CN-121544906-B - Visual identification and positioning system for welding spot edge of condenser tube

Abstract

The invention relates to the technical field of detection of welding quality of a condenser pipe, and discloses a vision recognition and positioning system for the welding spot edge of the condenser pipe. The image acquisition module acquires a multi-angle gray level image and sets a scanning interval, the edge feature module divides outline nodes and analyzes gradient features to generate vectors, the association analysis module extracts key indexes, establishes a positioning rule and acquires correction parameters, the form optimization module identifies a space association mode, calibrates outline indexes and calculates offset, the reference positioning module derives an optimal positioning threshold and generates a deviation sequence, and the decision output module integrates the deviation sequence to form a positioning scheme. The system realizes high-precision identification and positioning of the welding spot edge characteristics through cooperation of multiple modules, improves the automation level and precision of the welding quality detection of the condenser tube, and is suitable for welding quality detection scenes of automatic production lines.

Inventors

• LV YONG
• KONG QINGJI
• WANG XIN
• YU QIAOLING

Assignees

• 辽宁西部动力科技有限公司

Dates

Publication Date

20260512

Application Date

20250729

Claims (5)

1. 1. A system for visual identification and location of condenser tube weld edges, comprising: the image acquisition module is used for acquiring gray image data of the welding line area of the multi-angle condenser tube and setting a contour scanning interval corresponding to the welding point positioning; The edge feature module is used for dividing a plurality of contour nodes in the contour scanning interval, carrying out gradient feature analysis on the image data of each contour node, and generating edge feature vectors corresponding to the contour nodes; the association analysis module is used for extracting key contour indexes from the edge feature vectors, establishing a positioning judgment rule associated with the contour nodes, defining the condition that the nodes belong to the welding spot edges, and acquiring geometric correction parameters corresponding to the rule; The form optimization module is used for identifying a space association mode in the geometric correction parameters, wherein the space association mode represents the distribution rule and the mutual influence relation of the parameters in space, dynamically calibrates key contour indexes according to the association mode, and calculates contour offset of each contour node under different calibration strategies; The reference positioning module is used for deducing an optimal positioning threshold value according to the profile offset and generating a profile deviation sequence by comparing the current profile characteristic value with the optimal positioning threshold value; the decision output module is used for analyzing the contour deviation sequence, integrating the contour deviation sequence into a welding spot positioning decision scheme based on the spatial deviation trend of the contour node, constructing a welding spot feature library corresponding to the contour node, wherein the welding spot feature library comprises geometric parameter vectors mapped by image data and gradient features; The geometric parameter vector is subjected to similar contour matching, and a contour cluster group of the geometric parameter vector is divided according to a matching result; extracting curvature distribution, contour polarity and morphological density parameters according to the spatial attribute and gradient strength in the geometric parameter vector, and generating an edge feature label based on the parameters; correlating the edge feature labels with the geometric parameter vectors, and screening parameter vectors with similarity higher than a preset contour threshold value to form a contour cluster group by calculating the spatial similarity among the feature labels; Separating main contour duty ratio, secondary contour duty ratio and morphological fluctuation parameters from the edge feature vector, and generating a positioning judgment rule of contour nodes based on the parameters; If the number of the contour nodes covered by the current positioning judgment rule is smaller than a preset positioning threshold, traversing edge feature vectors of adjacent contour nodes, and adding geometric indexes which are not contained in the judgment rule of the adjacent nodes into the current rule; extracting a morphological mode closest to the current contour offset in a standard welding spot database, and calculating cosine similarity of the morphological mode and the current contour offset on spatial distribution to be used as a first dynamic reference value; counting the difference between the current profile offset and the number of profile peaks of the standard form mode, and taking the number of the difference as a second dynamic reference value; Based on the linear combination of the first dynamic reference value and the second dynamic reference value, matching an optimal positioning threshold value in a preset positioning threshold value table; Dividing a main offset interval and a secondary offset interval according to the space offset direction of each node in the profile offset sequence, wherein the main offset interval corresponds to the main offset trend of the welding spot edge, and the secondary offset interval reflects a local abnormality or transition region; And extracting the convergence rate of the contour deviation in the main offset interval and the diffusion rate of the contour deviation in the secondary offset interval, and carrying out fusion calculation on the convergence rate and the diffusion rate according to the spatial weight of the contour node to generate the adjustment parameters of the welding spot positioning decision scheme.
2. 2. The system for visual identification and localization of condenser tube solder joint edges of claim 1, wherein the implementation of generating edge feature vectors corresponding to contour nodes comprises: aiming at each contour node, acquiring gradient fluctuation data of the node in a preset window according to the spatial position of the node in a contour scanning interval, and calculating a gradient fluctuation coefficient of the node; When the gradient fluctuation coefficient is lower than the first gradient threshold value, marking the node as a steady-state node, carrying out gradient superposition on the image data of the adjacent nodes of the node, and reconstructing the superposed data into an edge feature vector.
3. 3. The visual recognition and positioning system for the welding spot edge of the condensing tube according to claim 1, wherein the form optimization module is realized by acquiring a space factor of form switching frequency and a fluctuation factor of gradient jump amplitude in a space correlation mode; and constructing a focusing weight matrix associated with the space factor and the fluctuation factor, and determining contour offset under different calibration strategies according to the distribution probability of each element in the matrix.
4. 4. The system for visual identification and localization of condenser tube solder joint edges of claim 3, wherein constructing a focus weight matrix further comprises: identifying morphological periodic characteristics of the space factors, and setting the space factors as initial indexes of a focusing weight matrix if the current periodic characteristics are completely matched with a preset contour period; calculating the association matching degree of the space factor and the fluctuation factor, and sequentially generating an intermediate index and a termination index of the focusing weight matrix according to the matching degree from high to low; And carrying out path backtracking on the termination index, and outputting the termination index as the final distribution of the focusing weight matrix when the matching degree of the termination index is lower than a preset matching threshold value.
5. 5. The system for visual identification and localization of condenser tube solder joint edges of claim 4, wherein the calculation of the profile offset for each profile node under different calibration strategies comprises: counting the spatial factor mean value and fluctuation factor range of each termination index in the focusing weight matrix, and calculating the global variance of all index factors; And simultaneously calculating the ratio of the fluctuation factor range to the global variance, and taking the weighted sum of the fluctuation factor range and the global variance as the profile offset of the index.

Description

Visual identification and positioning system for welding spot edge of condenser tube Technical Field The invention relates to the technical field of detection of welding quality of a condenser pipe, in particular to a visual identification and positioning system for the edge of a welding spot of the condenser pipe. Background In the field of refrigeration equipment manufacture, a condenser tube is used as a key heat exchange component, and the welding quality of the condenser tube directly influences the performance and reliability of equipment. The accurate identification and positioning of the welding spot edge are the core links for evaluating welding quality, and the traditional manual detection mode has the problems of low efficiency, strong subjectivity, easiness in interference by human factors and the like, and is difficult to meet the detection requirements of high precision and high efficiency on an automatic production line. With the development of industrial automation and machine vision technologies, image recognition-based detection methods are becoming mainstream, but the prior art still faces many challenges when dealing with complex morphological features of the welding spot edge of a condenser tube. On the one hand, the geometric forms of the welding spot edges of the condensing tube have various characteristics, such as curvature change, contour polarity difference, non-uniform form density and the like, of the welding spot edges caused by different welding process parameters, and the traditional fixed-threshold edge detection algorithm is difficult to adapt to the dynamic identification requirement of the multi-form characteristics. For example, detection methods based on a single gradient threshold are prone to edge miss or false detection problems when faced with solder joint edges where the curvature change is large. On the other hand, when the existing visual positioning system processes multi-angle image data, effective analysis on a contour node space association mode is lacking, and accurate positioning judgment rules are difficult to establish, so that positioning accuracy is insufficient. For example, in the process of multi-view image stitching, due to lack of dynamic calibration of the spatial offset trend of the contour nodes, accumulated errors are easily generated, and accuracy of the final positioning result is affected. In addition, when gradient fluctuation data is processed, the dynamic characteristics of contour nodes cannot be fully considered in the existing system, and the differential processing mechanism of strong contour nodes and steady-state nodes is not perfect enough, so that the characterization capability of edge feature vectors is insufficient. For example, for a steady state node, simply ignoring gradient overlay information for its neighbors may result in loss of edge features, affecting the accuracy of subsequent correlation analysis. Meanwhile, in the derivation process of the positioning threshold, the existing method is mostly based on fixed standard template matching, and is lack of self-adaptive adjustment of the dynamic characteristics of the current profile offset, so that the optimal positioning threshold is difficult to quickly match in a complex welding environment. The existing detection technology for the welding spot edge of the condenser tube has obvious defects in the aspects of complex morphological feature processing, multi-view data association analysis, dynamic feature self-adaptive adjustment and the like, and a visual identification and positioning system capable of realizing multi-angle image data fusion, dynamic calibration of profile features and accurate deduction of a positioning threshold value is needed to meet the actual requirements of efficient detection of the welding quality of the condenser tube on an automatic production line. Disclosure of Invention The invention aims to provide a visual recognition and positioning system for the welding spot edge of a condensing tube, so as to solve the problems in the background art. In order to achieve the purpose, the invention provides the following technical scheme that the system for visually identifying and positioning the welding spot edge of the condensing tube comprises: the image acquisition module is used for acquiring gray image data of the welding line area of the multi-angle condenser tube and setting a contour scanning interval corresponding to the welding point positioning; The edge feature module is used for dividing a plurality of contour nodes in the contour scanning interval, carrying out gradient feature analysis on the image data of each contour node, and generating edge feature vectors corresponding to the contour nodes; the association analysis module is used for extracting key contour indexes from the edge feature vectors, establishing a positioning judgment rule associated with the contour nodes and acquiring geometric correction parameters corresponding to