Steel structure construction welding quality evaluation method and system based on visual analysis

Abstract

The invention discloses a steel structure construction welding quality evaluation method and system based on visual analysis, and relates to the technical field of welding engineering management. Comprising the steps that a historical steel structure model is constructed, a plurality of supplementary steel structure models are generated through constraint variation on the basis, and the historical steel structure model and the supplementary steel structure models are collectively called as a virtual steel structure model; performing stress analog simulation on each virtual steel structure model, and determining stress accumulation parameters of different position blocks under a preset load; performing image processing and feature recognition on the current steel structure welding seam by adopting a visual analysis technology, determining the welding quality of the welding seam at each position, and performing comprehensive welding quality evaluation in combination with the current stress distribution feature; according to the method, through organic combination of virtual model stress simulation and visual inspection, accurate evaluation of the welding seam quality considering structural stress distribution is achieved, the objectivity and reliability of evaluation are improved, and the method is suitable for large steel structure engineering construction quality control.

Description

Technical Field

[0001] This invention relates to the field of welding engineering management technology, and in particular to a method and system for assessing the welding quality of steel structure construction based on visual analysis. Background Technology

[0002] In the field of steel structure construction, welding quality is a key factor affecting the overall load-bearing capacity, durability, and safety of the structure. Current technologies primarily employ traditional methods for weld quality assessment, such as manual visual inspection, ultrasonic testing, radiographic testing, or magnetic particle testing. While these methods can detect surface or internal defects, they suffer from low efficiency, high cost, strong subjectivity, and incomplete coverage of complex structures. With advancements in machine vision technology, image processing-based weld quality inspection methods are gradually being applied. For example, by capturing weld images and performing grayscale processing, edge extraction, and defect feature recognition, automated detection of defects such as porosity, cracks, and lack of fusion on the weld surface can be achieved. However, these methods typically focus only on the local appearance features and grayscale distribution of the weld, failing to consider the stress state and stress distribution differences of the weld within the overall steel structure. In practical engineering, the stress loads borne by welds at different locations in a steel structure vary significantly. Even minor defects in welds in high-stress concentration areas (such as node intersections, bifurcation points, or end connections) can lead to fatigue failure or overall failure, while similar defects in low-stress areas have a relatively smaller impact. Existing assessment methods lack comprehensive consideration of structural stress distribution, resulting in a mismatch between welding quality assessment results and actual structural performance. This leads to problems such as large deviations in safety risk assessment and inaccurate classification of defect hazard, making it difficult to meet the needs of large steel structure projects for high-quality and high-reliability welding assessment. Summary of the Invention

[0003] The purpose of this invention is to provide a comprehensive evaluation method that combines the visual inspection results of welds with the structural stress distribution characteristics, so as to achieve a more accurate and objective evaluation of the welding quality of steel structure construction.

[0004] This invention discloses a method for assessing the welding quality of steel structure construction based on visual analysis, including: Step S100: For the historical engineering records that are required, obtain a number of historical steel structure data, construct a corresponding historical steel structure model based on the historical steel structure data, and construct a number of supplementary steel structure models based on the historical steel structure models. The historical steel structure model and the supplementary steel structure models are collectively referred to as the virtual steel structure model. Step S200: Perform stress simulation on each virtual steel structure model to determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load. Correlate the mapping between the virtual steel structure model and the cumulative stress parameters to construct a virtual steel structure stress analysis model. Step S300: Using a virtual steel structure stress analysis model, analyze the currently constructed steel structure to determine the current stress distribution characteristics; Step S400: Using visual analysis technology, determine the welding quality of welds at different locations of the steel structure currently being constructed, and combine this with the current stress distribution characteristics to conduct a comprehensive welding quality assessment of the steel structure currently being constructed.

[0005] In some embodiments disclosed in this invention, the method for constructing a corresponding historical steel structure model based on historical steel structure data includes: Step S101: Construct a spatial coordinate system, analyze historical steel structure data, identify several individual steel structure components, determine the main extension direction of each individual steel structure component, and replace it with structural extension lines. Step S102: Align the steel structure unit components and the structural extension lines, and connect the structural extension lines point-to-point based on the combined welding positions of the steel structure unit components to obtain the historical steel structure model. Based on the historical steel structure data, determine the load intervention characteristics of different nodes in the historical steel structure model, including the stress changes of different nodes.

[0006] In some embodiments disclosed in this invention, the method for constructing several supplementary steel structure models based on historical steel structure models includes: Step S103: Perform topological abstraction on the linear connection structure of the historical steel structure model, identify key nodes, including endpoints, intersections and bifurcation points, linear segment lengths, angle distributions and load intervention characteristics at the nodes, including force direction and intensity changes, to form a parameterized historical linear topology template. Step S104 introduces a constrained mutation generation mechanism. Based on the parameterized historical linear topology template, the global topology parameters are simultaneously subjected to constrained random perturbations, including the number of key nodes, bifurcation mode, span ratio between nodes, line segment length, line segment angle offset, load force direction of position nodes, and load intensity changes, to obtain a supplementary linear topology template. Step S105: Based on the supplementary linear topology template, the historical steel structure model corresponding to the historical topology template is structurally adjusted to obtain the supplementary steel structure model.

[0007] In some embodiments disclosed in this invention, the method for stress simulation of a virtual steel structure model includes: Step S201: Perform finite element mesh adaptive generation on the linear segments and key nodes in the virtual steel structure model. Introduce multi-scale nested meshes, including macroscopic overall meshes and local sub-model meshes, near nodes with high load intervention features and weld connection locations. Step S202: Based on the load intervention characteristics of the virtual steel structure model, apply a multi-physics coupled dynamic loading sequence, including static load, cyclic fatigue load and random vibration load, analyze the stress superposition effect of key nodes connected by the structural extension line, and obtain the stress accumulation parameters of the key nodes.

[0008] In some embodiments disclosed in this invention, the method for determining the welding quality of welds at different locations in a currently constructed steel structure using visual analysis technology includes: Step S401: Take a picture of the welded joint of the steel structure being constructed to obtain an image of the welded joint, and use visual analysis technology to perform grayscale processing on the image of the welded joint to obtain a grayscale image of the welded joint. Step S402: Perform edge analysis on the grayscale image of the weld joint to determine the weld edge in the grayscale image of the weld joint, and then calibrate the weld edge in the grayscale image of the weld joint to determine the grayscale area of ​​the weld within the weld edge. Step S403: Analyze the edge structure features of the weld edge, and find a suitable comparison weld gray area in the preset weld gray area library based on the edge structure features. Select the most suitable weld gray area from the comparison weld gray area, and evaluate the welding quality corresponding to the most suitable weld gray area as the weld welding quality of the current weld gray area.

[0009] In some embodiments disclosed in this invention, the method for analyzing the edge structural features of the weld edge includes: Step S4031: Determine the center parallel line between the weld edges, uniformly set a number of width detection points on the center parallel line, and analyze the weld spacing between the weld edges at the width detection points. Step S4032: The central parallel line representation diagram formed by the central parallel lines and the weld spacing set on the central parallel lines are identified as edge structure features.

[0010] In some embodiments disclosed in this invention, the method for finding a suitable comparison weld grayscale region in a preset weld grayscale region library based on edge structural features includes: Step S40321: Based on the matching degree between the center parallel line representation images, the corresponding weld grayscale area is selected as the comparison weld grayscale area. The method for analyzing the matching degree between the center parallel line representation images includes: Align the representation diagrams of the central parallel lines, compare each relative central parallel line, analyze the intersection angle and length difference between them, and determine the line overlap parameter between the central parallel lines based on the value range of the intersection angle and length difference. If there is a central parallel line that does not have a corresponding central parallel line, then the overlap parameter corresponding to the central parallel line is determined to be the preset minimum value. Analyze the spacing difference of weld seams on the corresponding center parallel lines, and calculate the average and variance of the spacing difference, denoted as the average spacing difference and the variance of the spacing difference. Determine the value range to which the average spacing difference and the variance of the spacing difference belong, and output the weld seam spacing matching parameters between the center parallel lines based on their respective value ranges. Based on the weld spacing matching parameter, the line overlap parameter is corrected to obtain the corrected line overlap parameter. Based on the average overlap parameter, overlap parameter variance and minimum overlap parameter of the corrected line overlap parameters of all center parallel lines, the matching degree between the center parallel line representation diagrams is determined. The expression for calculating the overlap parameter between lines is as follows: ; Where H is the corrected inter-line coincidence parameter, h is the inter-line coincidence parameter, W is the weld spacing coincidence parameter, K is the coincidence parameter influence adjustment coefficient, and b is the coincidence parameter influence adjustment constant. The expression for calculating the degree of matching is: ; Where P represents the degree of matching. For the average coincident parameter, For the variance of the overlapping parameters, For the lowest overlap parameter, This is a variance correction coefficient determination function that outputs the corresponding variance correction coefficient based on the preset interval to which the variance of the coincident parameter belongs. This is the function for determining the correction coefficient of the lowest coincident parameter. Based on the preset interval to which the lowest coincident parameter belongs, it outputs the corresponding correction coefficient of the lowest coincident parameter.

[0011] In some embodiments disclosed in this invention, the method for selecting the most suitable weld grayscale region from the comparison weld grayscale region includes: Step S4033: Compare the current weld grayscale area with the comparison weld grayscale area, including comparing the average grayscale values ​​of different blocks between the two, calculating the difference in the average grayscale value of each block, identifying the grayscale protruding blocks between the two, determining the block center distance of the grayscale protruding blocks, and the block area difference. Step S4034: Compare the difference in the average gray value of each location block, record it as the average gray value difference, and determine the location blocks whose average gray value difference is less than or equal to a preset value, record them as average gray value equal location blocks, and calculate the ratio of average gray value equal location blocks to the number of average gray value equal locations in all location blocks. Step S4035: If the distance between the center of the blocks and the difference in the area of ​​the blocks are both less than or equal to the preset values, then the gray-scale protruding blocks are identified as equal gray-scale protruding blocks. The ratio of equal gray-scale protruding blocks to the total number of gray-scale protruding blocks is calculated and recorded as the ratio of equal gray-scale protruding blocks. Step S4036: Based on the ratio of the number of equal grayscale values ​​and the ratio of the number of outstanding grayscale blocks, determine the selected weld grayscale region for comparison. The methods for determining grayscale highlighted areas include: The current weld grayscale area and the weld grayscale area for comparison are divided into a unified grid to obtain several detection blocks of the same size and position. Calculate the average gray value of each location block, as well as the overall average gray value and gray standard deviation of the weld gray area; For each detection block, if the absolute difference between its average gray value and the overall average gray value exceeds the preset gray value difference threshold, then the block at that location is identified as a gray-prominent block. Merge adjacent grayscale highlight blocks to form connected grayscale highlight blocks.

[0012] In some embodiments disclosed in this invention, a method for comprehensively evaluating the welding quality of a currently constructed steel structure, taking into account the current stress distribution characteristics, includes: Step S401: Analyze the welding characteristics of welds at different locations of the steel structure being constructed, including determining the weld routing length and width, and determining the spatial characteristics of the weld. The spatial characteristics include the spatial routing of the weld, referred to as the spatial welding routing structure. Step S402: The spatial welding wiring structure, wiring length, and wiring width are compared with the preset weld strength database to determine the welding strength of the weld. The preset weld strength database is a database of preset combinations of welding features and welding strengths. The combination of welding features and welding strengths is derived from preset strength tests. Step S403: Calculate the ratio of welding strength to cumulative stress parameter, and record it as the welding quality sensitive weight. Based on the welding quality sensitive weight, correct the quality difference between the corresponding weld quality and the standard weld quality to obtain a comprehensive welding quality assessment.

[0013] In some embodiments disclosed in this invention, a steel structure construction welding quality assessment system based on visual analysis includes: The first module is used to obtain a number of historical steel structure data for projects that require historical engineering records, and to build a corresponding historical steel structure model based on the historical steel structure data. Based on the historical steel structure model, a number of supplementary steel structure models are also built. The historical steel structure model and the supplementary steel structure models are collectively referred to as the virtual steel structure model. The second module is used to perform stress simulation on each virtual steel structure model, determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load, associate the mapping between the virtual steel structure model and the cumulative stress parameters, and construct a virtual steel structure stress analysis model. The third module is used to analyze the currently constructed steel structure using a virtual steel structure stress analysis model to determine the current stress distribution characteristics. The fourth module is used to determine the welding quality of welds at different locations in the currently constructed steel structure using visual analysis technology, and to conduct a comprehensive welding quality assessment of the currently constructed steel structure in combination with the current stress distribution characteristics.

[0014] This invention discloses a method and system for assessing the welding quality of steel structure construction based on visual analysis, belonging to the field of welding engineering management technology. It includes constructing a historical steel structure model and generating several supplementary steel structure models based on this model through constrained mutations; these two are collectively referred to as virtual steel structure models. Stress simulation is performed on each virtual steel structure model to determine the cumulative stress parameters of different locations under preset loads. Visual analysis technology is used to perform image processing and feature recognition on the current steel structure welds to determine the welding quality of welds at each location, and a comprehensive welding quality assessment is performed based on the current stress distribution characteristics. This invention achieves accurate weld quality assessment considering structural stress distribution through the organic combination of virtual model stress simulation and visual inspection, improving the objectivity and reliability of the assessment, and is suitable for quality control in large-scale steel structure engineering construction.

[0015] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating the steps of a visual analysis-based method for assessing the welding quality of steel structure construction, as disclosed in an embodiment of the present invention. Detailed Implementation

[0017] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings and specific embodiments. It should be understood that the preferred embodiments described herein are only for illustration and explanation of the present invention and should not be construed as limiting the scope of protection of the present invention. Those skilled in the art can make some non-essential improvements and adjustments based on the following content of the present invention. In the present invention, unless otherwise expressly specified and limited, the technical terms used in the present invention should have the ordinary meaning understood by those skilled in the art.

[0019] Example: Example

[0020] The purpose of this invention is to provide a comprehensive evaluation method that combines the visual inspection results of welds with the structural stress distribution characteristics, so as to achieve a more accurate and objective evaluation of the welding quality of steel structure construction.

[0021] This invention discloses a method for assessing the welding quality of steel structure construction based on visual analysis. (See reference...) Figure 1 ,include: Step S100: For the historical engineering records required, obtain a number of historical steel structure data, construct a corresponding historical steel structure model based on the historical steel structure data, and construct a number of supplementary steel structure models based on the historical steel structure models. The historical steel structure models and the supplementary steel structure models are collectively referred to as the virtual steel structure model.

[0022] The core principle of step S100 is to accumulate real steel structure data from historical engineering records and construct a parameterized historical steel structure model based on this data. Simultaneously, a constrained mutation mechanism is introduced to generate diverse supplementary steel structure models, ultimately forming a rich dataset of virtual steel structure models. The technical principle involves first decomposing the historical steel structure data into components and replacing linear extension lines in a spatial coordinate system to achieve topological abstraction of the model (identifying key nodes such as endpoints, intersections, bifurcation points, and linear segment lengths, angles, and load characteristics). Then, while maintaining engineering rationality, constrained random perturbations are applied to key parameters (such as the number of nodes, bifurcation patterns, span ratios, angle offsets, and load direction and strength) to avoid generating physically impossible structures. This data augmentation method is similar to constrained generative adversarial thinking, but specifically designed for steel structure topology, significantly expanding the virtual sample size. For example, based on historical data from actual bridges or high-rise buildings, hundreds of variant models can be generated, covering different spans and node load scenarios, thus providing diverse and high-fidelity virtual samples for subsequent stress simulation. This improves the model's generalization ability to real construction variations and avoids overfitting problems caused by relying solely on limited historical data.

[0023] Step S200: Perform stress simulation on each virtual steel structure model to determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load. Correlate the mapping between the virtual steel structure model and the cumulative stress parameters to construct a virtual steel structure stress analysis model.

[0024] Step S200 is based on finite element analysis (FEA) to perform multi-physics coupled stress simulation on a virtual steel structure model, establishing an accurate mapping relationship between the model structure and stress accumulation parameters, and forming a transferable virtual steel structure stress analysis model. Specifically, firstly, adaptive multi-scale meshing is performed on linear segments and key nodes, and nested sub-mesh is densified at high-load nodes and potential weld locations to capture local stress concentrations; then, a sequence of real engineering loads, including static loads, cyclic fatigue, and random vibrations, is applied to calculate the stress superposition effect (such as Von Mises stress accumulation) at key nodes.

[0025] Step S300: Using a virtual steel structure stress analysis model, analyze the currently constructed steel structure to determine the current stress distribution characteristics.

[0026] Step S300 utilizes a pre-constructed virtual steel structure stress analysis model. Through structural similarity matching or parameter mapping, it rapidly infers the stress distribution of the current steel structure under construction, avoiding time-consuming full-scale finite element simulations for each actual structure. Specifically, it compares the topological features (nodes, linear segments, load involvement) of the current steel structure with a virtual model library, selects the stress mapping relationship of the most similar virtual model, and directly derives the stress distribution characteristics of the current structure, including the identification of high stress concentration areas. For example, in the current steel frame structure scanned at the construction site, if its node bifurcation pattern highly matches a certain virtual model, the stress accumulation parameter distribution of that virtual model under a preset load can be directly inherited. This step significantly improves evaluation efficiency.

[0027] Step S400: Using visual analysis technology, determine the welding quality of welds at different locations of the steel structure currently being constructed, and combine this with the current stress distribution characteristics to conduct a comprehensive welding quality assessment of the steel structure currently being constructed.

[0028] The principle of step S400 is to perform a weighted fusion of the local quality features of the weld seam extracted by machine vision technology and the stress distribution features obtained in step S300 to achieve a comprehensive welding quality assessment, avoiding the underestimation of risk caused by simply ignoring positional stress differences in visual inspection. Specifically, this includes: firstly, accurately identifying the welding quality level from a preset database through weld seam image grayscale conversion, edge extraction, center parallel line spacing analysis, and grayscale block matching (combining a corrected overlap parameter formula and comparison of prominent blocks); then analyzing the weld seam spatial routing, length, width, and other features, and retrieving the welding strength from a strength database; finally, calculating the sensitive weights of strength and cumulative stress parameters to correct for quality differences. For example, a slight lack of fusion defect (medium visual score) at a high-stress node will be upgraded to a severe level due to its higher weight, while the same defect in a low-stress area will retain its original rating. The technological innovation of this fusion mechanism lies in the introduction of stress-sensitive weight correction, which realizes a graded and quantitative assessment of defect risk, significantly improving the objectivity and safety of the assessment, and meeting the needs of high-quality welding control in large steel structure projects.

[0029] In some embodiments disclosed in this invention, the method for constructing a corresponding historical steel structure model based on historical steel structure data includes: Step S101: Construct a spatial coordinate system, analyze historical steel structure data, identify several individual steel structure components, determine the main extension direction of each individual steel structure component, and replace it with structural extension lines.

[0030] Step S102: Align the steel structure unit components and the structural extension lines, and connect the structural extension lines point-to-point based on the combined welding positions of the steel structure unit components to obtain the historical steel structure model. Based on the historical steel structure data, determine the load intervention characteristics of different nodes in the historical steel structure model, including the stress changes of different nodes.

[0031] In some embodiments disclosed in this invention, the method for constructing several supplementary steel structure models based on historical steel structure models includes: Step S103: Perform topological abstraction on the linear connection structure of the historical steel structure model, identify key nodes, including endpoints, intersections and bifurcation points, linear segment lengths, angle distributions and load intervention characteristics at the nodes, including force direction and intensity changes, to form a parameterized historical linear topology template.

[0032] Step S104 introduces a constrained mutation generation mechanism. Based on the parameterized historical linear topology template, the global topology parameters are simultaneously subjected to restricted random perturbations, including the number of key nodes, bifurcation mode, span ratio between nodes, line segment length, line segment angle offset, load force direction of location nodes, and load intensity changes, to obtain a supplementary linear topology template.

[0033] Step S105: Based on the supplementary linear topology template, the historical steel structure model corresponding to the historical topology template is structurally adjusted to obtain the supplementary steel structure model.

[0034] In some embodiments disclosed in this invention, the method for stress simulation of a virtual steel structure model includes: Step S201: Perform finite element mesh adaptive generation on the linear segments and key nodes in the virtual steel structure model. Introduce multi-scale nested meshes near nodes with high load intervention features and at weld connection locations, including macroscopic overall meshes and local sub-model meshes.

[0035] The core principle of step S201 is to use adaptive meshing technology in finite element analysis (FEA) to intelligently optimize the mesh of linear segments and key nodes of the virtual steel structure model, and to introduce multi-scale nested mesh structures near nodes with high load characteristics and at weld joint locations. Specifically, adaptive meshing automatically adjusts the mesh density based on the initially calculated stress gradient or error estimate, avoiding the waste of computational resources caused by global uniform refinement; multi-scale nesting includes a macroscopic overall mesh to capture the global deformation and load distribution of the structure, and a local sub-model mesh (sub-model method) to simulate stress concentration phenomena in the weld region with high accuracy.

[0036] Step S202: Based on the load intervention characteristics of the virtual steel structure model, apply a multi-physics coupled dynamic loading sequence, including static load, cyclic fatigue load and random vibration load, analyze the stress superposition effect of key nodes connected by the structural extension line, and obtain the stress accumulation parameters of the key nodes.

[0037] Step S202 is based on the load intervention characteristics of the virtual steel structure model. It constructs a dynamic loading sequence coupled with multiple physics fields to simulate the entire life cycle of stress. The focus is on analyzing the stress superposition effect at key nodes connecting the structural extension lines, ultimately quantifying the cumulative stress parameters. Specifically, the loading sequence comprehensively considers multiple load sources from actual engineering projects: static loads simulate constant gravity or preload, cyclic fatigue loads simulate repeated traffic or mechanical vibrations, and random vibration loads simulate wind loads or seismic excitations. Through multi-physics field coupling (such as force-thermal-vibration interaction), the accumulation of Von Mises equivalent stress in the time domain is calculated (e.g., based on Miner's linear cumulative damage theory or rainflow counting method). For example, in a virtual high-rise building model, wind-induced random vibrations and cyclic fatigue caused by vehicles can lead to multiple stress superpositions at nodes, potentially exceeding the material fatigue limit, while remaining safe under a single static load. The innovative value of this step lies in realistically reproducing the dynamic multi-field coupling effect during the service of steel structures, accurately quantifying the long-term damage risk of key nodes (especially welded joints), avoiding the problem of underestimating fatigue failure in traditional static analysis, and thus providing high-confidence stress distribution data for subsequent comprehensive quality assessment combined with visual inspection.

[0038] In some embodiments disclosed in this invention, the method for determining the welding quality of welds at different locations in a currently constructed steel structure using visual analysis technology includes: Step S401: Take a picture of the welded joint of the steel structure being constructed to obtain an image of the welded joint, and use visual analysis technology to perform grayscale processing on the image of the welded joint to obtain a grayscale image of the welded joint.

[0039] Step S402: Perform edge analysis on the grayscale image of the weld to determine the weld edge in the grayscale image of the weld, and calibrate the weld edge in the grayscale image of the weld to determine the grayscale area of ​​the weld within the weld edge.

[0040] Step S403: Analyze the edge structure features of the weld edge, and find a suitable comparison weld gray area in the preset weld gray area library based on the edge structure features. Select the most suitable weld gray area from the comparison weld gray area, and evaluate the welding quality corresponding to the most suitable weld gray area as the weld welding quality of the current weld gray area.

[0041] In some embodiments disclosed in this invention, the method for analyzing the edge structural features of the weld edge includes: Step S4031: Determine the center parallel line between the weld edges, uniformly set a number of width detection points on the center parallel line, and analyze the weld spacing between the weld edges at the width detection points.

[0042] Step S4032: The central parallel line representation diagram formed by the central parallel lines and the weld spacing set on the central parallel lines are identified as edge structure features.

[0043] In some embodiments disclosed in this invention, the method for finding a suitable comparison weld grayscale region in a preset weld grayscale region library based on edge structural features includes: Step S40321: Based on the matching degree between the center parallel line representation images, the corresponding weld grayscale area is selected as the comparison weld grayscale area. The method for analyzing the matching degree between the center parallel line representation images includes: Align the representation diagrams of the central parallel lines, compare each relative central parallel line, analyze the intersection angle and length difference between them, and determine the line overlap parameter between the central parallel lines based on the value range of the intersection angle and length difference. If there is a central parallel line that does not have a corresponding central parallel line, then the overlap parameter corresponding to the central parallel line is determined to be the preset minimum value. Analyze the spacing difference of weld seams on the corresponding center parallel lines, and calculate the average and variance of the spacing difference, denoted as the average spacing difference and the variance of the spacing difference. Determine the value range to which the average spacing difference and the variance of the spacing difference belong, and output the weld seam spacing matching parameters between the center parallel lines based on their respective value ranges. Based on the weld spacing matching parameter, the line overlap parameter is corrected to obtain the corrected line overlap parameter. Based on the average overlap parameter, overlap parameter variance and minimum overlap parameter of the corrected line overlap parameters of all center parallel lines, the matching degree between the center parallel line representation diagrams is determined. The expression for calculating the overlap parameter between lines is as follows: ; Where H is the corrected inter-line coincidence parameter, h is the inter-line coincidence parameter, W is the weld spacing coincidence parameter, K is the coincidence parameter influence adjustment coefficient, and b is the coincidence parameter influence adjustment constant. The expression for calculating the degree of matching is: ; Where P represents the degree of matching. For the average coincident parameter, For the variance of the overlapping parameters, For the lowest overlap parameter, This is a variance correction coefficient determination function that outputs the corresponding variance correction coefficient based on the preset interval to which the variance of the coincident parameter belongs. This is the function for determining the correction coefficient of the lowest coincident parameter. Based on the preset interval to which the lowest coincident parameter belongs, it outputs the corresponding correction coefficient of the lowest coincident parameter.

[0044] K: The constant coefficient for adjusting the influence of the weld spacing matching parameter W on the overall coincidence parameter. The larger the K value, the higher the weight of the spacing matching degree, reflecting the importance attached to the consistency of weld width.

[0045] b: The constant bias term, which is used to adjust the baseline offset of the exponential function, prevents over- or under-correction when W is small. It is equivalent to adjusting the "noise floor" level of the correction curve.

[0046] In some embodiments disclosed in this invention, the method for selecting the most suitable weld grayscale region from the comparison weld grayscale region includes: Step S4033: Compare the current weld grayscale area with the comparison weld grayscale area, including comparing the average grayscale values ​​of different blocks between the two, calculating the difference in the average grayscale value of each block, identifying the grayscale protruding blocks between the two, determining the center distance of the grayscale protruding blocks, and the difference in block area.

[0047] Step S4034: Compare the difference in the average gray value of each location block and record it as the average gray value difference. Determine the location blocks whose average gray value difference is less than or equal to a preset value and record them as average gray value equal location blocks. Calculate the ratio of average gray value equal location blocks to the number of average gray value equal locations in all location blocks.

[0048] Step S4035: If the distance between the center of the blocks and the difference in the area of ​​the blocks are both less than or equal to the preset values, then the gray-scale protruding blocks are identified as equal gray-scale protruding blocks. The ratio of equal gray-scale protruding blocks to the total number of gray-scale protruding blocks is calculated and recorded as the ratio of equal gray-scale protruding blocks.

[0049] Step S4036: Based on the ratio of the number of equal grayscale values ​​and the ratio of the number of outstanding grayscale blocks, determine the selected weld grayscale region for comparison. The methods for determining grayscale highlighted areas include: The current weld grayscale area and the comparison weld grayscale area are divided into a unified grid to obtain several detection blocks of the same size and position.

[0050] Calculate the average gray value of each location block, as well as the overall average gray value and gray standard deviation of the weld gray area; For each detection block, if the absolute difference between its average gray value and the overall average gray value exceeds a preset gray difference threshold, then the block at that location is identified as a gray-prominent block.

[0051] Merge adjacent grayscale highlight blocks to form connected grayscale highlight blocks.

[0052] In some embodiments disclosed in this invention, a method for comprehensively evaluating the welding quality of a currently constructed steel structure, taking into account the current stress distribution characteristics, includes: Step S401: Analyze the welding characteristics of welds at different locations of the currently constructed steel structure, including determining the weld routing length and width, and determining the spatial characteristics of the weld. The spatial characteristics include the spatial routing of the weld, denoted as the spatial welding routing structure.

[0053] Step S402: The spatial welding trace structure, trace length, and trace width are compared with a preset weld strength database to determine the weld strength. The preset weld strength database is a database of preset combinations of weld welding features and weld strengths. The correspondence between the weld welding features and weld strengths comes from preset strength tests.

[0054] Step S403: Calculate the ratio of welding strength to cumulative stress parameter, and record it as the welding quality sensitive weight. Based on the welding quality sensitive weight, correct the quality difference between the corresponding weld quality and the standard weld quality to obtain a comprehensive welding quality assessment.

[0055] In some embodiments disclosed in this invention, a steel structure construction welding quality assessment system based on visual analysis includes: The first module is used to obtain a number of historical steel structure data for projects that require historical engineering records, and to build a corresponding historical steel structure model based on the historical steel structure data. Based on the historical steel structure model, a number of supplementary steel structure models are also built. The historical steel structure model and the supplementary steel structure models are collectively referred to as the virtual steel structure model. The second module is used to perform stress simulation on each virtual steel structure model, determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load, associate the mapping between the virtual steel structure model and the cumulative stress parameters, and construct a virtual steel structure stress analysis model. The third module is used to analyze the currently constructed steel structure using a virtual steel structure stress analysis model to determine the current stress distribution characteristics. The fourth module is used to determine the welding quality of welds at different locations in the currently constructed steel structure using visual analysis technology, and to conduct a comprehensive welding quality assessment of the currently constructed steel structure in combination with the current stress distribution characteristics.

[0056] This invention discloses a method and system for assessing the welding quality of steel structure construction based on visual analysis, belonging to the field of welding engineering management technology. It includes constructing a historical steel structure model and generating several supplementary steel structure models based on this model through constrained mutations; these two are collectively referred to as virtual steel structure models. Stress simulation is performed on each virtual steel structure model to determine the cumulative stress parameters of different locations under preset loads. Visual analysis technology is used to perform image processing and feature recognition on the current steel structure welds to determine the welding quality of welds at each location, and a comprehensive welding quality assessment is performed based on the current stress distribution characteristics. This invention achieves accurate weld quality assessment considering structural stress distribution through the organic combination of virtual model stress simulation and visual inspection, improving the objectivity and reliability of the assessment, and is suitable for quality control in large-scale steel structure engineering construction.

[0057] Through the above description of the embodiments, those skilled in the art can clearly understand that the present invention can be implemented in hardware or by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) and includes several instructions to cause a computer device (such as a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for assessing the welding quality of steel structure construction based on visual analysis, characterized in that, include: Step S100: For the historical engineering records that are required, obtain a number of historical steel structure data, construct a corresponding historical steel structure model based on the historical steel structure data, and construct a number of supplementary steel structure models based on the historical steel structure models. The historical steel structure model and the supplementary steel structure models are collectively referred to as the virtual steel structure model. Step S200: Perform stress simulation on each virtual steel structure model to determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load. Correlate the mapping between the virtual steel structure model and the cumulative stress parameters to construct a virtual steel structure stress analysis model. Step S300: Using a virtual steel structure stress analysis model, analyze the currently constructed steel structure to determine the current stress distribution characteristics; Step S400: Using visual analysis technology, determine the welding quality of welds at different locations of the steel structure currently being constructed, and combine this with the current stress distribution characteristics to conduct a comprehensive welding quality assessment of the steel structure currently being constructed.

2. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 1, characterized in that, Methods for constructing corresponding historical steel structure models based on historical steel structure data include: Step S101: Construct a spatial coordinate system, analyze historical steel structure data, identify several individual steel structure components, determine the main extension direction of each individual steel structure component, and replace it with structural extension lines. Step S102: Align the steel structure unit components and the structural extension lines, and connect the structural extension lines point-to-point based on the combined welding positions of the steel structure unit components to obtain the historical steel structure model. Based on the historical steel structure data, determine the load intervention characteristics of different nodes in the historical steel structure model, including the stress changes of different nodes.

3. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 2, characterized in that, The methods for constructing several supplementary steel structure models based on historical steel structure models include: Step S103: Perform topological abstraction on the linear connection structure of the historical steel structure model, identify key nodes, including endpoints, intersections and bifurcation points, linear segment lengths, angle distributions and load intervention characteristics at the nodes, including force direction and intensity changes, to form a parameterized historical linear topology template. Step S104 introduces a constrained mutation generation mechanism. Based on the parameterized historical linear topology template, the global topology parameters are simultaneously subjected to constrained random perturbations, including the number of key nodes, bifurcation mode, span ratio between nodes, line segment length, line segment angle offset, load force direction of position nodes, and load intensity changes, to obtain a supplementary linear topology template. Step S105: Based on the supplementary linear topology template, the historical steel structure model corresponding to the historical topology template is structurally adjusted to obtain the supplementary steel structure model.

4. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 2, characterized in that, Methods for stress simulation of virtual steel structure models include: Step S201: Perform finite element mesh adaptive generation on the linear segments and key nodes in the virtual steel structure model. Introduce multi-scale nested meshes, including macroscopic overall meshes and local sub-model meshes, near nodes with high load intervention features and weld connection locations. Step S202: Based on the load intervention characteristics of the virtual steel structure model, apply a multi-physics coupled dynamic loading sequence, including static load, cyclic fatigue load and random vibration load, analyze the stress superposition effect of key nodes connected by the structural extension line, and obtain the stress accumulation parameters of the key nodes.

5. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 2, characterized in that, Methods for determining the welding quality of welds at different locations in a currently constructed steel structure using visual analysis technology include: Step S401: Take a picture of the welded joint of the steel structure being constructed to obtain an image of the welded joint, and use visual analysis technology to perform grayscale processing on the image of the welded joint to obtain a grayscale image of the welded joint. Step S402: Perform edge analysis on the grayscale image of the weld joint to determine the weld edge in the grayscale image of the weld joint, and then calibrate the weld edge in the grayscale image of the weld joint to determine the grayscale area of ​​the weld within the weld edge. Step S403: Analyze the edge structure features of the weld edge, and find a suitable comparison weld gray area in the preset weld gray area library based on the edge structure features. Select the most suitable weld gray area from the comparison weld gray area, and evaluate the welding quality corresponding to the most suitable weld gray area as the weld welding quality of the current weld gray area.

6. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 5, characterized in that, Methods for analyzing the edge structure characteristics of weld seams include: Step S4031: Determine the center parallel line between the weld edges, uniformly set a number of width detection points on the center parallel line, and analyze the weld spacing between the weld edges at the width detection points. Step S4032: The central parallel line representation diagram formed by the central parallel lines and the weld spacing set on the central parallel lines are identified as edge structure features.

7. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 6, characterized in that, Methods for finding suitable weld grayscale regions for comparison based on edge structure features in a pre-defined weld grayscale region library include: Step S40321: Based on the matching degree between the center parallel line representation images, the corresponding weld grayscale area is selected as the comparison weld grayscale area. The method for analyzing the matching degree between the center parallel line representation images includes: Align the representation diagrams of the central parallel lines, compare each relative central parallel line, analyze the intersection angle and length difference between them, and determine the line overlap parameter between the central parallel lines based on the value range of the intersection angle and length difference. If there is a central parallel line that does not have a corresponding central parallel line, then the overlap parameter corresponding to the central parallel line is determined to be the preset minimum value. Analyze the spacing difference of weld seams on the corresponding center parallel lines, and calculate the average and variance of the spacing difference, denoted as the average spacing difference and the variance of the spacing difference. Determine the value range to which the average spacing difference and the variance of the spacing difference belong, and output the weld seam spacing matching parameters between the center parallel lines based on their respective value ranges. Based on the weld spacing matching parameter, the line overlap parameter is corrected to obtain the corrected line overlap parameter. Based on the average overlap parameter, overlap parameter variance and minimum overlap parameter of the corrected line overlap parameters of all center parallel lines, the matching degree between the center parallel line representation diagrams is determined. The expression for calculating the overlap parameter between lines is as follows: ； Where H is the corrected inter-line coincidence parameter, h is the inter-line coincidence parameter, W is the weld spacing coincidence parameter, K is the coincidence parameter influence adjustment coefficient, and b is the coincidence parameter influence adjustment constant. The expression for calculating the degree of matching is: ； Where P represents the degree of matching. For the average coincident parameter, For the variance of the overlapping parameters, For the lowest overlap parameter, This is a variance correction coefficient determination function that outputs the corresponding variance correction coefficient based on the preset interval to which the variance of the coincident parameter belongs. This is the function for determining the correction coefficient of the lowest coincident parameter. Based on the preset interval to which the lowest coincident parameter belongs, it outputs the corresponding correction coefficient of the lowest coincident parameter.

8. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 5, characterized in that, Methods for selecting the most suitable weld grayscale region in the comparison weld grayscale region include: Step S4033: Compare the current weld grayscale area with the comparison weld grayscale area, including comparing the average grayscale values ​​of different blocks between the two, calculating the difference in the average grayscale value of each block, identifying the grayscale protruding blocks between the two, determining the block center distance of the grayscale protruding blocks, and the block area difference. Step S4034: Compare the difference in the average gray value of each location block, record it as the average gray value difference, and determine the location blocks whose average gray value difference is less than or equal to a preset value, record them as average gray value equal location blocks, and calculate the ratio of average gray value equal location blocks to the number of average gray value equal locations in all location blocks. Step S4035: If the distance between the center of the blocks and the difference in the area of ​​the blocks are both less than or equal to the preset values, then the gray-scale protruding blocks are identified as equal gray-scale protruding blocks. The ratio of equal gray-scale protruding blocks to the total number of gray-scale protruding blocks is calculated and recorded as the ratio of equal gray-scale protruding blocks. Step S4036: Based on the ratio of the number of equal grayscale values ​​and the ratio of the number of outstanding grayscale blocks, determine the selected weld grayscale region for comparison. The methods for determining grayscale highlighted areas include: The current weld grayscale area and the weld grayscale area for comparison are divided into a unified grid to obtain several detection blocks of the same size and position. Calculate the average gray value of each location block, as well as the overall average gray value and gray standard deviation of the weld gray area; For each detection block, if the absolute difference between its average gray value and the overall average gray value exceeds the preset gray value difference threshold, then the block at that location is identified as a gray-prominent block. Merge adjacent grayscale highlight blocks to form connected grayscale highlight blocks.

9. The method for assessing the welding quality of steel structure construction based on visual analysis according to claim 1, characterized in that, Methods for comprehensively assessing the welding quality of currently constructed steel structures, taking into account current stress distribution characteristics, include: Step S401: Analyze the welding characteristics of welds at different locations of the steel structure being constructed, including determining the weld routing length and width, and determining the spatial characteristics of the weld. The spatial characteristics include the spatial routing of the weld, referred to as the spatial welding routing structure. Step S402: The spatial welding wiring structure, wiring length, and wiring width are compared with the preset weld strength database to determine the welding strength of the weld. The preset weld strength database is a database of preset combinations of welding features and welding strengths. The combination of welding features and welding strengths is derived from preset strength tests. Step S403: Calculate the ratio of welding strength to cumulative stress parameter, and record it as the welding quality sensitive weight. Based on the welding quality sensitive weight, correct the quality difference between the corresponding weld quality and the standard weld quality to obtain a comprehensive welding quality assessment.

10. A steel structure construction welding quality assessment system based on visual analysis, characterized in that, A method for evaluating the welding quality of steel structure construction for performing any one of claims 1-9, comprising: The first module is used to obtain a number of historical steel structure data for projects that require historical engineering records, and to build a corresponding historical steel structure model based on the historical steel structure data. Based on the historical steel structure model, a number of supplementary steel structure models are also built. The historical steel structure model and the supplementary steel structure models are collectively referred to as the virtual steel structure model. The second module is used to perform stress simulation on each virtual steel structure model, determine the cumulative stress parameters of the virtual steel structure model in different locations under the preset load, associate the mapping between the virtual steel structure model and the cumulative stress parameters, and construct a virtual steel structure stress analysis model. The third module is used to analyze the currently constructed steel structure using a virtual steel structure stress analysis model to determine the current stress distribution characteristics. The fourth module is used to determine the welding quality of welds at different locations in the currently constructed steel structure using visual analysis technology, and to conduct a comprehensive welding quality assessment of the currently constructed steel structure in combination with the current stress distribution characteristics.

Images