A carbon steel pipe welding quality evaluation system

Abstract

The present application relates to the technical field of welding quality evaluation, in particular to a carbon steel pipe welding quality evaluation system, which comprises a welding area three-dimensional scanning module, a weld dynamic grid generation module, a welding form change analysis module, a weld structure graph construction module, a weld quality graph convolution network module and a welding defect intelligent identification module.In the present application, the real-time adjustment of the dynamic grid generation module can capture the slight changes of the weld geometry, the generation of the welding form feature data and the construction of the weld structure graph provide a new dimension for the welding quality evaluation, the introduction of the weld quality graph convolution network module analyzes the node relationship and structure characteristics in the weld graph structure through the deep learning technology, which provides an intelligent means for the welding quality evaluation, the welding defect intelligent identification module optimizes the identification process and can accurately classify various welding defects, which provides strong technical support for the guarantee of welding quality.

Description

Technical Field

[0001] This invention relates to the field of welding quality assessment technology, and in particular to a carbon steel pipe welding quality assessment system. Background Technology

[0002] The field of welding quality assessment technology focuses on inspecting and verifying the integrity, stability, and strength of welded joints or welds to ensure they meet specific industry standards and usage requirements. This field combines materials science, mechanical engineering, and physical testing techniques to identify potential defects in welds, such as cracks, voids, and inclusions, through a range of testing methods. These defects affect the safety and reliability of the structure. Assessment techniques include, but are not limited to, non-destructive testing (NDT), destructive testing, and chemical and physical analysis methods. With technological advancements, the field of welding quality assessment continuously incorporates new technologies and methods, such as digital imaging, automated inspection, and machine learning algorithms, to improve the accuracy and efficiency of inspections.

[0003] A carbon steel pipe welding quality assessment system is a system specifically designed to evaluate the quality of carbon steel pipe welds. Its purpose is to ensure that the welds formed during the welding process meet predetermined safety and performance standards. This system typically assesses the uniformity, continuity, and presence of defects in the weld to ensure the safety and reliability of the pipeline when transporting fluids (whether gas or liquid). By using this assessment system, welding problems that lead to pipeline failures, such as cracks, porosity, or incomplete penetration, can be detected and corrected in a timely manner, thereby reducing maintenance costs, extending pipeline lifespan, and ensuring transportation safety.

[0004] Traditional systems often struggle to accurately capture and assess minute defects within welds due to technological limitations, especially when weld morphology is complex or varied. Existing methods cannot adapt and accurately assess these defects in real time, leading to potentially inaccurate results and failing to fully guarantee structural safety and reliability. The lack of efficient automated and intelligent assessment tools makes defect identification time-consuming and experience-dependent, reducing assessment efficiency and increasing maintenance costs. Furthermore, insufficient ability to identify complex defects affects the comprehensiveness and accuracy of the assessment, limiting its application in high-standard and demanding scenarios. These shortcomings result in increased safety risks, higher maintenance costs, and shorter pipeline lifespans, adversely impacting production and operations. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a carbon steel pipe welding quality assessment system.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a carbon steel pipe welding quality assessment system comprising:

[0007] The 3D scanning module for the welding area uses 3D scanning technology to scan the welding area of ​​carbon steel pipes, capture the structural features of the weld, record geometric data, and generate 3D welding scanning data.

[0008] The weld dynamic mesh generation module uses the three-dimensional welding scan data to divide the mesh, adjust each mesh unit in real time to match the changes in weld geometry, refine the mesh units and capture detailed information to generate a weld dynamic mesh model.

[0009] The welding morphology change analysis module, based on the weld dynamic mesh model, monitors the changes in weld width, depth, and edge contour, extracts the changing features, and generates welding morphology feature data.

[0010] The weld structure diagram construction module maps the weld and defect areas to graph nodes using the welding morphology feature data, defines edges based on the geometric characteristics and positional relationships between nodes, constructs a graph representation of the weld, and generates a carbon steel pipe weld diagram structure.

[0011] The weld quality graph convolutional network module uses the aforementioned carbon steel pipe weld graph structure to extract features, analyze the node relationships and structural features in the graph representation of the weld, extract weld quality association information, and generate weld quality feature representation.

[0012] The intelligent welding defect identification module identifies and classifies defect types based on the weld quality characteristics, distinguishes different types of welding defects, and generates welding defect identification results.

[0013] As a further aspect of the present invention, the three-dimensional welding scanning data includes a set of position coordinates, depth values, width values, and a set of surface features; the weld dynamic mesh model includes mesh cell size, a weld morphology description set, and a cell connection diagram; the welding morphology feature data includes width variation, depth variation, and an edge contour morphology set; the carbon steel pipe weld image structure includes a spatial distribution diagram, a connection property diagram, and a regional node classification diagram; and the weld quality feature representation includes an overall evaluation index, a set of local defect feature vectors, and quality classification labels.

[0014] As a further aspect of the present invention, the weld dynamic mesh generation module includes:

[0015] The mesh initialization submodule uses the voxelization method based on the three-dimensional welding scan data to determine whether the data covers the entire weld area. If so, it creates an initial mesh frame and automatically divides the weld into multiple basic mesh units according to the geometry of the weld. It sets the initial size and shape parameters for each unit to generate an initial mesh model.

[0016] The mesh adjustment submodule uses the initial mesh model to monitor the real-time changes in the weld shape. If a change is detected, it dynamically adjusts the size and shape of the mesh cells to match the current geometric features of the weld. By calculating the amount of change in the weld geometry, it adjusts the distribution and density of adjacent mesh cells to obtain the adjusted mesh model.

[0017] The detail capture submodule iteratively improves the model's ability to capture weld details through the adjusted mesh model. By analyzing the geometric characteristics of the weld area, it identifies key geometric feature areas. If the model needs to capture geometric features, it increases the mesh density in the key areas, performs local refinement processing on the model, and fine-tunes the size of the mesh cells to obtain a dynamic mesh model of the weld.

[0018] As a further aspect of the present invention, the analysis of the geometric characteristics of the weld region is performed using the following formula:

[0019] ;

[0020] Calculate the geometric feature deviation value, combine it with the geometric characteristics of the weld area, calculate the local feature parameters through the key area data, identify the key geometric feature area, if the model needs to capture geometric features, increase the mesh density in the key area, perform local refinement processing on the model, fine-tune the size of the mesh cells, and obtain the dynamic mesh model of the weld.

[0021] in, Represents the geometric feature deviation value. Represents the weld area Local curvature length at each sampling point This represents the average curvature length of all weld sampling points. Representing the The area of ​​each mesh cell in the weld zone. Representing the The density of mesh cells in the weld area, This represents the total number of mesh cells within the weld area. This represents a summation operation on all grid cells. This represents the absolute deviation between the weld curvature length and the average curvature length.

[0022] As a further aspect of the present invention, the welding morphology change analysis module includes:

[0023] The width monitoring submodule analyzes the distribution characteristics of grid cells along the weld length based on the weld dynamic grid model, and records the width data at each position by comparing the distance changes of adjacent grid cells in the lateral direction, thereby generating weld width change data.

[0024] The depth monitoring submodule uses the weld width variation data to analyze the changes in grid cells along the weld depth. By measuring the height difference of differentiated depth grid cells, it records the distance from the surface of the weld to the deepest point, extracts the depth variation trend along the weld length, and obtains weld depth variation data.

[0025] The edge contour monitoring submodule identifies changes in the weld edge contour through the weld depth change data, analyzes the outer lines of the mesh model, compares the changes in edge contour lines at different time points, records the undulations and offsets of the edge contour, tracks the trajectory of edge contour changes, and obtains weld edge contour data.

[0026] The feature extraction submodule integrates the width, depth and edge contour variation information based on the weld edge contour data, analyzes the key variation features of the weld morphology, and generates welding morphology feature data by comparing and analyzing the variation rules and patterns of the width, depth and edge contour data.

[0027] As a further aspect of the present invention, the weld structure diagram construction module includes:

[0028] The welding morphology recognition submodule analyzes the geometric dimensions and morphology of the weld and defect areas based on the welding morphology feature data and sets node attributes. At the same time, it locates the position of the weld and defect, assigns a unique identifier to each graph node, maps the weld and defect features to the graph node attributes, and generates welding morphology feature analysis results.

[0029] The graph node mapping submodule uses the welding morphology feature analysis results and a mapping algorithm to convert the description of the weld and defect areas into graph nodes. Each graph node is assigned a unique identifier and a corresponding attribute value, and is quantified according to the morphological features of the weld and defects. At the same time, the nodes are connected to each other according to the spatial relationship between the weld and defects to generate graph node mapping results.

[0030] The edge definition and graph construction submodule sorts the nodes and establishes connection paths based on the graph node mapping results using spatial location information. It adjusts the connection strength parameters according to the characteristics of the connection paths and refines the connection paths and strength parameters through iterative optimization to obtain the carbon steel pipe weld diagram structure.

[0031] As a further aspect of the present invention, the weld quality map convolutional network module includes:

[0032] The structural analysis submodule is based on the carbon steel pipe weld diagram structure. It analyzes each node in the diagram and its connection method, identifies the position and role of the weld or defect area represented by each node in the overall weld structure, analyzes the direct and indirect connection paths between nodes, and generates weld structure depth analysis results.

[0033] The quality influencing factor identification submodule, based on the weld structure depth analysis results, applies network analysis algorithms to analyze the interaction between nodes and their potential impact on the overall weld quality. If the target node connection pattern is found to be related to quality defects, the node connection pattern is analyzed in detail to identify the key nodes and connection methods that affect the weld quality and obtain the key quality influencing factors.

[0034] The quality characteristic integration submodule integrates multiple sets of key quality influencing factors, analyzes the common effects of these factors on weld quality, studies the interaction between key nodes and connection modes, analyzes the impact factors of these interactions on weld quality, and generates a weld quality characteristic representation.

[0035] As a further aspect of the present invention, the network analysis algorithm adopts the formula

[0036]

[0037] Improved betweenness centrality of computing nodes, where, For nodes Improved middle centrality For nodes The weight, For nodes and The weights of the paths between them. For nodes Position parameters, For path arrive The length parameter, For the nodes The number of shortest paths, For nodes To the node The total number of shortest paths.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0039] In this invention, the real-time adjustment of the dynamic mesh generation module enables the precise capture of minute changes in weld geometry, thus making the analysis of weld morphology variations more accurate. The generation of weld morphology feature data and the construction of weld structure diagrams provide a new dimension for weld quality assessment, allowing defect identification to go beyond surface-level analysis and incorporate deep structural correlations, significantly improving the comprehensiveness and depth of defect identification. The introduction of the weld quality diagram convolutional network module, through deep learning technology to analyze the node relationships and structural features in the weld diagram structure, provides an intelligent means for weld quality assessment, effectively improving assessment speed and accuracy. The intelligent weld defect identification module further optimizes the identification process, enabling rapid and accurate classification of various weld defects, providing strong technical support for ensuring weld quality. Attached Figure Description

[0040] Figure 1 This is a system flowchart of the present invention;

[0041] Figure 2 This is a schematic diagram of the system framework of the present invention;

[0042] Figure 3 This is a flowchart of the three-dimensional scanning module for the welding area of ​​the present invention;

[0043] Figure 4 This is a flowchart of the weld dynamic mesh generation module of the present invention;

[0044] Figure 5 This is a flowchart of the welding morphology change analysis module of the present invention;

[0045] Figure 6 This is a flowchart of the weld structure diagram construction module of the present invention;

[0046] Figure 7 This is a flowchart of the weld quality image convolutional network module of the present invention;

[0047] Figure 8 This is a flowchart of the intelligent welding defect identification module of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Example 1

[0051] Please see Figures 1 to 2 A carbon steel pipe welding quality assessment system includes:

[0052] The 3D scanning module for the welding area uses 3D scanning technology to scan the welding area of ​​carbon steel pipes, capture the structural features of the weld, record geometric data, and generate 3D welding scanning data.

[0053] The weld dynamic mesh generation module uses 3D welding scan data to divide the mesh, adjusts each mesh unit in real time to match the changes in weld geometry, refines the mesh units and captures detailed information to generate a weld dynamic mesh model.

[0054] The welding morphology change analysis module is based on the dynamic mesh model of the weld, monitors the changes in weld width, depth and edge contour, extracts the features of the changes, and generates welding morphology feature data.

[0055] The weld structure diagram construction module maps welds and defect areas to graph nodes using welding morphology feature data, defines edges based on the geometric characteristics and positional relationships between nodes, constructs a graph representation of the weld, and generates a graph structure of carbon steel pipe welds.

[0056] The weld quality graph convolutional network module uses the weld graph structure of carbon steel pipe to extract features, analyze the node relationships and structural features in the graph representation of the weld, extract the weld quality correlation information, and generate the weld quality feature representation.

[0057] The intelligent welding defect identification module identifies and classifies defect types by representing weld quality characteristics, distinguishes different types of welding defects, and generates welding defect identification results.

[0058] The 3D welding scan data includes a set of location coordinates, depth values, width values, and surface feature sets. The weld dynamic mesh model includes mesh cell sizes, weld morphology description sets, and cell connection diagrams. Weld morphology feature data includes width variation, depth variation, and edge contour morphology sets. The carbon steel pipe weld image structure includes a spatial distribution map, connection property map, and regional node classification map. Weld quality feature representations include an overall evaluation index, a set of local defect feature vectors, and quality classification labels. Weld defect identification results include a type identification set, location coordinates, size estimates, and severity levels.

[0059] Please see Figure 3 The 3D scanning module for the welding area includes:

[0060] The weld positioning submodule is based on three-dimensional scanning technology. It performs preliminary positioning of the carbon steel pipe welding area, adjusts the scanning equipment to align with the welding line, records the coordinates of the welding area, measures the length and width of the welding area, controls the scanning path to cover the entire welding area, and generates weld positioning information.

[0061] The weld positioning submodule is based on 3D scanning technology to perform preliminary positioning of the carbon steel pipe welding area. It uses an optical 3D scanner to perform the scanning operation, setting the scanner parameters to include a scanning speed of 5mm / s and a scanning resolution of 0.01mm. The center point coordinates of the welding area are located through a 3D coordinate measurement system. The welding line is automatically aligned using QR code positioning technology, and the coordinate data of the welding area is collected. The length and width of the welding area are calculated using a linear interpolation algorithm, and the scanning path is controlled to cover the entire welding area in a spiral shape to generate weld positioning information.

[0062] The structural feature capture submodule uses weld positioning information, adjusts the scanning resolution to refine the capture of weld geometry, synchronously adjusts the scanning speed to match weld size, records the texture information of weld surface, and obtains weld geometry data.

[0063] The structural feature capture submodule uses weld seam positioning information, adjusts the scanning resolution to refine the capture of weld seam geometry, simultaneously adjusts the scanning speed to match weld seam size, records weld seam surface texture information, and acquires weld seam geometric structure data. During this process, a fine scanning mode is used, with a scanning resolution set to 0.005 mm and a scanning speed adjusted to 2 mm / s to match subtle dimensional changes in the weld seam. High dynamic range imaging (HDR) technology is used to capture weld seam surface texture, an edge detection algorithm (Canny algorithm) is applied to refine weld seam edges, and a surface fitting algorithm (least squares method) is simultaneously applied to obtain accurate weld seam geometric structure data.

[0064] The 3D data generation submodule refines the recorded data using weld geometry data, adjusts the data model to match the actual weld geometry, reconstructs the 3D model of the weld, and obtains 3D welding scan data.

[0065] The 3D data generation submodule refines the recorded data using weld geometry data, adjusts the data model to match the actual weld geometry, and reconstructs the 3D model of the weld to obtain 3D welding scan data. In this step, a 3D reconstruction algorithm (Voxel-based Marching Cubes algorithm) is used, with a voxel size of 0.001 mm³. The spatial distribution of 3D data points is adjusted using the acquired geometric data. Point cloud data processing software (e.g., PointCloudLibrary, PCL) is used to filter and denoise the point cloud. Surface reconstruction techniques (e.g., Poisson reconstruction) are applied to convert the processed point cloud into a continuous 3D surface, ultimately generating 3D model data that matches the actual weld.

[0066] Please see Figure 4 The weld dynamic mesh generation module includes:

[0067] The mesh initialization submodule is based on 3D welding scan data. It uses a voxelization method to determine whether the data covers the entire weld area. If so, it creates an initial mesh framework and automatically divides the weld into multiple basic mesh units according to the geometry of the weld. It sets the initial size and shape parameters for each unit and generates an initial mesh model.

[0068] The mesh initialization submodule is based on 3D welding scan data. It uses a voxelization method to convert the 3D scan data into a voxel mesh. The voxel size is set to 1mm³ to determine whether the data covers the entire weld area. If so, an initial mesh framework is created. An automatic mesh generation algorithm (such as Octree generation) is used to automatically divide the weld into multiple basic mesh units according to the geometry of the weld. Initial size and shape parameters are set for each unit, including a unit side length of 2mm. The shape parameters are automatically adjusted according to the local geometric features of the weld to generate the initial mesh model.

[0069] The mesh adjustment submodule uses an initial mesh model to monitor real-time changes in the weld shape. If a change is detected, the size and shape of the mesh cells are dynamically adjusted to match the current geometric features of the weld. By calculating the amount of change in the weld geometry, the distribution and density of adjacent mesh cells are adjusted to obtain the adjusted mesh model.

[0070] The mesh adjustment submodule uses an initial mesh model to monitor real-time changes in the weld shape. Utilizing real-time 3D scanning feedback and combined with dynamic mesh adjustment algorithms, such as the elastic mesh method, it dynamically adjusts the size and shape of mesh cells to match the current geometric features of the weld if a change is detected. By calculating the amount of change in the weld geometry, it uses dynamic interpolation technology to adjust the distribution and density of adjacent mesh cells, including adjusting the mesh cell spacing to 1.5 mm and dynamically increasing or decreasing the density according to the weld changes, thus obtaining the adjusted mesh model.

[0071] The detail capture submodule iteratively improves the model's ability to capture weld details by adjusting the mesh model. It analyzes the geometric characteristics of the weld area, identifies key geometric feature areas, and increases the mesh density in key areas if the model needs to capture geometric features. It performs local refinement of the model, fine-tunes the size of the mesh cells, and obtains a dynamic mesh model of the weld.

[0072] The detail capture submodule iteratively improves the model's ability to capture weld details by adjusting the mesh model. It uses detail enhancement algorithms, such as mesh refinement technology, to analyze the geometric characteristics of the weld area and combine them with local refinement strategies to identify key geometric feature areas. If the model needs to capture geometric features, the mesh density is increased in the key areas, adjusting the density to have more mesh units within 2 mm². The model is then locally refined by fine-tuning the size of the mesh units, such as adjusting the unit size to 1 mm in the detail area, to obtain a dynamic mesh model of the weld.

[0073] The geometric characteristics of the weld area are analyzed using the following formula:

[0074] ;

[0075] Calculate the geometric feature deviation value, combine it with the geometric characteristics of the weld area, calculate the local feature parameters through the key area data, identify the key geometric feature area, if the model needs to capture geometric features, increase the mesh density in the key area, perform local refinement processing on the model, fine-tune the size of the mesh cells, and obtain the dynamic mesh model of the weld.

[0076] in, Represents the geometric feature deviation value. Represents the weld area Local curvature length at each sampling point This represents the average curvature length of all weld sampling points. Representing the The area of ​​each mesh cell in the weld zone. Representing the The density of mesh cells in the weld area, This represents the total number of mesh cells within the weld area. This represents a summation operation on all grid cells. This represents the absolute deviation between the weld curvature length and the average curvature length.

[0077] To analyze the geometric characteristics of the weld region, an initial mesh model of the weld region is first obtained. Discrete point coordinate data for the weld edge and internal regions are selected. Based on the coordinate data, the rate of curvature change of adjacent mesh cells is calculated. For the rate of curvature change of the weld region, the local mean curvature of each mesh cell is calculated. By comparing the changes in the mean curvature of adjacent cells, the geometric feature regions of the weld are obtained. Furthermore, the area of ​​each mesh cell is calculated. and grid density These data are used to evaluate the geometric variation characteristics of the weld region. For example, there is usually a large rate of curvature change in the transition region at the weld edge, while the rate of curvature change is relatively small in the weld center region. To determine whether a region is a critical feature region, the geometric feature deviation value of the weld region mesh element is calculated. , among which, if Higher than the set threshold If the region contains key geometric features, then the mesh density needs to be increased for local refinement, and this threshold is set. Based on the complexity of the weld shape, a higher value is taken when the weld shape is relatively simple, for example... When the weld shape is complex, a lower value should be used, for example... For example, if the mean curvature values ​​of five mesh elements in a certain weld area are respectively , ,

[0078] , , Then its mean curvature

[0079] Then calculate the absolute deviation value of each grid cell. Calculate the sum of these values, then divide by the sum of the area densities of all grid cells to obtain the result. If the value is higher than the set threshold, the mesh density in the local area needs to be increased; otherwise, it remains unchanged. Finally, based on the above calculation results, the mesh density is increased in the key geometric feature areas, and local refinement is performed. Finally, the size of the mesh cells is finely adjusted to obtain the dynamic mesh model of the weld.

[0080] Given that the mesh element areas of a certain weld region are respectively mm2, mm2, mm2,

[0081] mm2, mm2, mesh densities are respectively , , , , Calculate the geometric feature deviation value:

[0082] ;

[0083] ;

[0084] ;

[0085] Compare With threshold ,like Less than If the current region does not belong to the key geometric feature region, there is no need to increase the mesh density. If the mesh density in this area needs to be increased and local refinement is required, the result shows that the geometric features in the weld area change little and the existing mesh generation accuracy can meet the modeling requirements, so no refinement is needed.

[0086] The innovative aspects of the formula:

[0087] The advantage of the formula is that it quantifies the local changes in weld geometry by calculating local curvature changes and area density ratios, and determines the mesh refinement requirements by setting thresholds, thereby achieving accurate capture of key areas.

[0088] Please see Figure 5 The welding morphology change analysis module includes:

[0089] The width monitoring submodule is based on the weld dynamic mesh model. It analyzes the distribution characteristics of mesh cells along the weld length, and records the width data at each position by comparing the changes in the distance between adjacent mesh cells in the lateral direction, thus generating weld width change data.

[0090] The width monitoring submodule is based on the weld dynamic mesh model. It uses a mesh analysis algorithm and calls the NumPy library through the Python programming language to process the mesh data. It traverses the mesh cells along the weld length, calculates the Euclidean distance between adjacent mesh cells in the lateral direction, uses a for loop to traverse each position, and records the width data of each position as a NumPy array to generate weld width change data.

[0091] The depth monitoring submodule uses weld width variation data to analyze the changes in grid cells along the weld depth. By measuring the height difference of differentiated depth grid cells, it records the distance from the surface of the weld to the deepest point, extracts the depth variation trend along the weld length, and obtains weld depth variation data.

[0092] The depth monitoring submodule uses weld width variation data, applies a depth measurement algorithm, and uses the Matplotlib library to draw a depth-width plot. It analyzes the variation trend of grid cells along the weld depth, calculates the height difference between adjacent depth grid cells using the difference method, and uses the map function to map each height difference to depth data. It records the distance from the weld surface to the deepest point as a list to generate weld depth variation data.

[0093] The edge contour monitoring submodule identifies changes in weld edge contours by using weld depth variation data, analyzes the outer lines of the mesh model, compares edge contour line changes at different time points, records edge contour undulations and offsets, tracks the trajectory of edge contour changes, and obtains weld edge contour data.

[0094] The edge contour monitoring submodule uses weld depth change data and edge detection algorithms. It identifies weld edge contours using the Canny edge detection function in the OpenCV library, sets thresholds of 100 and 200, applies edge detection to the weld model, analyzes the obtained edge image, and records the undulations and offsets of the edge contours as a two-dimensional array by traversing the edge pixels. It then tracks the trajectory of edge contour changes to generate weld edge contour data.

[0095] The feature extraction submodule integrates the width, depth, and edge contour variation information based on weld edge contour data to analyze the key variation features of weld morphology. By comparing and analyzing the variation patterns and rules of width, depth, and edge contour data, it generates welding morphology feature data.

[0096] The feature extraction submodule integrates width, depth, and edge contour variation information based on weld edge contour data. It uses a data clustering algorithm, specifically the K-means algorithm executed through the Scikit-learn library, with a cluster size of 3, to perform cluster analysis on key variation features of weld morphology. The fit method is then applied to the set of width, depth, and edge contour data to compare and analyze the variation patterns and rules of width, depth, and edge contour data, generating weld morphology feature data.

[0097] Please see Figure 6 The weld structure diagram construction module includes:

[0098] The welding morphology recognition submodule analyzes the geometric dimensions and shape of the weld and defect areas based on welding morphology feature data and sets node attributes. At the same time, it locates the position of the weld and defect, assigns a unique identifier to each graph node, maps the weld and defect features to graph node attributes, and generates welding morphology feature analysis results.

[0099] The welding morphology recognition submodule is based on welding morphology feature data and uses graph theory analysis algorithms. It uses the Python programming language to call the NetworkX library to create a graph model, quantitatively analyze the geometric dimensions and morphology of the weld and defect areas, assign a unique identifier to each graph node, and use the set_node_attributes method to map the weld and defect features to the graph node attributes. The node attributes include geometric dimensions and morphological descriptions, and the welding morphology feature analysis results are generated.

[0100] The graph node mapping submodule uses the welding morphology feature analysis results and a mapping algorithm to convert the description of the weld and defect areas into graph nodes. Each graph node is assigned a unique identifier and a corresponding attribute value, and is quantified according to the morphological features of the weld and defects. At the same time, the nodes are connected to each other according to the spatial relationship between the weld and defects to generate graph node mapping results.

[0101] The graph node mapping submodule uses the welding morphology feature analysis results and a data mapping algorithm. It then uses the NetworkX library in Python to process the graph data, converting the descriptions of weld and defect areas into graph nodes. Each node is assigned a unique identifier and corresponding attribute values. The `add_node` method is used to add nodes to the graph, and the morphological features of the weld and defects are quantified. Finally, the `add_edge` method is used to connect the nodes according to the spatial relationship between the weld and defects, generating the graph node mapping result.

[0102] The edge definition and graph construction submodule sorts the nodes and establishes connection paths based on the graph node mapping results and spatial location information. It adjusts the connection strength parameters according to the characteristics of the connection paths and refines the connection paths and strength parameters through iterative optimization to obtain the graph structure of carbon steel pipe welds.

[0103] The edge definition and graph construction submodule, based on the graph node mapping results, employs a spatial relationship analysis algorithm and utilizes the NetworkX library in Python to further manipulate the graph. It sorts the nodes and establishes connection paths, using the sorted function to sort them according to their spatial location information. Based on the characteristics of the connection paths, it adjusts the connection strength parameters using the set_edge_attributes method, and iteratively optimizes the connection paths and strength parameters to refine them, generating the carbon steel pipe weld diagram structure.

[0104] Please see Figure 7The weld quality image convolutional network module includes:

[0105] The structural analysis submodule is based on the carbon steel pipe weld diagram structure. It analyzes each node in the diagram and its connection method, identifies the position and role of the weld or defect area represented by each node in the overall weld structure, analyzes the direct and indirect connection paths between nodes, and generates weld structure depth analysis results.

[0106] The structural analysis submodule is based on the weld seam diagram of carbon steel pipes. It adopts graph theory analysis methods and uses the NetworkX library in Python programming language to analyze the graph structure. It conducts in-depth research on each node and its connection method, extracts each node and its attributes using the graph.nodes(data=True) method, analyzes the direct and indirect connection paths between nodes using the graph.edges method, identifies the position and role of the weld seam or defect area represented by each node in the overall weld seam structure, records the connection information and path characteristics between nodes, and generates in-depth analysis results of the weld seam structure.

[0107] The quality influencing factor identification submodule, based on the weld structure depth analysis results, applies network analysis algorithms to analyze the interaction between nodes and their potential impact on the overall weld quality. If the target node connection pattern is found to be related to quality defects, the node connection pattern is analyzed in detail to identify the key nodes and connection methods that affect weld quality and obtain key quality influencing factors.

[0108] The quality influencing factor identification submodule, based on the weld structure depth analysis results, applies network analysis algorithms and continues to process data using the Python programming language and NetworkX library. It analyzes the interactions between nodes and their potential impact on the overall weld quality, and uses the nx.betweenness_centrality(graph) method to calculate the betweenness centrality of nodes to identify key nodes and connection methods that significantly affect weld quality. It focuses on analyzing node connection patterns related to quality defects and identifies key quality influencing factors.

[0109] Network analysis algorithms use formulas

[0110]

[0111] Improved betweenness centrality of computing nodes, where, For nodes Improved middle centrality For nodes The weight, For nodes and The weights of the paths between them. For nodes Position parameters, For path arrive The length parameter, For the nodes The number of shortest paths, For nodes To the node The total number of shortest paths.

[0112] First, the betweenness centrality of each node is calculated. Then, new weights and parameters are introduced based on the node's position and path characteristics, including the node's... weight This reflects the importance or functionality of the node. and Weight of the path between Considering the importance or accessibility of the path, nodes Position parameters This indicates the geographical or logical location of a node in the network, and the path. arrive Length parameter Taking into account the actual length or cost of the path, all these parameters are combined to adjust the calculation of betweenness centrality, making it more accurately reflect the influence of a node in the network, and the weighting coefficients. and The specific values ​​can be obtained through analysis of node and path attributes. For example, weight values ​​can be determined based on node connectivity, path congestion, or transmission rate, as well as location parameters. This can be determined by analyzing the position of nodes in the network layout. For example, edge nodes and core nodes have different position parameters and length parameters. These improvements, which can be determined based on the actual distance or cost of the path, not only enhance the accuracy of betweenness centrality calculations but also provide more dimensions of consideration for network analysis.

[0113] The quality characteristic integration submodule integrates multiple sets of key quality influencing factors, analyzes the common effects of these factors on weld quality, studies the interaction between key nodes and connection modes, analyzes the impact of these interactions on weld quality, and generates a representation of weld quality characteristics.

[0114] The quality characteristic integration submodule integrates multiple sets of key quality influencing factors, employs statistical analysis methods, and aggregates data on these factors using the Python programming language and Pandas library. It analyzes the combined effects of these factors on weld quality, and uses the `df.groupby(['factor'])['influence'].aggregate('mean').reset_index()` method to study the interactions between key nodes and connection patterns. This analysis examines the impact of these interactions on weld quality and generates a representation of weld quality characteristics.

[0115] Please see Figure 8 The intelligent welding defect identification module includes:

[0116] The feature parsing submodule analyzes the dimensions in the feature vector based on the weld quality feature representation, identifies key features and analyzes their impact on weld quality. Through the feature analysis process, it parses how each dimension in the feature vector reflects the weld quality state, determines the decisive factors affecting weld quality, and generates key weld quality factor data.

[0117] The feature parsing submodule is based on weld quality feature representation. It applies the Principal Component Analysis (PCA) algorithm and executes PCA using the Scikit-learn library in Python. The n_components parameter is set to automatic selection to analyze the dimensions in the feature vector. The PCA.fit_transform method is used to reduce the dimensionality of the feature data, identify and analyze the impact of key features on weld quality, analyze how each dimension in the feature vector reflects the weld quality state, determine the decisive factors affecting weld quality, and generate key weld quality factor data.

[0118] The defect classification submodule is based on the key quality factor data of welds. It uses logical analysis to set the correspondence between the quality characteristics of multiple welds and the differentiated defect types, subdivides the welds, classifies the welds into defect categories according to their quality characteristics, and outputs detailed defect classification information.

[0119] The defect classification submodule is based on the key quality factor data of welds. It adopts the decision tree classification algorithm and uses the Scikit-learn library to execute DecisionTreeClassifier in Python. The criterion parameter is set to "gini" to optimize the classification accuracy. The fit method is applied to the data between weld quality features and differential defect types. Welds are classified into defect categories according to quality features, and detailed defect classification information is output.

[0120] The results output submodule performs in-depth analysis of defects based on detailed defect classification information, identifies the causes of defects and analyzes their formation mechanisms, provides category reports for each defect type, constructs a welding defect database, and generates welding defect identification results.

[0121] The output module, based on refined defect classification information, applies Natural Language Processing (NLP) algorithms and uses the Natural Language Toolkit (NLTK) library in Python to analyze defect description text. It sets appropriate text processing parameters for word segmentation and annotation, identifies defect causes and analyzes formation mechanisms, provides detailed category reports based on the description text of each defect type, constructs a welding defect database, and generates welding defect identification results.

[0122] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A carbon steel pipe welding quality assessment system, characterized in that, The system includes: The 3D scanning module for the welding area uses 3D scanning technology to scan the welding area of ​​carbon steel pipes, capture the structural features of the weld, record geometric data, and generate 3D welding scanning data. The weld dynamic mesh generation module uses the three-dimensional welding scan data to divide the mesh, adjust each mesh unit in real time to match the changes in weld geometry, refine the mesh units and capture detailed information to generate a weld dynamic mesh model. The welding morphology change analysis module, based on the weld dynamic mesh model, monitors the changes in weld width, depth, and edge contour, extracts the changing features, and generates welding morphology feature data. The weld structure diagram construction module maps the weld and defect areas to graph nodes using the welding morphology feature data, defines edges based on the geometric characteristics and positional relationships between nodes, constructs a graph representation of the weld, and generates a carbon steel pipe weld diagram structure. The weld quality graph convolutional network module uses the aforementioned carbon steel pipe weld graph structure to extract features, analyze the node relationships and structural features in the graph representation of the weld, extract weld quality association information, and generate weld quality feature representation. The weld quality map convolutional network module includes: The structural analysis submodule is based on the carbon steel pipe weld diagram structure. It analyzes each node in the diagram and its connection method, identifies the position and role of the weld or defect area represented by each node in the overall weld structure, analyzes the direct and indirect connection paths between nodes, and generates weld structure depth analysis results. The quality influencing factor identification submodule, based on the weld structure depth analysis results, applies network analysis algorithms to analyze the interaction between nodes and their potential impact on the overall weld quality. If the target node connection pattern is found to be related to quality defects, the node connection pattern is analyzed in detail to identify the key nodes and connection methods that affect the weld quality and obtain the key quality influencing factors. The quality characteristic integration submodule integrates multiple sets of key quality influencing factors, analyzes the common effects of these factors on weld quality, studies the interaction between key nodes and connection modes, analyzes the impact factors of this interaction on weld quality, and generates a weld quality characteristic representation. The network analysis algorithm uses the formula ； Improved betweenness centrality of computing nodes, where, For nodes Improved middle centrality For nodes The weight, For nodes and The weights of the paths between them. For nodes Position parameters, For path arrive The length parameter, For the nodes The number of shortest paths, For nodes To the node The total number of shortest paths; The intelligent welding defect identification module identifies and classifies defect types based on the weld quality characteristics, distinguishes different types of welding defects, and generates welding defect identification results.

2. The carbon steel pipe welding quality assessment system according to claim 1, characterized in that, The three-dimensional welding scanning data includes a set of position coordinates, depth values, width values, and a set of surface features. The weld dynamic mesh model includes mesh cell size, weld morphology description set, and cell connection diagram. The welding morphology feature data includes width variation, depth variation, and edge contour morphology set. The carbon steel pipe weld image structure includes a spatial distribution map, a connection property map, and a regional node classification map. The weld quality feature representation includes an overall evaluation index, a set of local defect feature vectors, and quality classification labels. The welding defect identification results include a type identification set, position coordinates, size estimate, and severity level.

3. The carbon steel pipe welding quality assessment system according to claim 1, characterized in that, The weld dynamic mesh generation module includes: The mesh initialization submodule uses the voxelization method based on the three-dimensional welding scan data to determine whether the data covers the entire weld area. If so, it creates an initial mesh frame and automatically divides the weld into multiple basic mesh units according to the geometry of the weld. It sets the initial size and shape parameters for each unit to generate an initial mesh model. The mesh adjustment submodule uses the initial mesh model to monitor the real-time changes in the weld shape. If a change is detected, it dynamically adjusts the size and shape of the mesh cells to match the current geometric features of the weld. By calculating the amount of change in the weld geometry, it adjusts the distribution and density of adjacent mesh cells to obtain the adjusted mesh model. The detail capture submodule iteratively improves the model's ability to capture weld details through the adjusted mesh model. By analyzing the geometric characteristics of the weld area, it identifies key geometric feature areas. If the model needs to capture geometric features, it increases the mesh density in the key areas, performs local refinement processing on the model, and fine-tunes the size of the mesh cells to obtain a dynamic mesh model of the weld.

4. The carbon steel pipe welding quality assessment system according to claim 1, characterized in that, The geometric characteristics of the weld area are analyzed using the following formula: ； Calculate the geometric feature deviation value, combine it with the geometric characteristics of the weld area, calculate the local feature parameters through the key area data, identify the key geometric feature area, if the model needs to capture geometric features, increase the mesh density in the key area, perform local refinement processing on the model, fine-tune the size of the mesh cells, and obtain the dynamic mesh model of the weld. in, Represents the geometric feature deviation value. Represents the weld area Local curvature length at each sampling point This represents the average curvature length of all weld sampling points. Representing the The area of ​​each mesh cell in the weld zone. Representing the The density of mesh cells in the weld area, This represents the total number of mesh cells within the weld area. This represents a summation operation on all grid cells. This represents the absolute deviation between the weld curvature length and the average curvature length.

5. The carbon steel pipe welding quality assessment system according to claim 1, characterized in that, The welding morphology change analysis module includes: The width monitoring submodule analyzes the distribution characteristics of grid cells along the weld length based on the weld dynamic grid model, and records the width data at each position by comparing the distance changes of adjacent grid cells in the lateral direction, thereby generating weld width change data. The depth monitoring submodule uses the weld width variation data to analyze the changes in grid cells along the weld depth. By measuring the height difference of differentiated depth grid cells, it records the distance from the surface of the weld to the deepest point, extracts the depth variation trend along the weld length, and obtains weld depth variation data. The edge contour monitoring submodule identifies changes in the weld edge contour through the weld depth change data, analyzes the outer lines of the mesh model, compares the changes in edge contour lines at different time points, records the undulations and offsets of the edge contour, tracks the trajectory of edge contour changes, and obtains weld edge contour data. The feature extraction submodule integrates the width, depth and edge contour variation information based on the weld edge contour data, analyzes the key variation features of the weld morphology, and generates welding morphology feature data by comparing and analyzing the variation rules and patterns of the width, depth and edge contour data.

6. The carbon steel pipe welding quality assessment system according to claim 1, characterized in that, The weld structure diagram construction module includes: The welding morphology recognition submodule analyzes the geometric dimensions and morphology of the weld and defect areas based on the welding morphology feature data and sets node attributes. At the same time, it locates the position of the weld and defect, assigns a unique identifier to each graph node, maps the weld and defect features to the graph node attributes, and generates welding morphology feature analysis results. The graph node mapping submodule uses the welding morphology feature analysis results and a mapping algorithm to convert the description of the weld and defect areas into graph nodes. Each graph node is assigned a unique identifier and a corresponding attribute value, and is quantified according to the morphological features of the weld and defects. At the same time, the nodes are connected to each other according to the spatial relationship between the weld and defects to generate graph node mapping results. The edge definition and graph construction submodule sorts the nodes and establishes connection paths based on the graph node mapping results using spatial location information. It adjusts the connection strength parameters according to the characteristics of the connection paths and refines the connection paths and strength parameters through iterative optimization to obtain the carbon steel pipe weld diagram structure.

Images