A method and system for steel structure installation and adjustment based on dynamic tracking of column axis.

Abstract

This application relates to the field of image processing technology and discloses a method and system for adjusting steel structure installation based on dynamic tracking of column axes. The method includes: acquiring images of steel structure columns to obtain contour data; segmenting the images to extract column position features; matching the actual and designed column positions to calculate the deviation; analyzing the deviation trajectory to generate a displacement distribution map; calculating adjustment parameters to form instruction data; and constructing a virtual model to verify the adjustment effect and optimize the parameters. This application establishes a complete closed-loop system from measurement, analysis, optimization to verification through the comprehensive application of dynamic tracking of column axes, image processing, and multi-objective optimization algorithms, achieving precise control and intelligent adjustment of steel structure column installation.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method and system for adjusting the installation of steel structures based on dynamic tracking of column axes. Background Technology

[0002] Steel structure buildings are widely used in modern industrial plants, commercial buildings, and high-rise buildings due to their advantages such as short construction cycle, good seismic performance, and high strength-to-weight ratio. During steel structure construction, the columns, as the main load-bearing components, directly affect the safety and service life of the entire structure through their installation accuracy. Traditional steel structure column installation methods mainly rely on manual measurement and experience-based adjustments. Typically, theodolites or total stations are used to measure the spatial position of the columns, and deviations are manually calculated based on the measurement data. Adjustments are then made using hydraulic equipment or jacks. In recent years, with the development of 3D laser scanning technology and BIM technology, some advanced steel structure construction companies have begun to adopt digital measurement methods, improving measurement accuracy and efficiency. However, the adjustment process still mainly relies on manual experience and simple calculation tools.

[0003] However, existing technologies have significant shortcomings. First, traditional manual measurement and adjustment methods have limited accuracy, making it difficult to meet the increasingly stringent precision requirements for steel structure installation. Second, while current digital measurement methods improve accuracy, they lack real-time dynamic tracking capabilities, failing to effectively monitor changes in column position during the adjustment process. Third, existing adjustment methods typically involve static adjustments to individual columns, lacking consideration for the interrelationships within the overall structure, potentially causing new deviations in other areas when adjusting one location. Fourth, the lack of systematic adjustment optimization algorithms results in low efficiency and a high dependence on operator experience. Finally, existing methods lack effective verification and feedback mechanisms, making it difficult to guarantee the consistency and reliability of the final installation quality. These shortcomings lead to low installation accuracy, inefficiency, and increased costs for steel structures, severely impacting the overall quality and schedule of steel structure projects. Summary of the Invention

[0004] This application provides a steel structure installation adjustment method and system based on dynamic tracking of column axis. It is used to establish a complete closed-loop system from measurement, analysis, optimization to verification through the comprehensive application of dynamic tracking of column axis, image processing and multi-objective optimization algorithm, so as to realize precise control and intelligent adjustment of steel structure column installation.

[0005] In a first aspect, this application provides a steel structure installation and adjustment method based on dynamic tracking of column axes. The method includes: acquiring images of steel structure columns to obtain an image dataset containing column outlines, target point features, temporal information, and environmental parameters; performing image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing column position nodes and connection relationships; establishing a computer vision matching algorithm based on the temporal feature map to obtain the correspondence between the actual column image and the design model, and visual deviation data; performing bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain a column displacement vector field and a three-dimensional deviation distribution map; applying an image processing algorithm to calculate column adjustment parameters based on the three-dimensional deviation distribution map to obtain adjustment command data containing a spatial transformation matrix; constructing a virtual guidance model based on the adjustment command data and performing real-time image feedback verification to obtain adjustment accuracy evaluation results and model optimization parameters.

[0006] Secondly, this application provides a steel structure installation adjustment system based on dynamic tracking of column axes, the steel structure installation adjustment system based on dynamic tracking of column axes includes:

[0007] The acquisition module is used to acquire data from images of steel structure columns, and obtain an image dataset containing column outlines, target point features, time series information and environmental parameters.

[0008] The extraction module is used to perform image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing the column location nodes and connection relationships;

[0009] The matching module is used to establish a computer vision matching algorithm based on the time-series feature map to obtain the correspondence between the actual column image and the design model and the visual deviation data.

[0010] The analysis module is used to perform bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map;

[0011] The calculation module is used to calculate the column adjustment parameters based on the three-dimensional deviation distribution map using an image processing algorithm, and obtain adjustment instruction data including a spatial transformation matrix;

[0012] The verification module is used to construct a virtual guidance model based on the adjustment instruction data and perform real-time image feedback verification to obtain the adjustment accuracy evaluation results and model optimization parameters.

[0013] Thirdly, a computer device is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor invokes the instructions in the memory to cause the computer device to execute the above-described steel structure installation and adjustment method based on dynamic tracking of column axis.

[0014] Fourthly, a computer-readable storage medium is provided, wherein instructions are stored therein, which, when executed on a computer, cause the computer to perform the above-described steel structure installation and adjustment method based on dynamic tracking of column axes.

[0015] The technical solution provided in this application acquires image data of steel structure columns, obtaining an image dataset containing column outlines, target point features, temporal information, and environmental parameters, thus solving the problems of insufficient accuracy and incomplete data in traditional measurement methods. Image segmentation and feature extraction are performed on the image dataset to obtain a temporal feature map representing the column position nodes and connection relationships, achieving an accurate description of the column's spatial position and interrelationships. A computer vision matching algorithm is established based on the temporal feature map to obtain the correspondence between the actual column image and the design model, as well as visual deviation data, making the matching process between the column and the design position more accurate and reliable. Two-way spatiotemporal trajectory analysis is performed on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map, comprehensively understanding the dynamic characteristics of column position changes. Based on the three-dimensional deviation distribution map, image processing algorithms are applied to calculate column adjustment parameters, obtaining adjustment command data containing a spatial transformation matrix, achieving accurate conversion from measurement data to a specific adjustment scheme. A virtual guidance model is constructed based on the adjustment command data, and real-time image feedback verification is performed to obtain adjustment accuracy evaluation results and model optimization parameters, forming a complete closed-loop verification and continuous optimization mechanism. Of particular note is the innovative application of various artificial intelligence algorithms in the field of steel structure installation. These include improved accuracy of computer vision matching algorithms in column identification and matching, effective capture of dynamic changes through bidirectional spatiotemporal trajectory analysis algorithms, and intelligent adjustment strategy generation based on multi-objective optimization and reinforcement learning. These algorithms contribute to the solution in two ways: firstly, they solve the problem of complex column interaction effects that traditional methods cannot handle, improving the coordination of overall structural adjustments; secondly, through continuous optimization of adjustment parameters via machine learning, they accumulate and apply experiential knowledge, significantly improving adjustment efficiency and accuracy. In summary, this solution, by constructing a complete technology chain from data acquisition, feature extraction, matching analysis to adjustment verification, and combining the advantages of artificial intelligence algorithms, achieves precision, intelligence, and automation in the steel structure column installation process, effectively solving the technical problems of insufficient accuracy, low efficiency, and unstable quality in traditional steel structure installation. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of an embodiment of the steel structure installation and adjustment method based on dynamic tracking of column axis in this application.

[0018] Figure 2 This is a schematic diagram of one embodiment of the steel structure installation and adjustment system based on dynamic tracking of column axis in this application.

[0019] Figure 3 This is a schematic block diagram of the structure of the computer device in an embodiment of the present invention. Detailed Implementation

[0020] This application provides a method and system for adjusting the installation of steel structures based on dynamic tracking of column axes. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0021] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the steel structure installation and adjustment method based on dynamic tracking of column axis in this application includes:

[0022] Step S101: Collect data from the steel structure column images to obtain an image dataset containing column outlines, target point features, temporal information, and environmental parameters;

[0023] Step S102: Perform image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing the column location nodes and connection relationships;

[0024] Step S103: Establish a computer vision matching algorithm based on the temporal feature map to obtain the correspondence between the actual column image and the design model and the visual deviation data;

[0025] Step S104: Perform bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map;

[0026] Step S105: Calculate the column adjustment parameters using an image processing algorithm based on the three-dimensional deviation distribution map to obtain adjustment instruction data containing the spatial transformation matrix;

[0027] Step S106: Construct a virtual guidance model based on the adjustment instruction data and perform real-time image feedback verification to obtain the adjustment accuracy evaluation results and model optimization parameters.

[0028] It is understood that the executing entity of this application can be a steel structure installation and adjustment system based on dynamic tracking of column axis, or it can be a terminal or a server; the specific implementation is not limited here. This application's embodiment uses a server as an example for illustration.

[0029] Specifically, multi-view high-definition cameras are deployed at the construction site to capture reflective target points on the surface of the pillars. The cameras are typically positioned at different locations on the site to ensure comprehensive coverage of each pillar. The acquired images undergo geometric correction to eliminate lens distortion, forming a corrected image sequence. Simultaneously, the system records the timestamp of each frame, aligning images acquired by different cameras according to time points. The images also undergo background separation processing to extract pillar outlines and target point features. Furthermore, ambient light, temperature, and vibration data are collected and correlated with the image data to form an environmental influence factor matrix. The final image dataset contains pillar outlines, target point features, temporal information, and environmental parameters. Image segmentation and feature extraction are then performed based on the image dataset. This step first uses adaptive watershed segmentation to segment the pillar outlines, obtaining a pillar region mask. The watershed algorithm treats the image as a terrain surface, where high-gray-value areas form "ridges" and low-gray-value areas form "valleys," segmenting the image into different regions through a "flooding" process. From the segmented column region mask image, the centerline of the column is extracted and the axis equation is calculated using the Hough transform algorithm. The Hough transform identifies straight-line structures in the image through a voting method in parameter space, making it particularly suitable for axis extraction of regular geometric shapes such as steel structure columns. Based on the target point features and the column axis equation, column position node data containing three-dimensional spatial coordinates and attitude angles is generated. Then, the column position node data is used to construct a steel structure topology graph, and the connection relationship between adjacent columns is identified using a spatial distance threshold algorithm. The column position node data and connection relationships are correlated with temporal information to generate a temporal node feature sequence, which is then subjected to temporal domain filtering to eliminate abnormal jitter, and finally, a temporal feature map is synthesized.

[0030] When establishing a computer vision matching algorithm based on time-series feature maps, the design drawings are imported into the computer, and the column design axis data is extracted. This data is then converted into a design feature map in the image space coordinate system through coordinate transformation. Feature descriptors are extracted from both the time-series and design feature maps, and a feature index table is constructed using the Locality Sensitive Hashing (LSH) algorithm. LSH hashes similar data points into the same "bucket," accelerating the feature matching process. The feature similarity matrix is ​​calculated using the feature index table, and the maximum weighted binary matching problem is solved using an improved Hungarian algorithm. The Hungarian algorithm, a combinatorial optimization algorithm used to solve allocation problems, is used in this method to determine the optimal match between the actual and designed columns. The correspondence between the actual and designed columns is obtained from the matching results, generating a column pairing table. The positional and angular differences of each pair of columns in the pairing table are calculated to generate raw deviation data. This raw deviation data is then statistically filtered and spatially interpolated to form visual deviation data. When performing bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data, a time-series data matrix is ​​first established, and the main change patterns are extracted using principal component analysis. Principal component analysis (PCA) reduces data dimensionality and highlights key changes. The trajectory of the column's position change is calculated forward over time, and random noise is eliminated using a Kalman filter to generate a forward displacement vector. Kalman filtering is a recursive filtering algorithm capable of estimating the state of a dynamic system from noisy measurements. Similarly, the trajectory of the column's position change is calculated backward over time, and Bayesian estimation improves the accuracy of backward prediction, generating a backward displacement vector. The forward and backward displacement vectors are weighted and fused to generate a column displacement vector field. Spatial interpolation of the displacement vector field is performed, and a continuous displacement field representation is constructed using radial basis functions (RBFs). RBF interpolation can handle irregularly distributed data points, generating a smooth, continuous surface. The spatial distribution of positional deviations is calculated based on the displacement field and the original position data, generating a three-dimensional deviation distribution map.

[0031] When calculating column adjustment parameters based on the 3D deviation distribution map, gradient analysis is first performed on the map to obtain the deviation change rate and main deformation direction through directional derivative calculation. A multi-objective optimization function is constructed based on the deviation change rate, comprising three sub-objectives: minimizing column position deviation, minimizing adjustment amount, and maximizing structural stability. The multi-objective optimization function is solved using a particle swarm optimization algorithm to obtain a Pareto optimal solution set balancing the objectives. The particle swarm optimization algorithm, which simulates bird flock foraging behavior to find the optimal solution, is suitable for handling multi-objective optimization problems. Column adjustment vectors are extracted from the Pareto optimal solution set and their feasibility is screened based on engineering constraints. Rigid body transformation matrices are calculated on the screened column adjustment vectors to generate a spatial transformation matrix containing translation and rotation components. The spatial transformation matrix is ​​converted into executable adjustment instructions, forming adjustment instruction data. A virtual guidance model is constructed based on the adjustment instruction data, and real-time image feedback verification is performed. First, a virtual contour of the column target position is constructed based on the adjustment instruction data, and then augmented reality technology is used to overlay the virtual contour onto a real-time image. Column adjustment operations are implemented at the construction site, and real-time image acquisition records the changes in column position during the adjustment process. The actual positions of the columns in the real-time images are registered with the virtual contours, and the positional overlap index is calculated. An adjustment accuracy assessment report, including residual deviation and spatial distribution characteristics, is generated based on the positional overlap index. Based on the residual deviation and historical adjustment data, the adjustment coefficients are updated using a reinforcement learning algorithm to generate model optimization parameters. These model optimization parameters are fed back to the column adjustment database, forming a closed-loop adjustment optimization mechanism to complete the entire steel structure installation and adjustment process.

[0032] In this embodiment, by acquiring images of steel structure columns, an image dataset containing column outlines, target point features, temporal information, and environmental parameters is obtained, solving the problems of insufficient accuracy and incomplete data in traditional measurement methods. Image segmentation and feature extraction are performed on the image dataset to obtain a temporal feature map representing the column position nodes and connection relationships, achieving an accurate description of the column's spatial position and interrelationships. A computer vision matching algorithm is established based on the temporal feature map to obtain the correspondence between the actual column image and the design model, as well as visual deviation data, making the matching process between the column and the design position more accurate and reliable. Two-way spatiotemporal trajectory analysis is performed on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map, comprehensively understanding the dynamic characteristics of column position changes. Based on the three-dimensional deviation distribution map, image processing algorithms are applied to calculate column adjustment parameters, obtaining adjustment command data containing a spatial transformation matrix, achieving accurate conversion from measurement data to specific adjustment schemes. A virtual guidance model is constructed based on the adjustment command data and real-time image feedback verification is performed to obtain adjustment accuracy evaluation results and model optimization parameters, forming a complete closed-loop verification and continuous optimization mechanism. Of particular note is the innovative application of various artificial intelligence algorithms in the field of steel structure installation. These include improved accuracy of computer vision matching algorithms in column identification and matching, effective capture of dynamic changes through bidirectional spatiotemporal trajectory analysis algorithms, and intelligent adjustment strategy generation based on multi-objective optimization and reinforcement learning. These algorithms contribute to the solution in two ways: firstly, they solve the problem of complex column interaction effects that traditional methods cannot handle, improving the coordination of overall structural adjustments; secondly, through continuous optimization of adjustment parameters via machine learning, they accumulate and apply experiential knowledge, significantly improving adjustment efficiency and accuracy. In summary, this solution, by constructing a complete technology chain from data acquisition, feature extraction, matching analysis to adjustment verification, and combining the advantages of artificial intelligence algorithms, achieves precision, intelligence, and automation in the steel structure column installation process, effectively solving the technical problems of insufficient accuracy, low efficiency, and unstable quality in traditional steel structure installation.

[0033] In one specific embodiment, the process of performing step S101 may specifically include the following steps:

[0034] (1) Set up multi-view high-definition camera equipment around the steel structure column and capture the reflective target points on the surface of the column through image tracking algorithm;

[0035] (2) Perform geometric correction on the acquired image sequence, and eliminate lens distortion through a binocular vision calibration algorithm to obtain the corrected image sequence;

[0036] (3) Extract timestamp information from the corrected image sequence and align the images acquired by different camera devices according to time points using a time synchronization algorithm;

[0037] (4) Perform background separation on time-aligned images and extract column contours and target point features using an adaptive threshold segmentation algorithm;

[0038] (5) Collect ambient light, temperature and vibration parameter data, and associate them with image data through data fusion algorithm to form an environmental impact factor matrix;

[0039] (6) Integrate the column outline, target point features, time series information and environmental influence factor matrix to obtain the image dataset.

[0040] Specifically, multi-view high-definition camera equipment is installed around the steel structure columns. This equipment typically includes multiple high-definition industrial cameras positioned at different locations on the construction site to ensure adequate visual coverage of each steel structure column. Reflective target points are pre-attached to the column surfaces. These target points are circular or square markers made of a specially formulated highly reflective material, maintaining good visibility under various lighting conditions. Image tracking algorithms, a technique in computer vision, track the movement of target objects by matching feature points between consecutive frames. In this method, the image tracking algorithm extracts these target points from the background by recognizing their brightness and shape features, recording their position coordinates in the image. This coordinate data forms the basis for subsequent calculations of the column axis.

[0041] The acquired raw image sequences often contain distortions, requiring geometric correction. Lens distortion is an image deformation caused by the non-ideal characteristics of the camera's optical system, mainly including radial and tangential distortion. Binocular vision calibration algorithms use a calibration board (usually a checkerboard pattern) to capture multiple sets of images at different positions and angles, calculating the camera's intrinsic parameter matrix and distortion coefficients. The intrinsic parameter matrix contains the focal length and principal point coordinates, while the distortion coefficients describe the mathematical model of the distortion. Using these parameters, each frame of the image undergoes anti-distortion processing to obtain the corrected image sequence. In the corrected images, straight lines still appear as straight lines, and the angle and distance ratios remain unchanged, laying the foundation for subsequent accurate measurements. Extracting timestamp information from the corrected image sequence is a crucial step in establishing temporal relationships. Timestamps are precise time records of each image frame's acquisition, typically with millisecond-level accuracy. Since multiple cameras are used simultaneously at the construction site, slight time differences may exist between the devices. Time synchronization algorithms establish a unified time reference system by analyzing the timestamps in the image sequences acquired by all devices. The specific method involves selecting one camera device as the master clock, and aligning the time series of other devices to the master clock's time point using linear interpolation or spline interpolation methods. After this processing, images captured by all cameras at the same time point are organized together to form a time-aligned multi-view image group.

[0042] Background separation of time-aligned images is a crucial step in extracting column information. The purpose of background separation is to isolate columns and target points from the complex background of a construction site. An adaptive thresholding algorithm dynamically adjusts the threshold value based on the local characteristics of the image, adapting to the segmentation needs under different lighting conditions. This algorithm first calculates the image's grayscale histogram, then finds the optimal threshold through an iterative method, classifying image pixels into foreground (columns and target points) and background. For reflective target points, due to their high reflectivity, they appear as areas with high brightness in the image and can be extracted by setting a high brightness threshold. The extracted column outlines typically appear as connected regions, while target points appear as discrete highlights. These features are further refined through region connectivity analysis and shape analysis to obtain accurate column outlines and target point features.

[0043] Environmental factors significantly impact the installation accuracy of steel structures, necessitating the collection of data on parameters such as ambient light, temperature, and vibration. Ambient light is measured using light sensors to record changes in light intensity at the construction site; temperature is measured using temperature sensors to record ambient temperature and column surface temperature; and vibration is measured using accelerometers to record the vibration frequency and amplitude at the construction site. The correlation between these environmental parameter data and image data requires a data fusion algorithm. This algorithm first aligns all environmental parameters by timestamp, then calculates the correlation between environmental parameters and image features, establishing a model of the impact of environmental factors on column position changes. This correlation analysis is achieved through multiple regression or neural network methods, ultimately forming an environmental impact factor matrix that describes the degree of influence of different environmental factors on the column position and shape.

[0044] The final step in the data acquisition phase is to integrate the column outlines, target point features, time-series information, and environmental influence factor matrices to form the final image dataset. This integration process requires establishing a unified data structure to organize various data types according to their temporal and spatial relationships. Image datasets are typically stored as multidimensional arrays or structured databases, containing the spatial coordinates of the columns, the location information of the target points, time-series data, and corresponding environmental parameters. The data structure design must consider the needs of subsequent processing, such as feature extraction and matching analysis, ensuring data integrity and accessibility.

[0045] For example, in a large steel structure project, the construction team deployed four high-definition industrial cameras around each column, covering a 360-degree view. Each column surface was affixed with 12 reflective target points distributed according to a specific geometric pattern. Image tracking algorithms recorded the real-time status of the columns by identifying the positions of these target points. The original images contained approximately 3% distortion; after correction using a binocular vision calibration algorithm to calculate camera parameters, the distortion was reduced to below 0.2%. The time synchronization accuracy of the four cameras reached 5 milliseconds, ensuring temporal consistency across multiple viewpoints. Background separation processing extracted the column outlines and target points from the complex environment, achieving an accuracy rate exceeding 98%. Simultaneously recorded environmental data showed that for every 1 degree Celsius change in ambient temperature, the column length changed by approximately 0.01 millimeters; this data was integrated into an environmental impact factor matrix. The resulting image dataset contained dynamic column tracking information.

[0046] In one specific embodiment, the process of performing step S102 may specifically include the following steps:

[0047] (1) Adaptive watershed segmentation is performed on the column outlines in the image dataset to obtain the column region mask map;

[0048] (2) Extract the center line of the column from the mask image of the column area, and calculate the equation of the column axis using the Hough transform algorithm;

[0049] (3) Generate column position node data based on the target point characteristics and column axis equation, wherein each node contains three-dimensional spatial coordinates and attitude angle;

[0050] (4) Construct a steel structure topology diagram using column location node data, and identify the connection relationship between adjacent columns using a spatial distance threshold algorithm;

[0051] (5) Associate the column location node data and connection relationship with the time series information to generate a time series node feature sequence;

[0052] (6) Perform time-domain filtering on the time-series node feature sequences to eliminate abnormal jitter and synthesize time-series feature maps.

[0053] Specifically, the adaptive watershed segmentation algorithm treats the image as a terrain surface, with areas of high grayscale forming "ridges" and areas of low grayscale forming "valleys." The algorithm first calculates the gradient of the image to obtain a gradient image, and then applies a watershed transform to this gradient image. In the processing of steel structure column images, gradient calculation is performed using the Sobel operator or the Canny edge detector to highlight the column's edge features. To avoid over-segmentation, the algorithm also introduces a marker control mechanism, using the rough location of the column outline extracted in the previous step as initial marker points. The watershed algorithm simulates a rising water level, "flooding" the gradient image from the marker points, establishing a watershed line when different water areas are about to meet, forming the segmentation boundary. After processing, a column region mask image is obtained, which is a binary image where pixels with a value of 1 represent the column region and pixels with a value of 0 represent the background.

[0054] Extracting the center line of the pillar from the pillar region mask and then calculating the pillar axis equation using the Hough transform algorithm is a crucial step in accurately locating the pillar. The Hough transform is a feature extraction technique used to detect straight lines, circles, or other parametric shapes in an image. In pillar axis extraction, the pillar region mask is first thinned to obtain a skeleton line with a single pixel width. Then, the Hough transform is applied to transform the points in the image space to the parameter space, and straight lines are identified through a voting mechanism in the parameter space. The mathematical basis of the Hough transform is the parametric equation of a straight line:

[0055]

[0056] in, This represents the distance from the line to the origin of the coordinate system. This represents the angle between the straight line normal and the x-axis. During processing, for each foreground pixel in the image, all possible angles are calculated. , The parameters are combined and accumulated at corresponding positions in the parameter space. The peak points in the parameter space correspond to straight lines in the original image. In the extraction of the column axis, the parameter space is discretized into an accumulator matrix. ,in and They are and The discretized value is obtained. The parameters of the column axis are determined by finding local maxima in the accumulator matrix. For the column axis in three-dimensional space, the following parametric equation can be used:

[0057]

[0058] in, It is the position vector of a point on the axis. is the direction vector of the axis, and t is a parameter variable. The axis parameters in three-dimensional space are calculated by triangulating the two-dimensional axes extracted from multiple viewpoint images.

[0059] Generating column position node data based on target point features and column axis equations is fundamental to constructing the spatial model of the steel structure. Target point features contain the position information of each target point in the image; through the geometric relationships of multi-view cameras, the coordinates of the target points in three-dimensional space can be calculated. Column position node data is a data structure describing the spatial position and orientation of the columns; each node contains three-dimensional spatial coordinates (X, Y, Z) and orientation angle (...). ), representing the position of the center point of the column and the spatial direction of the column axis, respectively. The calculation process for the node data is as follows: First, the spatial direction of the column is determined according to the column axis equation. Then, the three-dimensional coordinates of the target point are fitted using the least squares method to determine the center position of the column. The attitude angle is calculated by the angle between the direction vector of the column axis and the coordinate axis.

[0060] Constructing a steel structure topology diagram using column location node data is a crucial process for analyzing the connections between columns. A topology diagram is a graph structure where nodes represent columns and edges represent connections between them. The construction process first creates a set of nodes based on the column location node data, then uses a spatial distance threshold algorithm to determine if connections exist between columns. This algorithm calculates the Euclidean distance between any two column nodes; if the distance is less than a preset threshold and conforms to the connection rules in steel structure design, a connection edge is established between these two nodes. The threshold setting considers the requirements for connector lengths in steel structure design codes, as well as the impact of measurement errors. The topology diagram not only contains spatial location information but also records attributes such as connection type and connection strength, comprehensively describing the spatial configuration of the steel structure.

[0061] Associating column position node data and their connections with temporal information to generate a temporal node feature sequence is fundamental to analyzing the time-varying characteristics of steel structures. The temporal node feature sequence records the position and orientation information of all columns at each time point, as well as the connection relationships between them. During the association process, the column position node data is first sorted by timestamp, and then a snapshot is created for each time point, containing the state of all columns at that time. By comparing data from adjacent time points, the rate of change of column position and orientation is calculated to identify movement trends and abnormal changes. The temporal node feature sequence is stored using a multi-dimensional array structure, facilitating subsequent temporal and spatial analysis.

[0062] Temporal filtering of the time-series node feature sequences to eliminate abnormal jitter and synthesize a time-series feature map is a crucial step in improving data quality. The temporal filtering process employs a low-pass filter to remove high-frequency noise and transient interference. First, the position and attitude data of each pillar node are subjected to Fourier transform to obtain a frequency domain representation. Then, a frequency threshold is applied to filter out high-frequency components. Finally, the time domain signal is recovered through inverse Fourier transform. The filtered data retains the main trend of pillar position changes and eliminates abnormal jitter caused by measurement errors, environmental vibrations, and other factors. The time-series feature map is a graphical representation of the filtered time-series node feature sequences, visually displaying the changes in pillar position and attitude over time, as well as the dynamic changes in the connection relationships between pillars.

[0063] In one specific embodiment, the process of performing step S103 may specifically include the following steps:

[0064] (1) Import the design drawings into the computer and extract the column design axis data, and convert them into design feature maps in the image space coordinate system through coordinate transformation;

[0065] (2) Extract feature descriptors from the time-series feature map and the design feature map, and construct a feature index table using the locality-sensitive hashing algorithm;

[0066] (3) Calculate the feature similarity matrix using the feature index table, and solve the maximum weighted binary matching problem using the improved Hungarian algorithm;

[0067] (4) Obtain the correspondence between the actual columns and the designed columns from the maximum weighted bisection matching results obtained from solving the maximum weighted bisection matching problem, and generate a column pairing table;

[0068] (5) Calculate the positional and angular differences of each pair of columns in the column pairing table and generate the original deviation data;

[0069] (6) Perform statistical filtering and spatial interpolation on the original deviation data to form visual deviation data.

[0070] Specifically, importing design drawings into the computer and extracting the column design axis data is the foundational work for steel structure installation and adjustment. Design drawings are typically in CAD format, containing the design positions and geometric parameters of each steel structure component. The import process first converts the CAD file into a processable data format, then extracts the column design axis data using image analysis techniques. This data includes the spatial position and orientation of each column in the world coordinate system, represented by three-dimensional coordinates and direction vectors. After extraction, this data needs to be transformed from the world coordinate system to the image space coordinate system—a process called coordinate transformation. Coordinate transformation involves translation, rotation, and scaling, achieved through transformation matrices. The transformed data forms a design feature map, a graphical representation showing the projection of the column design positions into image space, facilitating comparison and matching with the actual captured image.

[0071] Extracting feature descriptors from temporal and design feature maps is a crucial step in image matching. Feature descriptors are numerical vectors that describe the characteristics of local regions in an image, uniquely identifying feature points within the image. In steel structure column matching, feature descriptors mainly include the column's geometric features (such as length, width, and orientation) and topological features (such as its connection to adjacent columns). The extraction process first identifies key points in the temporal and design feature maps, then calculates feature descriptors around each key point. Locality-Sensitive Hash (LSH) is a technique for similarity search that quickly finds similar feature descriptors. This algorithm divides the high-dimensional feature space into multiple "buckets," ensuring that similar feature descriptors fall into the same bucket. Specifically, multiple hash functions are selected, each mapping a feature descriptor to an integer value. These integer values ​​are combined to form a hash code, determining the bucket to which the feature descriptor belongs. This method of constructing a feature index table significantly accelerates the feature matching process.

[0072] Calculating the feature similarity matrix using the feature index table is a step in quantifying the matching degree of the columns. The feature similarity matrix is ​​a two-dimensional array where each element represents the similarity between the actual column and the designed column. During calculation, all pairs of actual and designed columns are traversed, the feature index table is queried, feature descriptors sharing the same bucket are identified, and their similarity is calculated. Similarity is typically measured using Euclidean distance or cosine similarity; a higher value indicates a higher degree of matching. The maximum weighted bipartite matching problem is a combinatorial optimization problem. The goal is to find a set of edges in a bipartite graph such that the sum of the edge weights is maximized, and each node is associated with at most one edge. In steel structure column matching, the two sides of the bipartite graph represent the actual column and the designed column, respectively, and the edge weights are the similarity of the corresponding column pairs. The improved Hungarian algorithm is an efficient method for solving this problem; it iteratively searches for augmenting paths until the optimal match is found.

[0073] The correspondence between actual and designed columns is obtained from the maximum weighted binary matching results, and a column pairing table is generated as the output of the matching process. The matching result is a one-to-one mapping, representing the design column corresponding to each actual column. The column pairing table is a structured representation of this mapping, typically containing the actual column ID, the corresponding design column ID, and the matching confidence score. The matching confidence score is an indicator of matching reliability, derived from the values ​​in the similarity matrix. The mathematical representation of the pairing table can be described as follows:

[0074]

[0075] in, This is a column pairing table. The identifier represents the k-th actual column. Indicates and Matching design column identifiers, It is a permutation function that represents the optimal matching result. It is the matching confidence level. This represents the total number of actual columns. (Permutation function) This was obtained by solving the following maximization problem:

[0076]

[0077] in, It is an actual pillar With design columns The similarity between them.

[0078] Calculating the positional and angular differences of each pair of columns in the column pairing table to generate raw deviation data is a preparatory step for installation adjustments. The raw deviation data includes the differences in spatial position and orientation for each pair of matched columns. Positional differences are obtained by calculating the Euclidean distance between the actual column center point and the corresponding designed column center point, typically in millimeters or centimeters. Angular differences are obtained by calculating the angle between the actual column axis and the designed column axis, in degrees or radians. This difference data reflects the deviations during steel structure installation, providing a quantitative basis for subsequent adjustments.

[0079] The final step in data processing is to perform statistical filtering and spatial interpolation on the raw deviation data to form visual deviation data. Statistical filtering aims to eliminate outliers and noise, improving data reliability. Commonly used statistical filtering methods include median filtering, average filtering, and Gaussian filtering. The filtered data may contain spatial discontinuities or undersampling issues, which are addressed using spatial interpolation techniques. Spatial interpolation is a method of extrapolating data for unknown points based on data from known points; commonly used interpolation algorithms include inverse distance weighted interpolation, kriging, and radial basis function interpolation. The resulting visual deviation data is a visual representation, typically displayed as a heatmap or vector field, intuitively showing the distribution and trends of deviations in various parts of the steel structure.

[0080] In a large-scale steel structure project, the design drawings included design data for 35 columns. Through import processing, the coordinates and orientation information of each column were extracted and transformed into an image coordinate system to generate a design feature map. The time-series feature map collected from the actual construction site contained 32 identifiable column structures. Feature descriptors were extracted from these two sets of feature maps; each column's descriptor included geometric parameters and connectivity relationships. These high-dimensional descriptors were mapped to a low-dimensional space using a locality-sensitive hashing algorithm to construct a feature index table. The calculated similarity matrix showed that most columns had clear correspondences, but some areas exhibited significant confusion. An improved Hungarian algorithm was applied to solve the maximum weighted binary matching problem to obtain the optimal column correspondences. 31 column pairs were extracted from the matching results, of which one actual column did not find a match, and four design columns were not detected in the actual construction. The positional and angular deviations of these 31 column pairs were calculated, revealing that the maximum positional deviation reached 38 mm, occurring in the structural edge region; the maximum angular deviation was 1.8 degrees, occurring in areas significantly affected by external forces. Outliers were removed from these raw deviation data using mean value filtering, and then spatial interpolation was performed using radial basis functions to generate complete visual deviation data. This deviation data clearly revealed the main patterns of structural deformation, providing data support for subsequent precise adjustments.

[0081] In one specific embodiment, the process of executing step S104 may specifically include the following steps:

[0082] (1) Establish a time series data matrix based on the correspondence and visual deviation data, and extract the main change patterns by principal component analysis;

[0083] (2) Calculate the trajectory of the column position change along the forward time direction, and eliminate random noise by using the Kalman filter algorithm to form a positive displacement vector;

[0084] (3) Calculate the trajectory of the column position change in reverse along time, improve the accuracy of reverse prediction by Bayesian estimation method, and form a reverse displacement vector;

[0085] (4) Weight the forward displacement vector and the reverse displacement vector to generate a columnar displacement vector field;

[0086] (5) Spatial interpolation is performed on the displacement vector field of the column, and a continuous displacement field representation is constructed through radial basis functions;

[0087] (6) Calculate the spatial distribution of position deviation based on the displacement field and the original position data, and generate a three-dimensional deviation distribution map.

[0088] Specifically, establishing a time-series data matrix based on correspondences and visual deviation data is a crucial step in the dynamic tracking of steel structure column axes. A time-series data matrix is ​​a multi-dimensional data structure that records the positional deviation information of all columns at each time point. The construction process first arranges the visual deviation data in chronological order, then organizes the data for each time point into a matrix form. The rows of the matrix represent the time series, and the columns represent the position and attitude parameters of different columns. Principal component analysis (PCA) is a dimensionality reduction technique used to extract the main change patterns from high-dimensional data. In steel structure monitoring, PCA identifies the main directions of data change by calculating the eigenvalues ​​and eigenvectors of the data covariance matrix. These main directions typically correspond to the overall deformation patterns of the steel structure, such as overall offset, torsion, or local deformation. The extracted main change patterns help understand the underlying causes of column position changes, providing guidance for subsequent precise adjustments.

[0089] Calculating the trajectory of column position changes along the forward time path is a fundamental method for tracking the dynamic behavior of columns. The forward calculation begins at the starting point of the time series and proceeds sequentially, recording the changes in column position. This process can be described by the following mathematical model:

[0090]

[0091] in, It is the position vector of the column at time t. It is a velocity vector. It is the time step. It's random noise. The Kalman filter algorithm is a recursive estimation algorithm used to estimate the state of a dynamic system from noisy measurement data. The Kalman filter eliminates random noise through a predict-update cyclic process, resulting in a smooth, positive displacement vector.

[0092] Calculating the trajectory of the column's position change in reverse time is a supplementary method. It starts from the end of the time series and proceeds in reverse chronological order. The reverse calculation process is similar to the forward calculation, but in the opposite direction. The mathematical model is as follows:

[0093]

[0094] in, It is the position vector of the column at the reverse time t. It is the reverse velocity vector. It is random noise. Bayesian estimation methods improve the accuracy of state estimation by combining prior knowledge and observational data. In reverse trajectory calculation, Bayesian estimation uses structural characteristics and historical data as prior information, combined with real-time measurements, to obtain a more accurate reverse displacement vector.

[0095] Weighted fusion of forward and reverse displacement vectors is a key step in improving trajectory estimation accuracy. The fusion process employs a weighted averaging method, assigning weights based on the reliability of the forward and reverse estimates to calculate the final column displacement vector field. This displacement vector field provides a complete representation of the column's motion in space, containing information on the position, velocity, and acceleration of each column. Compared to trajectories in a single direction, the fused vector field more accurately reflects the true motion of the columns, reducing estimation bias and uncertainty.

[0096] Spatial interpolation of the column displacement vector field is a necessary step in constructing a continuous displacement field representation. Spatial interpolation is a method of estimating data at unsampled locations based on data from discrete sampled points. Radial basis functions (RBFs) are a commonly used interpolation function that uses the distance between sampled points as the independent variable to construct a smooth and continuous interpolation surface. In steel structure monitoring, RBF interpolation can estimate the displacement values ​​at any location within the entire structural space based on finite column displacement data, forming a continuous displacement field representation. This continuous representation facilitates subsequent analysis and visualization, providing comprehensive reference information for structural adjustments.

[0097] The final step in generating a 3D deviation distribution map is calculating the spatial distribution of positional deviations based on the displacement field and original position data. The calculation process first applies a continuous displacement field to the original position data to obtain the adjusted expected position; then, the expected position is compared with the design position, calculating the deviation value at each point; finally, these deviation values ​​are mapped into 3D space to form a 3D deviation distribution map. The 3D deviation distribution map is typically displayed in the form of a heat map or vector field, intuitively reflecting the magnitude and direction of deviations in various parts of the structure, and serves as an important reference for steel structure installation and adjustment.

[0098] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0099] (1) Perform gradient analysis on the three-dimensional deviation distribution map, and obtain the deviation change rate and main deformation direction by calculating the directional derivative;

[0100] (2) Construct a multi-objective optimization function based on the deviation change rate, which includes three sub-objectives: minimizing the column position deviation, minimizing the adjustment amount, and maximizing the structural stability;

[0101] (3) Solve the multi-objective optimization function by using the particle swarm optimization algorithm to obtain the Pareto optimal solution set that balances the objectives;

[0102] (4) Extract the column adjustment vector from the Pareto optimal solution set and perform feasibility screening based on the engineering constraints;

[0103] (5) Calculate the rigid body transformation matrix for the selected column adjustment vectors to generate a spatial transformation matrix containing translation and rotation components;

[0104] (6) Convert the spatial transformation matrix into an executable adjustment instruction to form adjustment instruction data.

[0105] Specifically, gradient analysis of the three-dimensional deviation distribution map is the first step in calculating the column adjustment parameters. The three-dimensional deviation distribution map is a spatial representation of the deviation between the actual and designed positions of the column, containing coordinate and deviation value information. Gradient analysis obtains the spatial distribution characteristics of the deviation by calculating the rate of change of the deviation field in various directions. The directional derivative is the rate of change of a function in a specified direction. By calculating the derivatives in different directions, the direction and rate of change of the deviation with the fastest change can be determined. In the specific calculation process, the three-dimensional deviation distribution map is first discretized into grid points, and then the partial derivatives of the deviation value in the x, y, and z directions are calculated for each grid point, synthesized into a gradient vector. The magnitude of the gradient vector represents the rate of change of the deviation, and the direction points to the direction of the fastest increase in deviation. By analyzing the gradient distribution throughout the space, the main deformation areas and deformation directions can be identified, providing a quantitative basis for column adjustment.

[0106] Constructing a multi-objective optimization function based on the rate of change of deviation is key to solving for the optimal adjustment scheme. The multi-objective optimization function integrates three competing sub-objectives: minimizing the column position deviation, minimizing the adjustment amount, and maximizing structural stability. The objective of minimizing the column position deviation aims to reduce the difference between the actual and designed column positions, typically represented by Euclidean distance. The objective of minimizing the adjustment amount aims to reduce the workload of the adjustment operation, usually measured by the distance the column moves and the change in angle. The objective of maximizing structural stability aims to maintain the overall stiffness and strength of the structure, typically evaluated by calculating the stress distribution and deformation energy of the adjusted structure. These three sub-objectives often conflict; for example, reducing the position deviation may require a large adjustment amount, while an excessively large adjustment amount may affect structural stability. Therefore, a weighted comprehensive objective function is needed to balance the importance of each sub-objective.

[0107] Particle Swarm Optimization (PSO) is an effective method for finding equilibrium solutions in multi-objective optimization functions. PSO is a heuristic optimization algorithm that simulates the foraging behavior of birds to find optimal solutions. In the steel structure column adjustment problem, each particle represents a possible adjustment scheme, encompassing the adjustment parameters of all columns. The algorithm first randomly initializes a swarm of particles, then iteratively updates the position and velocity of each particle. The update rule considers both the particle's own historical best position and the swarm's global best position, guiding particles to move towards better solution spaces. In multi-objective optimization, the algorithm maintains the diversity and convergence of the solution set through non-dominated sorting and crowding calculation, ultimately forming a Pareto optimal solution set. A Pareto optimal solution set is a set of solutions where each solution cannot further improve the values ​​of other objective functions without reducing the value of any of the other objective functions.

[0108] Extracting column adjustment vectors from the Pareto optimal solution set is a crucial step in transforming theoretical solutions into practical schemes. Each solution in the Pareto optimal solution set represents a possible adjustment scheme, but not all schemes are suitable for actual construction. The extraction process first selects several representative solutions from the solution set based on project priorities and decision preferences; then, these solutions are converted into column adjustment vectors, each containing the displacement and rotation parameters of the column. Engineering constraints include limitations on the adjustment equipment's capacity, construction space constraints, and structural safety margin requirements; these conditions are used to screen feasible schemes. Feasibility screening checks whether each adjustment vector satisfies all constraints, eliminating infeasible schemes and retaining adjustment vectors that conform to the actual engineering situation.

[0109] Calculating the rigid body transformation matrix from the selected column adjustment vector is fundamental to generating precise adjustment instructions. The column adjustment vector describes the distance the column needs to move and the angle of rotation, but construction sites require more specific transformation descriptions. The rigid body transformation matrix is ​​a mathematical tool used to describe the translation and rotation of an object in three-dimensional space. The calculation process first decomposes the column adjustment vector into translation and rotation components; then, it constructs translation and rotation matrices; finally, it multiplies the two matrices to obtain the complete spatial transformation matrix. The spatial transformation matrix is ​​a 4×4 matrix, with the top left 3×3 submatrix representing rotation and the top right 3×1 submatrix representing translation. This matrix allows for a precise description of the column's transformation from its current position to its target position.

[0110] Converting the spatial transformation matrix into executable adjustment instructions is the final step in achieving precise adjustments. The spatial transformation matrix is ​​a mathematical representation that needs to be translated into concrete operational instructions that engineers can understand and execute. The conversion process first extracts the translation vector and rotation angle from the transformation matrix; then, based on the characteristics of the adjustment equipment used on site, these parameters are converted into equipment operating parameters, such as the lifting height of hydraulic jacks, the lateral advance distance, and the thickness of the pad plate; finally, adjustment instruction data is generated, including the adjustment direction, adjustment amount, and execution sequence for each column. Adjustment instruction data is typically presented in tabular or graphical form for easy understanding and execution by construction personnel.

[0111] In one specific embodiment, the process of performing step S106 may specifically include the following steps:

[0112] (1) Construct a virtual outline of the target position of the column based on the adjustment instruction data, and overlay the virtual outline onto the real-time image using augmented reality technology;

[0113] (2) Implement column adjustment operations at the construction site and record the changes in column position during the adjustment process using a real-time image acquisition algorithm;

[0114] (3) Perform image registration between the actual position of the column in the real-time image and the virtual contour, and calculate the position overlap index;

[0115] (4) Generate an adjustment accuracy assessment report based on the position overlap index, including residual deviation and spatial distribution characteristics;

[0116] (5) Based on the residual deviation and historical adjustment data, the adjustment coefficients are updated through reinforcement learning algorithm to generate model optimization parameters;

[0117] (6) Feed the model optimization parameters back to the column adjustment database to form a closed-loop adjustment and optimization mechanism to complete the entire steel structure installation and adjustment process.

[0118] Specifically, a three-dimensional virtual model of the column is established, including its geometry, dimensions, and spatial position. The virtual outline is constructed using computer-aided design technology, combining the column's design parameters with target location data to generate the column's spatial outline at the target location. Augmented reality (AR) technology overlays virtual information onto a real-world view, implemented in steel structure installation using head-mounted displays or mobile devices such as tablets. The AR system first identifies feature points in the environment, establishing a coordinate correspondence between the real and virtual worlds, and then overlays the virtual column outline onto the real-time image according to this spatial correspondence. This intuitive visual guidance allows construction workers to clearly understand the target position of the column, greatly improving the accuracy and efficiency of adjustments. When adjusting columns on the construction site, real-time image acquisition algorithms are needed to record changes in column position during the adjustment process. These algorithms, based on computer vision technology, continuously capture image sequences of the column using multiple fixed or mobile cameras. The image acquisition frequency is typically set to multiple frames per second to ensure the capture of even minute changes in the column's position. The acquired images are first preprocessed, including noise reduction, brightness equalization, and geometric correction. Then, feature extraction algorithms are used to identify marker points or feature regions on the pillars. Feature extraction employs methods such as edge detection, corner detection, or feature descriptors to generate feature representations of the pillars in the images. By matching features between consecutive frames, the movement trajectory of the pillars during the adjustment process is tracked, and their positional changes are recorded. This positional change data is stored in time-series format for subsequent accuracy analysis and verification.

[0119] Image registration, specifically image registration between the actual position of the pillars in real-time images and the virtual contour, is a crucial step in evaluating the adjustment progress and accuracy. Image registration is the process of aligning two or more images; in this method, it refers to establishing a spatial correspondence between the real-time acquired pillar image and the virtually generated target contour. The registration process first extracts the pillar contour or feature points from the real-time image and then matches them with the corresponding features of the virtual contour. Matching algorithms typically employ feature point matching or contour matching methods to calculate the spatial transformation relationship between the two. The positional overlap index quantifies the degree of fit between the actual pillar and the target position, including positional deviation, angular deviation, and contour overlap rate. Positional deviation is obtained by calculating the Euclidean distance between corresponding feature points, angular deviation is obtained by calculating the angle along the principal axis, and contour overlap rate is determined by calculating the ratio of the overlapping area to the total area. These indices comprehensively reflect the accuracy of the pillar adjustment, providing a quantitative basis for subsequent fine-tuning.

[0120] Generating an adjustment accuracy assessment report based on the positional overlap index is a crucial step in evaluating the adjustment effect. The adjustment accuracy assessment report is a systematic summary of the column adjustment results, containing two main parts: residual deviation and spatial distribution characteristics. Residual deviation refers to the remaining positional and orientation deviations after adjustment, calculated using various parameters in the positional overlap index. Spatial distribution characteristics describe the distribution pattern of residual deviations throughout the structure, including the magnitude range, main distribution areas, and trends of the deviations. The report generation process first performs statistical analysis on the positional overlap index data, calculating the average, maximum, minimum, and standard deviation of each deviation; then, spatial clustering analysis identifies areas and patterns of concentrated deviations; finally, intuitive charts and textual descriptions are generated, forming a complete assessment report. The assessment report not only records the final adjustment results but also provides data support for subsequent optimization and improvement. Updating adjustment coefficients and generating model optimization parameters based on residual deviation and historical adjustment data using reinforcement learning algorithms is an innovative method to improve adjustment efficiency and accuracy. Reinforcement learning is a machine learning method that learns optimal decision-making strategies through trial and error and reward mechanisms. In steel structure column adjustment, reinforcement learning algorithms treat the adjustment operation as a decision-making process, using the reduction in residual deviation as a reward signal. By continuously trying different adjustment strategies, the optimal adjustment method is learned. Specifically, the process first constructs a state space (current column position and orientation) and an action space (possible adjustment operations); then, the learning model is initialized based on historical adjustment data; next, Q-learning or deep reinforcement learning algorithms are used to iteratively update the action value function; finally, the optimal adjustment strategy for different states is obtained. Model optimization parameters include the adjustment coefficient matrix, operation order weights, and convergence threshold, which directly affect the efficiency and accuracy of the adjustment.

[0121] Feeding model optimization parameters back to the column adjustment database to form a closed-loop adjustment and optimization mechanism is a crucial step in achieving continuous improvement. The column adjustment database is a centralized repository storing data related to steel structure installation and adjustment, including historical adjustment records, model parameters, and experiential knowledge. The feedback process first formats the newly generated model optimization parameters into a format acceptable to the database; then it compares and verifies them with historical parameters in the database; finally, it updates the parameter records in the database and marks them with timestamps and applicable conditions. The closed-loop adjustment and optimization mechanism is a self-improving working mode. By continuously accumulating adjustment experience and optimizing adjustment strategies, the system can draw on past experience when handling new adjustment tasks, improving adjustment efficiency and accuracy. This mechanism is particularly suitable for continuous steel structure installation projects, where subsequent adjustment operations can benefit from previous experience, gradually improving the overall installation quality.

[0122] The steel structure installation and adjustment method based on dynamic tracking of column axis in the embodiments of this application has been described above. The steel structure installation and adjustment system based on dynamic tracking of column axis in the embodiments of this application is described below. Please refer to [link / reference]. Figure 2 One embodiment of the steel structure installation adjustment system based on dynamic tracking of column axis in this application includes:

[0123] The acquisition module is used to acquire data from images of steel structure columns, and obtain an image dataset containing column outlines, target point features, time series information and environmental parameters.

[0124] The extraction module is used to perform image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing the column location nodes and connection relationships;

[0125] The matching module is used to establish a computer vision matching algorithm based on the time-series feature map to obtain the correspondence between the actual column image and the design model and the visual deviation data.

[0126] The analysis module is used to perform bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map;

[0127] The calculation module is used to calculate the column adjustment parameters based on the three-dimensional deviation distribution map using an image processing algorithm, and obtain adjustment instruction data including a spatial transformation matrix;

[0128] The verification module is used to construct a virtual guidance model based on the adjustment instruction data and perform real-time image feedback verification to obtain the adjustment accuracy evaluation results and model optimization parameters.

[0129] Through the collaborative efforts of the aforementioned components, image datasets containing column outlines, target point features, temporal information, and environmental parameters are obtained by acquiring images of steel structure columns, thus solving the problems of insufficient accuracy and incomplete data in traditional measurement methods. Image segmentation and feature extraction are performed based on the image dataset to obtain temporal feature maps representing column position nodes and connection relationships, achieving an accurate description of the spatial position and interrelationships of the columns. A computer vision matching algorithm is established based on the temporal feature maps to obtain the correspondence between the actual column images and the design model, as well as visual deviation data, making the matching process between the columns and the design positions more accurate and reliable. Two-way spatiotemporal trajectory analysis is performed on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map, comprehensively understanding the dynamic characteristics of column position changes. Based on the three-dimensional deviation distribution map, image processing algorithms are applied to calculate column adjustment parameters, obtaining adjustment command data containing spatial transformation matrices, achieving accurate conversion from measurement data to specific adjustment schemes. A virtual guidance model is constructed based on the adjustment command data and real-time image feedback verification is performed to obtain adjustment accuracy evaluation results and model optimization parameters, forming a complete closed-loop verification and continuous optimization mechanism. Of particular note is the innovative application of various artificial intelligence algorithms in the field of steel structure installation. These include improved accuracy of computer vision matching algorithms in column identification and matching, effective capture of dynamic changes through bidirectional spatiotemporal trajectory analysis algorithms, and intelligent adjustment strategy generation based on multi-objective optimization and reinforcement learning. These algorithms contribute to the solution in two ways: firstly, they solve the problem of complex column interaction effects that traditional methods cannot handle, improving the coordination of overall structural adjustments; secondly, through continuous optimization of adjustment parameters via machine learning, they accumulate and apply experiential knowledge, significantly improving adjustment efficiency and accuracy. In summary, this solution, by constructing a complete technology chain from data acquisition, feature extraction, matching analysis to adjustment verification, and combining the advantages of artificial intelligence algorithms, achieves precision, intelligence, and automation in the steel structure column installation process, effectively solving the technical problems of insufficient accuracy, low efficiency, and unstable quality in traditional steel structure installation.

[0130] Reference Figure 3 This invention also provides a computer device, which can be a server, and its internal structure can be as follows: Figure 3As shown, the computer device includes a processor, memory, display screen, input device, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database stores the data corresponding to this embodiment. The network interface is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, it implements the above-described method.

[0131] Those skilled in the art will understand that Figure 3 The structures shown are merely block diagrams of some structures related to the present invention and do not constitute a limitation on the computer devices on which the present invention is applied.

[0132] An embodiment of the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method. It is understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0133] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the present invention and embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM, etc.

[0134] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0135] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0136] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for adjusting the installation of steel structures based on dynamic tracking of column axes, characterized in that, The steel structure installation and adjustment method based on dynamic tracking of column axis includes: Data was collected from images of steel structure columns to obtain an image dataset containing column outlines, target point features, temporal information, and environmental parameters. Based on the image dataset, image segmentation and feature extraction are performed to obtain a temporal feature map representing the column location nodes and connection relationships; A computer vision matching algorithm is established based on the time-series feature map to obtain the correspondence between the actual column image and the design model and the visual deviation data. A two-way spatiotemporal trajectory analysis was performed on the correspondence and visual deviation data to obtain the column displacement vector field and a three-dimensional deviation distribution map. Based on the three-dimensional deviation distribution map, an image processing algorithm is applied to calculate the column adjustment parameters, resulting in adjustment instruction data containing a spatial transformation matrix; Based on the adjustment instruction data, a virtual guidance model is constructed and verified using real-time image feedback to obtain adjustment accuracy evaluation results and model optimization parameters. This includes: constructing a virtual outline of the column target position according to the adjustment instruction data; overlaying the virtual outline onto a real-time image using augmented reality technology; performing column adjustment operations at the construction site and recording column position changes during the adjustment process using a real-time image acquisition algorithm; performing image registration between the actual column position in the real-time image and the virtual outline, and calculating the position overlap index; generating an adjustment accuracy evaluation report based on the position overlap index, including residual deviation and spatial distribution characteristics; updating the adjustment coefficients using a reinforcement learning algorithm based on the residual deviation and historical adjustment data to generate model optimization parameters; and feeding the model optimization parameters back to the column adjustment database to form a closed-loop adjustment optimization mechanism, completing the entire steel structure installation and adjustment process.

2. The steel structure installation and adjustment method based on dynamic tracking of column axis as described in claim 1, characterized in that, The data acquisition of steel structure column images yields an image dataset containing column outlines, target point features, temporal information, and environmental parameters, including: Multi-view high-definition camera equipment is set up around the steel structure column, and the reflective target points on the surface of the column are captured by the image tracking algorithm; The acquired image sequence is geometrically corrected, and lens distortion is eliminated through a binocular vision calibration algorithm to obtain the corrected image sequence; Timestamp information is extracted from the corrected image sequence, and images acquired by different camera devices are aligned according to time points using a time synchronization algorithm; Background separation is performed on time-aligned images, and the column outline and target point features are extracted using an adaptive threshold segmentation algorithm. Ambient light, temperature, and vibration parameter data are collected and correlated with image data using a data fusion algorithm to form an environmental impact factor matrix; The image dataset is obtained by integrating the column outline, target point features, temporal information, and environmental influence factor matrix.

3. The steel structure installation and adjustment method based on dynamic tracking of column axis as described in claim 1, characterized in that, The step of performing image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing the column location nodes and connection relationships includes: Adaptive watershed segmentation is performed on the column outlines in the image dataset to obtain a column region mask image; The center line of the column is extracted from the mask image of the column area, and the equation of the column axis is calculated by the Hough transform algorithm. Based on the target point features and the column axis equation, column position node data is generated, wherein each node contains three-dimensional spatial coordinates and attitude angle; A steel structure topology diagram is constructed using the column location node data, and the connection relationship between adjacent columns is identified using a spatial distance threshold algorithm. The column location node data and connection relationships are associated with the time-series information to generate a time-series node feature sequence. The time-series node feature sequence is subjected to time-domain filtering to eliminate abnormal jitter and synthesize the time-series feature map.

4. The steel structure installation and adjustment method based on dynamic tracking of column axis as described in claim 1, characterized in that, The computer vision matching algorithm established based on the temporal feature map, which obtains the correspondence between the actual column image and the design model and the visual deviation data, includes: Import the design drawings into the computer and extract the column design axis data, then convert them into a design feature map in the image space coordinate system through coordinate transformation. Feature descriptors are extracted from the time-series feature map and the design feature map, and a feature index table is constructed using the locality-sensitive hashing algorithm; The feature similarity matrix is ​​calculated using the feature index table, and the maximum weighted binary matching problem is solved using the improved Hungarian algorithm. The correspondence between actual columns and designed columns is obtained from the maximum weighted bisection matching results obtained by solving the maximum weighted bisection matching problem, and a column pairing table is generated. Calculate the positional and angular differences of each pair of columns in the column pairing table to generate raw deviation data; The original deviation data is statistically filtered and spatially interpolated to form the visual deviation data.

5. The steel structure installation and adjustment method based on dynamic tracking of column axis as described in claim 1, characterized in that, The step of calculating the column adjustment parameters based on the three-dimensional deviation distribution map using an image processing algorithm to obtain adjustment instruction data containing a spatial transformation matrix includes: A time series data matrix is ​​established based on the aforementioned correspondence and visual deviation data, and the main change patterns are extracted using principal component analysis. The trajectory of the column's position change is calculated along the forward time direction, and random noise is eliminated by the Kalman filter algorithm to form a positive displacement vector; The trajectory of the column's position change is calculated in reverse along time, and the accuracy of the reverse prediction is improved by Bayesian estimation method to form a reverse displacement vector. The positive displacement vector and the negative displacement vector are weighted and fused to generate a columnar displacement vector field; Spatial interpolation is performed on the displacement vector field of the column, and a continuous displacement field representation is constructed through radial basis functions; The spatial distribution of position deviation is calculated based on the displacement field and the original position data, and the three-dimensional deviation distribution map is generated.

6. The steel structure installation and adjustment method based on dynamic tracking of column axis as described in claim 1, characterized in that, The step of performing bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map includes: Gradient analysis is performed on the three-dimensional deviation distribution map, and the deviation change rate and main deformation direction are obtained by calculating the directional derivative. A multi-objective optimization function is constructed based on the deviation change rate, which includes three sub-objectives: minimizing the column position deviation, minimizing the adjustment amount, and maximizing structural stability. The multi-objective optimization function is solved by particle swarm optimization algorithm to obtain a Pareto optimal solution set that balances the objectives. Extract the column adjustment vector from the Pareto optimal solution set, and perform feasibility screening based on engineering constraints; Calculate the rigid body transformation matrix for the selected column adjustment vectors to generate a spatial transformation matrix containing translation and rotation components. The spatial transformation matrix is ​​converted into an engineering-executable adjustment instruction, forming the adjustment instruction data.

7. A steel structure installation adjustment system based on dynamic tracking of column axis, used to implement the steel structure installation adjustment method based on dynamic tracking of column axis as described in any one of claims 1-6, characterized in that, The steel structure installation and adjustment system based on dynamic tracking of column axis includes: The acquisition module is used to acquire data from images of steel structure columns, and obtain an image dataset containing column outlines, target point features, time series information and environmental parameters. The extraction module is used to perform image segmentation and feature extraction based on the image dataset to obtain a temporal feature map representing the column location nodes and connection relationships; The matching module is used to establish a computer vision matching algorithm based on the time-series feature map to obtain the correspondence between the actual column image and the design model and the visual deviation data. The analysis module is used to perform bidirectional spatiotemporal trajectory analysis on the correspondence and visual deviation data to obtain the column displacement vector field and three-dimensional deviation distribution map; The calculation module is used to calculate the column adjustment parameters based on the three-dimensional deviation distribution map using an image processing algorithm, and obtain adjustment instruction data including a spatial transformation matrix; The verification module is used to construct a virtual guidance model based on the adjustment instruction data and perform real-time image feedback verification to obtain adjustment accuracy evaluation results and model optimization parameters. This includes: constructing a virtual outline of the column target position based on the adjustment instruction data; overlaying the virtual outline onto a real-time image using augmented reality technology; performing column adjustment operations at the construction site and recording column position changes during the adjustment process using a real-time image acquisition algorithm; performing image registration between the actual column position in the real-time image and the virtual outline, and calculating the position overlap index; generating an adjustment accuracy evaluation report based on the position overlap index, including residual deviation and spatial distribution characteristics; updating the adjustment coefficients based on the residual deviation and historical adjustment data using a reinforcement learning algorithm to generate model optimization parameters; and feeding the model optimization parameters back to the column adjustment database to form a closed-loop adjustment optimization mechanism, completing the entire steel structure installation and adjustment process.

8. A computer device, characterized in that, The system includes a memory and a processor, the memory storing a computer program that can run on the processor, characterized in that, when the processor executes the computer program, it implements the steel structure installation and adjustment method based on dynamic tracking of column axis as described in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, the computer program causing a processor, when executed by a processor, to perform the steel structure installation and adjustment method based on dynamic tracking of column axis as described in any one of claims 1 to 6.

Images