt angle change rate and displacement increment within the sliding time window is calculated in real time. When the absolute value of the correlation coefficient exceeds a preset threshold and the first derivative of the tilt angle change rate remains positive, an instability classification signal is triggered, and risk level parameters and the spatial coordinate codes of associated anchor points are output.

[0029] In this implementation scheme, a piecewise temperature strain compensation function is constructed. A database of material thermal expansion coefficients is used to establish compensation models for different temperature ranges for the steel or alloy material used. Temperature readings are used to correct the corresponding strain sensor, eliminating or reducing the influence of spurious deformation caused by thermal expansion. Let the temperature variable be denoted as Θ; the original measured strain as E; and the corrected strain as Ec. Let the temperature threshold be Θt. The piecewise compensation function f(Θ) is constructed: for the normal temperature range (Θ≤Θt), f(Θ)=C1+C2·Θ+C3·Θ 2 +C4·Θ 3 Where: C1; baseline compensation offset for normal temperature zone; C2; primary temperature compensation coefficient; C3; secondary temperature compensation coefficient; C4; tertiary temperature compensation coefficient, used for fine-tuning nonlinear response; Θ; current ambient temperature value; Θt; segmented threshold temperature, defining the normal temperature and high temperature ranges. High temperature range (Θ>Θt), f(Θ)=D1·exp(D2·(Θ-Θt)) t ))+D3·(Θ-Θ t +D4; where: D1; exponential compensation base coefficient; D2; exponential compensation growth rate coefficient; D3; high-temperature linear correction coefficient, reflecting the linear part of thermal expansion; D4; additional correction constant for the high-temperature zone; Θ; current ambient temperature value; Θt; piecewise threshold temperature, consistent with the normal temperature range. Based on this piecewise compensation function, the original strain Er is corrected to: E c =E r -f(Θ); where: E c ; Corrected strain; E r The sensor measures the original strain value; f(Θ); the temperature compensation function is calculated according to the above piecewise definition. The tilt angle change rate and displacement increment are calculated in real time, and dynamic correlation analysis is performed. The tilt angle change rate and displacement increment are obtained as follows: The tilt angle change rate is calculated from the tilt angle sequence continuously acquired by the attitude sensor, using the symbol X. l This represents the rate of change of tilt angle at time l within the sliding time window; Displacement increment: the incremental displacement value of the sliding component at the corresponding time, obtained through displacement sensors or position differential calculations, denoted by the symbol Y. lThis represents the displacement increment at time l within the sliding time window. Sliding time window: Defined to include L... s The time interval of continuous sampling points is used for dynamic analysis. The correlation analysis formula represents the Pearson correlation coefficient ρ calculated between the tilt angle change rate sequence and the displacement increment sequence within a sliding time window: Parameter description: X l : The rate of change of tilt angle at time l; Y l : The displacement increment at time l; L s : Total number of sampling points within the sliding time window; μ X : Mean value of the tilt angle change rate sequence; μ Y : Mean of the displacement increment sequence; ρ: Correlation coefficient between the rate of change of tilt and the displacement increment within the current time window. The first derivative of the rate of change of tilt is used for judgment, calculating the discrete first derivative sequence of the rate of change of tilt in real time, where l starts from 2; monitoring shows that if ΔX_l is positive for several consecutive sampling points, it indicates a continuous upward trend in the rate of change of tilt, potentially suggesting an intensified change in structural attitude. The instability triggering condition uses the correlation coefficient threshold ρ and the continuous increase of the first derivative to trigger a risk signal: when |ρ|>ρ1 and ΔX_l>0 is true for several recent moments, a potential instability trend is identified; ρ1: Correlation coefficient threshold, used to determine the high correlation between the rate of change of tilt and the displacement increment; ΔX_l: The value of the first derivative of the rate of change of tilt. Instability grading signal generation and output description: After the instability triggering condition is met, the risk level G can be further divided based on the absolute value of the correlation coefficient and the rate of increase of the rate of change of tilt. The level can be set according to the strategy, based on the combination of the maximum values ​​of |ρ| and ΔX_l. The system output includes the risk level parameter G and a list of associated dynamic anchor point spatial coordinate codes, which are used for subsequent control or early warning visualization.

[0030] Specifically, the logical process of inputting the steel box girder slip time series data and stability assessment results into a bidirectional long short-term memory neural network model to predict the cumulative deviation of the slip trajectory within the next ten seconds is as follows: One-dimensional convolutional noise reduction is performed on the raw data stream from the displacement sensors to extract key displacement feature waveforms, analyze the risk level parameters and spatial coordinate codes, generate regional sensitivity weighting coefficients, and fuse the track surface friction coefficient detection values ​​to construct a working condition feature vector; the weighted displacement sequence, spatial coordinate codes, and track friction coefficient vector are input into the bidirectional long short-term memory neural network, and an attention weight allocation mechanism is embedded through a gating unit to output the predicted value of the cumulative deviation within the next ten seconds.

[0031] In this implementation scheme, the convolutional noise reduction and feature extraction of the raw data stream from the displacement sensor are performed: the displacement data S acquired at high frequencies during the slip construction is processed... (τ)As the time series input, a one-dimensional convolutional neural network (1DCNN) is used to perform multi-scale kernel convolution operations, effectively filtering out transient disturbances and system noise, while extracting key displacement deformation waveform features to generate a feature sequence F. (τ) Analysis and weighted fusion of risk level and spatial coordinate coding: Extracting the risk level parameter R from stability assessment. (τ) Its value reflects the risk level of the current slip state, combined with the anchor point spatial coordinate encoding L. (τ) =[x (τ) ,y (τ) ,z (τ) ] Calculate the region sensitivity weighted coefficient vector W (τ) This is used to adjust the weights of subsequent input features in the attention mechanism. The formula for constructing the working condition feature vector is expressed as: X (τ) =Γ·(W (τ) ⊙F (τ) )+Λ·μ (τ) Where: X (τ) : The feature vector of the working condition at the current time step τ; W (τ) : Regional sensitivity weighted coefficient vector, generated from risk level and spatial location; F (τ) : The characteristic sequence of the displacement waveform after noise reduction; μ (τ) : Measured friction coefficient of the current slip segment; ⊙: Hadamard product, representing element-wise multiplication; Γ: Displacement feature weighting coefficient matrix; Λ: Linear fusion constant vector of friction coefficient. Bidirectional Long Short-Term Memory Neural Network (BiLSTM) predicts cumulative bias: X (τ) The input sequence is fed into a bidirectional LSTM model, recursively moving forward and backward in time to capture the slip history and future context information. The output feature sequence is aggregated through a fully connected layer to predict the cumulative deviation of the slip trajectory over the next ten seconds. The attention mechanism incorporates a weight allocation logic formula: for each time step of the BiLSTM hidden state h... (τ) An attention mechanism based on risk and sensitivity adjustment is introduced to assign prediction importance weights α. (τ) : Where: α (τ) Attention weights at time step τ; h (τ) : The hidden state vector output by the LSTM; W (τ) : Sensitivity weighted vector at time step τ; v: Attention score weight vector; Q: LSTM output projection matrix; K: Spatial weight projection matrix; tanh: Hyperbolic tangent activation function; exp: Exponential function; θ: Index variable, representing all time steps; ∑: Summation operation, used to normalize weights. The LSTM outputs at each time step are aggregated according to attention weights to obtain the cumulative bias prediction value for the next ten seconds. in: The cumulative deviation of the model's predicted slip trajectory over the next ten seconds; α (τ) : Weight coefficient, representing the importance of the feature at the current time step; h (τ) : LSTM hidden state; ∑ represents the weighted aggregation operation.

[0032] Specifically, the logical process for generating hydraulic jacking correction parameters when the cumulative deviation exceeds the safety threshold is as follows: the cumulative deviation is converted into normal pressure compensation and tangential speed adjustment at each jacking point based on a spatial vector decomposition algorithm; parameter generation is activated through a dual-condition triggering mechanism of static safety threshold and deviation growth rate dynamic threshold, and the output hydraulic jacking correction parameters include the target cylinder spatial topology code, control mode decision flag, and proportional-integral basic parameter set, wherein the proportional-integral basic parameter set includes the proportional coefficient reference value and the integral coefficient reference value.

[0033] In this implementation scheme, the cumulative deviation is decomposed into a space vector: the predicted slip cumulative deviation is then... Represented as a three-dimensional spatial vector, it is decomposed in the sliding coordinate system of the steel box girder to obtain the normal (vertical) component and the tangential (along the sliding axis) component in the direction of action of each hydraulic jacking unit, which serve as the basic input for subsequent control parameter generation. The formula for calculating the spatially decomposed control parameters is expressed as follows: Let the cumulative sliding deviation vector be: Then the direction vector of the unit force corresponding to any jacking control point λ is n. (λ) Its normal pressure compensation amount F can be calculated. (λ) With tangential speed adjustment V (λ) :F (λ) =κ1·(Δ (λ) ·n (λ) );V (λ) =κ2·||Δ (λ) -(Δ( λ )·n( λ) )·n (λ) ||;where: F (λ) : Normal pressure compensation at control point λ; V (λ) : Tangential velocity adjustment amount at control point λ; Δ (λ) : The cumulative deviation vector of control point λ; n (λ) : Unit direction vector of control point λ; · : Vector dot product; ∥·∥ : Vector magnitude; κ1 : Normal compensation gain coefficient, estimated from the compressive stiffness of the component; κ2 : Tangential adjustment gain coefficient, affected by friction characteristics. Triggering mechanism: Static + dynamic dual threshold judgment: The system adopts a combined triggering mechanism of "static threshold + deviation increase dynamic threshold". When one of the following two conditions is met, the generation of correction parameters is initiated: Ψ>Ξ s The cumulative deviation exceeds the static safety threshold; The deviation growth rate exceeds the dynamic rate threshold. Where: Ξ s This represents the structural safety limit deviation value; Ξ d This sets the maximum allowable speed increase for construction control. Hydraulic jacking correction parameter set generation: Once the correction condition is triggered, the system immediately generates a correction parameter set, mainly consisting of the following three parts: Target cylinder spatial topology encoding: Uniquely identifies the spatial position of each hydraulic cylinder control point in the three-dimensional topology, encoded in the form of C... (λ) Control mode decision flag: Determines whether to use "proportional control" or "proportional-integral control" based on the slip condition, and outputs flag M. (λ) ∈{P,PI}. 3. Calculation of the basic parameter set of proportional-integral (PI) controllers: If the current control point adopts PI mode, its proportional coefficient... Integral coefficient It can be dynamically adjusted based on the slip state: Formula expression: in: The proportional control coefficient for control point λ; Integral control coefficients for control point λ; θ1, θ2, θ3: weighting adjustment coefficients, which need to be adjusted according to the structural response sensitivity; Ξ s : Static safety threshold; Ξ d : Dynamic growth rate threshold; τ; Current time; τ'; Time integral variable.

[0034] Specifically, the process of mapping the cumulative deviation to the anchor point spatial coordinate set and dynamically mapping the stress distribution heatmap on the BIM model surface using the radial basis function interpolation algorithm is as follows: Based on the dynamic anchor point spatial coordinate set, a physical field coupling mapping relationship between the cumulative deviation and the corrected strain data is established; a continuous stress field distribution function is constructed using the radial basis function interpolation algorithm with the anchor points as control nodes; the stress transfer effect between adjacent anchor points is dynamically weighted using the Gaussian kernel function to generate a gradient stress heatmap on the BIM model surface; the slip trajectory change trend is integrated in real time, and high-stress areas are dynamically focused and rendered.

[0035] In this implementation plan, the deviation strain-stress coupling mapping relationship is constructed: based on the dynamic anchor point spatial coordinate set S={s (μ)}, where each anchor point μ has a spatial position and the corresponding cumulative deviation Compared with temperature-corrected strain data Based on the material constitutive relation, the stress estimates for the point set are established: Where: σ (μ) : Estimated axial stress at anchor point μ; E: Material elastic modulus, set based on actual batch of steel; The corrected anchor point strain value has been adjusted to account for temperature strain nonlinearity compensation; this stress value will be used as the objective function input in RBF interpolation. The RBF interpolation algorithm constructs the stress field function formula: based on the above anchor point stress value {σ... (μ) A radial basis function interpolation algorithm based on Gaussian kernel function is used to define a stress field distribution function on the surface of the BIM model. in: The estimated stress value at any position r; w (μ) : Weighting coefficient of anchor point μ, obtained based on stress value fitting; α: Gaussian kernel decay coefficient, controlling the locality of the interpolation function; ∥rs (μ) ∥: Euclidean distance between point r and anchor point μ; M: Total number of anchor points currently participating in interpolation; Weight w (μ) The stress value σ at all anchor points was calculated using the least squares method. (μ) The fitting yielded the following result: Dynamically Adjusting Thermal Weights Based on Trajectory Trends: To enhance the response to potential structural instability regions, a slip trajectory change trend field is introduced. The stress distribution weights are dynamically adjusted based on the trajectory variability rate. The updated interpolation function is as follows: in: Stress estimation value after trajectory trend enhancement; eβ: dynamic enhancement coefficient, reflecting the intensity of the influence of trajectory abrupt change on stress hotspots; The gradient modulus of the sliding trajectory at point r. Heatmap gradient rendering mechanism: Ultimately, the gradient modulus of the sliding trajectory at point r will be... The heatmap texture is generated by mapping the geometric mesh of the BIM model surface using an HSV tone mapping method. In this heatmap, red areas represent stress concentration areas (high values) and blue areas represent structural stress relaxation areas (low values). The heatmap is updated in real time for each frame based on the latest anchor point data, and supports dynamic focusing of areas and visual animation rendering.

[0036] Specifically, the process of marking collision risk coordinates in real time based on the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector is as follows: extract the spatial normal vectors of key structures including protective canopies and bridge piers from the obstacle library of the BIM model, calculate the dot product value of the current sliding direction vector of the steel box girder and the obstacle normal vector, and mark the three-dimensional risk cube in the anchor point coordinate system when the dot product result is greater than the preset collision risk coefficient and the real-time distance is less than the preset threshold; color-code the cube according to the risk level and mark the collision risk coordinates.

[0037] In this implementation plan, the obstacle normal vector set is extracted: all target components with rigid structural boundaries (such as protective canopies and bridge pier facades) are extracted from the predefined obstacle component library of the BIM model. For each obstacle component, the unit normal vector of its outer surface orientation is calculated to form the obstacle normal vector set. λ = 1, 2, ..., L; where: n( λ ): The unit normal vector of the λth obstacle component, pointing towards the side where a collision may occur; L: The total number of obstacles. Real-time calculation of the sliding direction vector: based on the centroid coordinates of the steel box girder at two consecutive moments on the current sliding path. Calculate the real-time slip direction unit vector: Where: d slide : The unit vector of the current sliding path direction; The coordinates of the slip center at two moments. The formula for determining collision risk based on the spatial dot product relationship is expressed as: for each obstacle normal vector n... (λ) Calculate its spatial dot product with the slip direction vector: γ (λ) =d slide ·n (λ) A region is identified as a potential collision risk area when the following dual criteria are met: Dot product criterion (angle less than threshold): γ (λ) >θ risk ;θ risk : Preset collision risk coefficient (usually set to cos(30°) or higher); indicates that the sliding direction tends towards the outer normal direction of the obstacle; real-time distance criterion: c (λ) : Coordinates of the nearest surface center of obstacle λ; d th Collision risk spatial distance threshold (set by the minimum structural tolerance spacing). Risk coordinate marking and visualization encoding: If both of the above conditions are met, the collision warning area will be marked as a three-dimensional risk cube in the coordinate system of the anchor point corresponding to the current sliding path. The risk level ρ is determined based on the magnitude of the dot product. (λ) The corresponding color rendering code is: γ (λ) >0.9: High risk (red); 0.7 <γ (λ) ≤0.9: Orange level, medium risk; θ risk <γ (λ) ≤0.7: Yellow - Low risk. Finally, risk cubes are marked in real-time within the BIM model interface. The three-dimensional spatial position and its color level are used to assist in adjusting the sliding path and providing early warning response.

[0038] Specifically, based on the hydraulic jacking correction parameters and collision risk coordinates, the specific process of constructing a response mechanism in which the proportional coefficient increases with the risk index and the integral coefficient dynamically adjusts with the prediction deviation is as follows: Based on the hydraulic jacking correction parameters, the reference values ​​of the proportional coefficient and the integral coefficient are extracted, and the target cylinder topology code and control mode flag are obtained simultaneously. The collision risk coordinates are converted into a spatial risk density index. According to the control mode flag, the corresponding proportional coefficient enhancement method is selected: under pressure control mode, the proportional coefficient increases geometrically with the risk index, and under speed control mode, the proportional coefficient increases linearly with the risk index. The integral coefficient is uniformly adjusted in reverse according to the cumulative deviation. When the deviation growth rate exceeds the critical slope, the integral action is locked, and the dynamic response parameter set bound to the cylinder topology code is output.

[0039] In this implementation plan, parameter extraction and spatial risk index calculation are performed by extracting the following key elements from the correction parameter package: Proportional coefficient benchmark value: Integral coefficient baseline value: Control cylinder topology coding: ξ (μ) Control mode flag: σ (μ) ∈{0,1}, where σ (μ) =0 indicates pressure control, σ (μ) =1 indicates speed control. Simultaneously, the spatial risk density index R centered on the cylinder control domain is calculated. (μ) The definition is as follows: Where: ρ (ν) : The grading coefficient of the νth risk cube (3 for high risk, 2 for medium risk, and 1 for low risk); c (ν) : Coordinates of the center point of the risk coordinate system; o (μ) : Coordinates of the action center of cylinder μ; N: Number of all risk cubes in the current area. This index is used to characterize the local risk density and its impact on the target cylinder. Proportional coefficient response mechanism construction: based on the control mode flag σ. (μ) The dynamic gain function that determines the proportional gain is: if σ (μ) =0 (Pressure control mode), using the geometric enhancement formula: If σ (μ) =1 (Speed ​​Control Mode), using the linear enhancement formula: in: The dynamic proportional gain after risk enhancement; qα; geometric enhancement factor, controlling the rate of increase of the gain; qβ; linear gain factor, setting the gain magnitude per unit of risk index. Integral coefficient response mechanism construction: real-time acquisition of the cumulative deviation Δ on the current cylinder control path. (μ) Its deviation rate of change The integral coefficients are adjusted in the following manner: in: Dynamically adjusted integral coefficient; η; integral suppression adjustment factor; Δ max The upper limit of the set deviation (normalized reference); The critical slope threshold for the rate of change of deviation; if the deviation increases too quickly (critical trend of instability), the integral action is automatically locked to prevent over-adjustment. Output dynamic response parameter set, combined with the topology code ξ for each cylinder. (μ) The following response parameter set will be output: This parameter set is updated in real time in the slip control system, driving the hydraulic control module to complete differentiated and precise adjustments, ensuring sufficient robustness in high-risk paths while avoiding over-response in low-deviation areas.

[0040] Specifically, the process of generating and sending the hydraulic cylinder pushing pressure or speed adjustment command to the pushing equipment controller for real-time correction is as follows: Parse the dynamic response parameter group and its bound cylinder topology code; classify and generate execution commands according to the control mode flag: for pressure control mode, convert the proportional coefficient and integral coefficient into a pressure correction amount containing integral operation; for speed control mode, convert the proportional coefficient into a speed adjustment amount; encapsulate the command data stream through the industrial real-time communication protocol; locate the target device according to the cylinder topology code; and send the command to the programmable controller.

[0041] In this implementation scheme, the dynamic response parameter set is parsed to extract variables from the response parameter set output in the previous step: Where: ξ (φ) : Hydraulic cylinder topology coding; Dynamic proportional coefficient; Dynamic integral coefficient; σ (φ) (∈{0,1}): Control mode flag (0 indicates pressure control mode, 1 indicates speed control mode). II. Execution instruction classification generates logical mode one: Pressure control mode (σ) (φ) =0) In pressure control mode, the command objective is to generate a corrected pressure increment to achieve force control correction. This is expressed using a proportional-integral control law as follows: Where: ΔF (φ) (t): Target pressure correction at time t; ε (φ) (t): Current cumulative deviation of the slip path (real-time error); t0: Correction start time. This correction is converted to the target hydraulic pressure value P. (φ) The calculation utilizes the linkage between the force-bearing area of ​​the hydraulic cylinder and the characteristics of the fluid medium. Mode 2: Speed ​​Control Mode (σ^(φ)=1) In speed control mode, only proportional control is used to generate the target speed: Where: V (φ) (t): Target velocity of the cylinder piston; ε(φ) (t): The deviation between the current position and the desired path (slip error); III. Instruction Encapsulation and Target Positioning 1. Industrial Communication Protocol Encapsulation encapsulates the above control quantities (ΔF) (φ) (t) or V (φ) (t) is encapsulated as a communication instruction data stream: Ψ (φ) =Pack(ξ (φ) ,χ (φ) (t),σ (φ) ,T (c) ); where: Ψ (φ) : Encapsulated communication commands; χ (φ) (t): Target adjustment value, ΔF in pressure control mode. (φ) (t), V in speed control mode (φ) (t); T (c) : Command execution timestamp or control cycle identifier. The protocol format follows real-time industrial communication standards such as Profinet, EtherCAT, and Modbus TCP / IP to ensure millisecond-level data synchronization. 2. Controller address mapping and command issuance are based on the cylinder topology encoding ξ. (φ) Locate the corresponding execution control unit (PLC or I / O control station) and set Ψ (φ) Write to the corresponding address channel to complete the instruction issuance.

[0042] Please see Figure 2 A BIM-based steel box girder slippage monitoring system includes the following modules: slippage assessment module, deviation early warning module, risk monitoring module, and closed-loop control module. The slippage assessment module uses a non-uniform rational B-spline surface to reconstruct and match the BIM model with the laser-scanned point cloud, generating a dynamic anchor point coordinate set at key sections of the steel box girder and binding the slippage track control points. It corrects strain measurements based on the nonlinear mapping relationship between temperature and strain, and assesses slippage stability by combining the correlation analysis between the tilt angle change rate and displacement increment. The deviation early warning module inputs the steel box girder slippage time-series data and stability assessment results into a bidirectional long short-term memory neural network model to predict slippage within the next ten seconds. The cumulative deviation of the trajectory is used to generate hydraulic jacking correction parameters when the cumulative deviation exceeds the safety threshold. The risk monitoring module maps the cumulative deviation to the anchor point spatial coordinate set, dynamically maps the stress distribution heat map on the BIM model surface through the radial basis function interpolation algorithm, and marks the collision risk coordinates in real time according to the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector. The closed-loop control module is used to construct a response mechanism based on the hydraulic jacking correction parameters and collision risk coordinates, in which the proportional coefficient increases with the risk exponential and the integral coefficient is dynamically adjusted with the prediction deviation. It generates adjustment commands for the hydraulic cylinder jacking pressure or speed and sends them to the jacking equipment controller to perform real-time correction.

[0043] In this implementation scheme, the slip evaluation module innovatively integrates the Non-Uniform Rational B-Spline Surface (NURBS) fitting mechanism with BIM control point topology consistency reconstruction technology, achieving high-precision matching between the laser-scanned point cloud and the BIM model. Unlike traditional methods using geometric bounding boxes or rigid matching algorithms, this scheme leverages NURBS surfaces to support complex curvature changes, constructs a dynamic anchor point control network, and spatially binds it to the slip track control points, thereby achieving full-process slip reference positioning within the high-order surface geometric domain. A material thermal expansion coefficient database is introduced to perform nonlinear correction on strain values ​​in different temperature ranges, combined with piecewise quadratic functions and exponential decay terms for modeling. This processing overcomes the limitations of traditional linear compensation mechanisms, ensuring the accuracy and reliability of strain variables under conditions such as high temperatures or alternating day and night. Deviation Early Warning Module: Utilizes a bidirectional long short-term memory neural network (Bi-LSTM) model to dynamically model slip time-series data, enabling deviation prediction of the slip trajectory in the next 10 seconds. Compared to traditional linear regression or unidirectional LSTM, it possesses stronger historical memory and forward-looking trend capabilities, making it particularly suitable for nonlinear systems like steel box girders where inertial lag and correction response delays coexist. Combining spatial coordinate encoding, friction coefficient detection values, and risk level parameters, a working condition feature vector matrix is ​​formed as network input, improving the model's sensitivity to local high-risk areas. When the predicted cumulative deviation exceeds the dynamic safety threshold, a set of correction parameters bound to the hydraulic cylinder encoding is automatically generated, ensuring rapid triggering of the closed-loop response between the prediction results and equipment execution. Risk Monitoring Module: Maps the cumulative deviation to the anchor point control network and constructs a continuous distribution image of BIM surface stress using the radial basis function (RBF) interpolation method. By adjusting the interpolation influence domain through a Gaussian kernel function, a smoother heatmap that conforms to the laws of physical transmission is achieved, overcoming the coarseness of traditional single-point value rendering methods. Based on the principle of vector dot product, the angle between the sliding direction and the normal vector of the obstacle surface is calculated in real time. A risk cube marker is triggered when the sliding path is about to orthogonally collide with the obstacle, and the risk level is reflected by color coding. The closed-loop control module constructs a control response mechanism that adjusts the proportional coefficient according to the spatial risk density index and the integral coefficient inversely adjusts according to the dynamic slope of the deviation, significantly outperforming traditional fixed PI parameter control logic. For the two control modes of pressure and speed, a differentiated strategy is used to generate commands separately: integral enhancement is used for pressure mode; proportional linear adjustment is used for speed mode.

[0044] In summary, this application has at least the following effects:

[0045] A BIM-based method and system for monitoring the sliding installation of steel box girders effectively improves the accuracy and reliability of dynamic anchor point positioning during the sliding process by matching the topology of the non-uniform rational B-spline surface with the BIM model. It introduces temperature-strain nonlinear compensation and tilt-displacement correlation analysis to enhance the scientific rigor and foresight of sliding stability assessment, significantly improving early anomaly identification capabilities. Based on bidirectional long short-term memory neural network-based sliding trajectory prediction, it achieves accurate early warning of future sliding deviations, providing a basis for risk management during the sliding process. Using radial basis function interpolation to construct stress heat maps and real-time collision risk marking of the sliding path, it achieves dynamic visualization of sliding stress distribution and spatial collision risk, improving the intuitiveness and practicality of monitoring. A dynamic linkage mechanism between proportional-integral control parameters and risk levels is constructed to effectively realize intelligent and adaptive adjustment of hydraulic jacking correction control, improving the accuracy and stability of correction response. Combined with industrial real-time communication protocols, it achieves precise issuance and execution of control commands, ensuring efficient and real-time response of the sliding correction closed-loop control, and improving the overall safety and automation level of the system.

[0046] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0047] This invention is described with reference to flowchart illustrations and / or block diagrams of systems, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0048] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The function specified in one or more boxes.

[0049] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0050] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0051] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A BIM-based method for monitoring the sliding installation of steel box girders, characterized in that, Includes the following steps: S1. By reconstructing and matching the BIM model and laser scanning point cloud through non-uniform rational B-spline surface, a dynamic anchor point coordinate set is generated at the key section of the steel box girder and the control points of the sliding track are bound. The strain measurement value is corrected according to the nonlinear mapping relationship of temperature and strain, and the sliding stability is evaluated by combining the correlation analysis of the tilt angle change rate and the displacement increment. S2. Input the sliding time series data of the steel box girder and the stability assessment results into the bidirectional long short-term memory neural network model to predict the cumulative deviation of the sliding trajectory in the next ten seconds. When the cumulative deviation exceeds the safety threshold, generate hydraulic jacking correction parameters. S3. Map the cumulative deviation to the anchor point spatial coordinate set, dynamically map the stress distribution heat map on the BIM model surface through the radial basis function interpolation algorithm, and mark the collision risk coordinates in real time according to the spatial dot product relationship between the slip path direction vector and the obstacle normal vector. S4. Based on the hydraulic jacking correction parameters and collision risk coordinates, a response mechanism is constructed in which the proportional coefficient increases with the risk index and the integral coefficient is dynamically adjusted with the prediction deviation. This mechanism generates adjustment commands for the hydraulic cylinder jacking pressure or speed and sends them to the jacking equipment controller to perform real-time correction. The specific process for evaluating slip stability based on correcting strain measurements according to the nonlinear mapping relationship of temperature and strain, combined with correlation analysis of the tilt angle change rate and displacement increment, is as follows: A piecewise temperature strain compensation function is constructed by calling the material thermal expansion coefficient database: the room temperature range is fitted by a quadratic polynomial, and the high temperature range introduces an exponential correction term based on the material thermal expansion characteristics. The dynamic Pearson correlation coefficient between the rate of change of tilt angle and the displacement increment within the sliding time window is calculated in real time. When the absolute value of the correlation coefficient exceeds the preset threshold and the first derivative of the rate of change of tilt angle remains positive, an instability classification signal is triggered and the risk level parameter and the spatial coordinate code of the associated anchor point are output. The logic process for generating hydraulic jacking correction parameters when the cumulative deviation exceeds the safety threshold is as follows: Based on the spatial vector decomposition algorithm, the cumulative deviation is converted into the normal pressure compensation and tangential velocity adjustment at each jacking point; The parameter generation is activated by a dual-condition triggering mechanism of static safety threshold and deviation growth rate dynamic threshold. The output includes hydraulic jacking correction parameters containing target cylinder space topology code, control mode decision flag and proportional-integral basic parameter set, where the proportional-integral basic parameter set includes proportional coefficient reference value and integral coefficient reference value. Based on the hydraulic jacking correction parameters and collision risk coordinates, the specific process of constructing a response mechanism in which the proportional coefficient increases with the risk index and the integral coefficient dynamically adjusts with the prediction deviation is as follows: Based on the hydraulic jacking correction parameters, the proportional coefficient reference value and integral coefficient reference value are extracted. At the same time, the target cylinder topology code and control mode flag are obtained, and the collision risk coordinates are converted into the spatial risk density index. Select the appropriate proportional coefficient enhancement method based on the control mode flag: under pressure control mode, the proportional coefficient increases geometrically with the risk index; under speed control mode, the proportional coefficient increases linearly with the risk index. The integral coefficient is uniformly adjusted in the opposite direction based on the cumulative deviation. When the deviation growth rate exceeds the critical slope, the integral action is locked, and the output is a set of dynamic response parameters bound to the cylinder topology code.

2. The BIM-based method for monitoring the sliding installation of steel box girders according to claim 1, characterized in that: The specific process of generating a dynamic anchor point coordinate set and binding the sliding track control points at key sections of the steel box girder by reconstructing and matching the BIM model and laser scanning point cloud through non-uniform rational B-spline surface is as follows: The laser scanning point cloud is subjected to curvature feature-based partitioned filtering to extract the feature point cloud clusters of the web and top and bottom plates of the steel box girder; An adaptive surface control network with the same topology as the control points of the BIM model is constructed using a non-uniform rational B-spline algorithm, and registration is performed through iterative nearest point optimization. Dynamic anchor point clusters are generated at equal intervals along the sliding axis at key sections of the registered BIM model. Each anchor point is assigned a unique spatial identifier code. The coordinates of the track control points measured by the total station are simultaneously integrated to establish a two-way coordinate transformation mapping relationship between the anchor points and the track control points, forming a dynamic spatial reference network for the entire sliding process.

3. The BIM-based method for monitoring the sliding installation of steel box girders according to claim 2, characterized in that: The logical process of inputting the slip time series data of the steel box girder and the stability assessment results into a bidirectional long short-term memory neural network model to predict the cumulative deviation of the slip trajectory within the next ten seconds is as follows: One-dimensional convolution noise reduction is performed on the raw data stream of the displacement sensor to extract key displacement feature waveforms, analyze the risk level parameters and spatial coordinate codes, generate regional sensitivity weighting coefficients, and fuse the track surface friction coefficient detection values ​​to construct a working condition feature vector. The weighted displacement sequence, spatial coordinate encoding, and track friction coefficient vector are input into a bidirectional long short-term memory neural network. An attention weight allocation mechanism is embedded through a gating unit to output the predicted value of the cumulative deviation for the next ten seconds.

4. The BIM-based method for monitoring the sliding installation of steel box girders according to claim 3, characterized in that: The specific process of mapping the accumulated deviation to the anchor point spatial coordinate set and dynamically mapping the stress distribution heatmap on the BIM model surface using the radial basis function interpolation algorithm is as follows: Based on the dynamic anchor point spatial coordinate set, a physical field coupling mapping relationship between the cumulative deviation and the corrected strain data is established. By using the radial basis function interpolation algorithm, a continuous stress field distribution function is constructed with anchor points as control nodes. The stress transfer effect between adjacent anchor points is dynamically weighted by the Gaussian kernel function, and a gradient stress heat map is generated on the surface of the BIM model. The slip trajectory change trend is integrated in real time, and high-stress areas are dynamically focused and rendered.

5. The BIM-based method for monitoring the sliding installation of steel box girders according to claim 4, characterized in that: The specific process of marking the collision risk coordinates in real time based on the spatial dot product relationship between the sliding path direction vector and the obstacle normal vector is as follows: Extract spatial normal vectors of key structures including protective canopies and bridge piers from the obstacle library of the BIM model, calculate the dot product of the current sliding direction vector of the steel box girder and the obstacle normal vector, and mark the three-dimensional risk cube in the anchor point coordinate system when the dot product result is greater than the preset collision risk coefficient and the real-time distance is less than the preset threshold. The cube is rendered with color coding based on the risk level, and the collision risk coordinates are marked.

6. The BIM-based method for monitoring the sliding installation of steel box girders according to claim 5, characterized in that: The specific process of generating adjustment commands for the hydraulic cylinder pushing pressure or speed and sending them to the pushing equipment controller to perform real-time correction is as follows: The dynamic response parameter group and its bound cylinder topology code are analyzed, and execution instructions are generated according to the control mode flag: the proportional coefficient and integral coefficient are converted into pressure correction amount with integral operation in the pressure control mode, and the proportional coefficient is converted into speed adjustment amount in the speed control mode. The instruction data stream is encapsulated using an industrial real-time communication protocol, and the target device is located and sent to the programmable controller based on the cylinder topology code.

7. A BIM-based steel box girder sliding installation monitoring system, applied to any one of the BIM-based steel box girder sliding installation monitoring methods according to claims 1-6, characterized in that, It includes the following modules: slip assessment module, deviation early warning module, risk monitoring module, and closed-loop control module; The slip assessment module is used to reconstruct and match the BIM model and laser scanning point cloud through non-uniform rational B-spline surface reconstruction, generate a dynamic anchor point coordinate set on the key section of the steel box girder and bind the slip track control points, correct the strain measurement value according to the nonlinear mapping relationship of temperature and strain, and evaluate the slip stability by combining the correlation analysis of the tilt angle change rate and displacement increment. The deviation early warning module is used to input the sliding time series data of the steel box girder and the stability assessment results into the bidirectional long short-term memory neural network model to predict the cumulative deviation of the sliding trajectory within the next ten seconds. When the cumulative deviation exceeds the safety threshold, hydraulic jacking correction parameters are generated. The risk monitoring module is used to map the cumulative deviation to the anchor point spatial coordinate set, dynamically map the stress distribution heat map on the BIM model surface through the radial basis function interpolation algorithm, and mark the collision risk coordinates in real time according to the spatial dot product relationship between the slip path direction vector and the obstacle normal vector. The closed-loop control module is used to construct a response mechanism based on the hydraulic jacking correction parameters and collision risk coordinates. The proportional coefficient increases with the risk index and the integral coefficient is dynamically adjusted with the prediction deviation. It generates adjustment commands for the hydraulic cylinder jacking pressure or speed and sends them to the jacking equipment controller to perform real-time correction.

Images