An intelligent welding system based on machine vision

Abstract

The application relates to the field of visual monitoring, in particular to an intelligent welding system based on machine vision, which comprises a production analysis module used for determining a production line monitoring strategy; a node monitoring module used for determining an interference analysis mode in response to an analysis setting condition; an interference analysis module used for determining a node adjustment strategy based on an abnormal reflection parameter and an abnormal reflection distribution index in response to a reflection analysis condition, and determining whether to adjust a light source distribution index based on the abnormal reflection parameter and an edge abnormal parameter in response to a quality analysis condition; a monitoring evaluation module used for determining whether to update a monitoring parameter for a monitoring execution node in response to a difference analysis condition; and a difference compensation module connected with the monitoring evaluation module and used for determining a difference compensation mode corresponding to each difference analysis node in response to a compensation setting condition. The application improves the accuracy of quality inspection results of target welding pieces.

Description

Technical Field

[0001] This invention relates to the field of visual monitoring, and more particularly to an intelligent welding system based on machine vision. Background Technology

[0002] Real-time inspection of the quality of welded products using machine vision technology can effectively ensure the quality of the welding production process and provide timely feedback. In actual production, to meet production demands, it is often necessary to produce multiple specifications or categories of products simultaneously, which can easily lead to mixing of different specifications or categories during quality inspection. However, existing machine vision-based welding systems often fail to effectively eliminate the interference caused by product mixing in the visual analysis process of quality inspection. Therefore, how to minimize the interference of product mixing in the visual analysis process of quality inspection to ensure the accuracy of quality inspection results in actual welding production is a problem that urgently needs to be solved by those skilled in the art.

[0003] Chinese patent application publication number CN117314829A discloses a computer vision-based method and system for quality inspection of industrial parts. The computer vision-based method includes: acquiring image data of industrial parts using a camera, preprocessing the image data to obtain preprocessed image data; extracting features from the preprocessed image data to obtain feature information of the industrial parts; analyzing the feature information using a convolutional neural network model to obtain defect detection results of the industrial parts; and generating an industrial parts quality inspection report based on the defect detection results. However, the above solution has the following problems: it fails to consider the interference caused by the mixing of different batches of products during the production process on visual analysis, resulting in low accuracy of the quality inspection results and consequently, low feedback quality on the production process. Summary of the Invention

[0004] To address this issue, the present invention provides an intelligent welding system based on machine vision, which overcomes the problem in the prior art that fails to consider the interference caused by the mixing of different batches of products during the production process on visual analysis, resulting in low accuracy of product quality inspection results and consequently low feedback quality of the welding process.

[0005] To achieve the above objectives, the present invention provides an intelligent welding system based on machine vision, comprising:

[0006] The production analysis module is used to periodically respond to production assessment conditions in order to determine production line monitoring strategies and to perform monitoring node analysis or monitoring quality analysis for the target production line.

[0007] The node monitoring module, which is connected to the production analysis module, is used to determine the category of each monitoring execution node in response to node status assessment conditions, and to determine the interference analysis method as either acquisition reflection analysis or acquisition quality analysis for node monitoring images in response to analysis setting conditions.

[0008] An interference analysis module, connected to the node monitoring module, is used to respond to reflection analysis conditions, determine node adjustment strategies based on abnormal reflection parameters and abnormal reflection distribution index, and respond to quality analysis conditions, determine whether to adjust the light source distribution index of the second-class monitoring execution node based on abnormal reflection parameters and edge anomaly parameters.

[0009] The monitoring and evaluation module, which is connected to the production analysis module, is used to identify the difference analysis nodes of the target monitored production line and, in response to the difference analysis conditions, to determine whether to update the monitoring parameters for each monitoring execution node.

[0010] The difference compensation module, which is connected to the monitoring and evaluation module, is used to determine the difference compensation method for each difference analysis node in response to the compensation setting conditions. This method is either to update the monitoring parameters of the difference analysis node based on the difference coefficient, or to adjust the light source richness parameters of the difference analysis node based on the image confidence stability coefficient.

[0011] Furthermore, if the production evaluation condition responded by the production analysis module is that the production alternation frequency is greater than the preset production alternation frequency or the production synchronization coefficient is greater than the preset production synchronization coefficient, then the node monitoring module will perform monitoring node analysis on the target supervised production line, including:

[0012] The node monitoring module responds to node analysis conditions and determines the monitoring interference status of each monitoring execution node within the target supervised production line based on the node association category coefficient and category interference index.

[0013] The node analysis method for each monitoring execution node is determined based on the monitoring interference status;

[0014] The node analysis conditions are determined by the production analysis module to enable the node monitoring module to perform monitoring node analysis on the target supervised production line.

[0015] Furthermore, the node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is greater than the preset node association category coefficient or the category interference index is greater than the preset category interference index, then the monitored execution node is determined to be a type of monitored execution node.

[0016] The node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is less than or equal to the preset node association category coefficient and the category interference index is less than or equal to the preset category interference index, then the monitored execution node is determined to be a Class II monitored execution node.

[0017] Furthermore, the analysis setting condition for the node monitoring module response is the existence of a type of monitoring execution node. The determination interference analysis module performs reflection analysis on the node monitoring images of each type of monitoring execution node, including:

[0018] The interference analysis module responds to the reflection analysis conditions and determines the abnormal reflection points based on the reflection difference coefficient corresponding to the point analysis range.

[0019] The interferometry analysis module determines the node adjustment strategy based on the abnormal reflection parameters and the abnormal reflection distribution index.

[0020] The reflection analysis condition is determined by the node monitoring module to be the result of the interference analysis module collecting and analyzing the reflection of the node monitoring images of each type of monitoring execution node.

[0021] Furthermore, the interference distribution condition responded by the interference analysis module is that there exists a type of monitoring execution node whose node monitoring image has an abnormal reflection parameter greater than a preset abnormal reflection parameter or an abnormal reflection distribution index greater than a preset abnormal reflection distribution index. Then, based on the abnormal reflection distribution index, the acquisition execution angle of this type of monitoring execution node is increased and adjusted.

[0022] The increase in the acquisition angle is positively correlated with the abnormal reflection distribution index.

[0023] Furthermore, the interference distribution condition of the interference analysis module response is that there exists a type of monitoring execution node whose abnormal reflection parameter of the node monitoring image is less than or equal to a preset abnormal reflection parameter and whose abnormal reflection distribution index is less than or equal to a preset abnormal reflection distribution index. Then, based on the reference reflection difference coefficient, the light source richness parameter of the downstream monitoring node of this type of monitoring execution node is increased and adjusted.

[0024] The increase in the richness parameters of the light source is positively correlated with the reference reflection difference coefficient.

[0025] Furthermore, if the node monitoring module's response is set to the presence of two types of monitoring execution nodes, then the interference analysis module will perform a quality analysis of the node monitoring images for each type of monitoring execution node, including:

[0026] The interferometric analysis module responds to quality analysis conditions and determines the acquisition quality coefficient of the monitoring image of each node based on abnormal reflection parameters and edge abnormal parameters.

[0027] The change setting condition for the response of the interference analysis module is that if the acquisition quality coefficient of the monitoring image of a node is less than the preset acquisition quality coefficient, then the light source distribution index of the second type of monitoring execution node corresponding to the monitoring image of that node is increased and adjusted according to the acquisition quality coefficient and the target reflection index.

[0028] The quality analysis condition node monitoring module determines that the interference analysis module performs quality analysis on the node monitoring images of each type II monitoring execution node.

[0029] Furthermore, if the production evaluation condition responded by the production analysis module is that the production alternation frequency is less than or equal to the preset production alternation frequency and the production synchronization coefficient is less than or equal to the preset production synchronization coefficient, then the monitoring and evaluation module will perform monitoring quality analysis on the target supervised production line, including:

[0030] The monitoring and evaluation module responds to monitoring and analysis conditions and determines the regulatory difference parameters for each monitoring execution node based on the node quality inspection coefficient and the production continuity index.

[0031] Determine whether to update monitoring parameters for each monitoring execution node based on regulatory difference parameters;

[0032] The monitoring and analysis conditions are determined by the production analysis module to enable the monitoring and evaluation module to perform monitoring quality analysis on the target supervised production line.

[0033] Furthermore, if the monitoring and evaluation module responds to the difference analysis condition that there is a regulatory execution node with a regulatory difference parameter greater than the preset regulatory difference parameter, then the difference compensation module will adjust the monitoring parameters of the difference analysis node and determine the difference compensation method for each difference analysis node based on the difference concentration index.

[0034] Regulatory execution nodes whose regulatory difference parameters are greater than the preset regulatory difference parameters are recorded as difference analysis nodes.

[0035] Furthermore, the compensation setting condition for the difference compensation module response is that if the difference concentration index of a difference analysis node is greater than the preset difference concentration index, then the monitoring parameters for that difference analysis node are updated according to the execution difference coefficient.

[0036] The compensation setting condition for the difference compensation module is that if the difference concentration index of a difference analysis node is less than or equal to a preset difference concentration index, then the light source richness parameter of that difference analysis node is adjusted according to the image confidence stability coefficient.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows: The technical solution of the present invention determines the production line monitoring strategy of the target supervision production line based on the production alternation frequency and the production synchronization coefficient. By adjusting the production line monitoring strategy of the target supervision production line in a timely manner, the monitoring method of the target supervision production line is more in line with the actual working scenario, and the mixed risk of target welded parts in the target supervision production line is avoided from interfering with visual analysis. The present invention improves the accuracy of the quality inspection results of the target welded parts.

[0038] Furthermore, in this invention, when the production alternation frequency of the target monitored production line is greater than the preset production alternation frequency or the production synchronization coefficient is greater than the preset production synchronization coefficient, it indicates that there is a significant product mixing risk within the production line evaluation period during the analysis of the target monitored production line. To avoid the impact of product mixing of different categories on the quality inspection process, the monitoring interference status of each monitoring execution node is determined based on the node association category coefficient and category interference index. Based on the monitoring interference status, targeted analysis is performed on each monitoring execution node to ensure that the node analysis method for each monitoring execution node is more consistent with the actual node status, thereby improving the execution efficiency of the quality inspection process and the reliability of the quality inspection results.

[0039] Furthermore, this invention determines the monitoring interference status of each monitoring execution node based on the node association category coefficient and the category interference index. This characterizes the degree of interference caused by product mixing during the product quality inspection process completed by the monitoring execution node through image analysis within the production line evaluation cycle during the analysis. This provides a basis for determining the node analysis method for each monitoring execution node, ensuring that the analysis process for each monitoring execution node is more in line with actual needs. This invention improves the accuracy of quality inspection results obtained based on image analysis while ensuring the execution efficiency of the quality inspection process.

[0040] Furthermore, this invention addresses monitoring execution nodes in a state of monitoring interference. These nodes experience a high degree of interference during product quality inspection analysis. It is necessary to further analyze the mixed situation of the target welded parts under the quality inspection of these monitoring execution nodes and adjust the monitoring parameters accordingly. This invention improves the execution efficiency of the quality inspection process based on image analysis and the reliability of the quality inspection results.

[0041] Furthermore, in this invention, when the production alternation frequency of the target monitored production line is less than or equal to the preset production alternation frequency and the production synchronization coefficient is less than or equal to the preset production synchronization coefficient, it indicates that the product mixing risk of the target monitored production line within the production line evaluation cycle being analyzed is low. By analyzing the differences between the quality inspection results of the monitoring execution nodes, it is determined whether to adjust the monitoring parameters of the monitoring execution nodes to ensure the quality of the product images obtained by each monitoring execution node for product quality inspection analysis, thereby improving the accuracy of the quality inspection results for the target welded parts. Attached Figure Description

[0042] Figure 1 This is a module connection diagram of the intelligent welding system based on machine vision according to the present invention;

[0043] Figure 2 This is a flowchart of the production analysis module of the present invention responding to production evaluation conditions to determine the production line monitoring strategy;

[0044] Figure 3 This is a flowchart of the node monitoring module of the present invention responding to analysis setting conditions to determine the interferometric analysis method;

[0045] Figure 4 This is a flowchart of the difference compensation module of the present invention responding to compensation setting conditions to determine the difference compensation method. Detailed Implementation

[0046] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0047] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0048] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0049] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0050] Please see Figures 1 to 4 As shown, the present invention provides an intelligent welding system based on machine vision, comprising:

[0051] The production analysis module is used to periodically respond to production assessment conditions in order to determine production line monitoring strategies and to perform monitoring node analysis or monitoring quality analysis for the target production line.

[0052] The node monitoring module, which is connected to the production analysis module, is used to determine the category of each monitoring execution node in response to node status assessment conditions, and to determine the interference analysis method as either acquisition reflection analysis or acquisition quality analysis for node monitoring images in response to analysis setting conditions.

[0053] An interference analysis module, connected to the node monitoring module, is used to respond to reflection analysis conditions, determine node adjustment strategies based on abnormal reflection parameters and abnormal reflection distribution index, and respond to quality analysis conditions, determine whether to adjust the light source distribution index of the second-class monitoring execution node based on abnormal reflection parameters and edge anomaly parameters.

[0054] The monitoring and evaluation module, which is connected to the production analysis module, is used to identify the difference analysis nodes of the target monitored production line and, in response to the difference analysis conditions, to determine whether to update the monitoring parameters for the monitoring execution nodes.

[0055] The difference compensation module, which is connected to the monitoring and evaluation module, is used to determine the difference compensation method for each difference analysis node in response to the compensation setting conditions. This method is either to update the monitoring parameters of the difference analysis node based on the difference coefficient, or to adjust the light source richness parameters of the difference analysis node based on the image confidence stability coefficient.

[0056] This invention is used to analyze the welding quality of multiple product categories. The production line undergoing welding is designated as the target monitoring production line. This target monitoring production line has several monitoring execution nodes to acquire product images during the production process of the target welded parts. Monitoring parameters are those that affect the imaging effect of the product images. These parameters include, but are not limited to, resolution, acquisition angle, light source richness parameters, light source distribution index, depth of field, aperture, and focal length. Image analysis determines whether defects exist in the target welded parts during the welding process. Each monitoring execution node corresponds to a monitored production area and a welding processing device. For a single monitoring execution node, the monitored production area is the range that the node can acquire within the target monitoring production line. The welding processing device is the device corresponding to the most recent welding processing step completed by the target monitored product category acquired by the monitoring execution node. This invention does not specify the location of each monitoring execution node or the monitoring device; users can adaptively set them according to the actual working scenario. In this invention, the target monitoring production line needs to complete the welding processing of multiple target welded parts.

[0057] This invention applies a cyclic production line evaluation cycle. The duration of the production line evaluation cycle can be determined by the user. The higher the user's requirements for the accuracy of the quality inspection results of the target welded parts, the shorter the duration of the production line evaluation cycle. A production line evaluation cycle of 24 hours is provided. At the end of each production line evaluation cycle, the production alternation frequency and production synchronization coefficient of the target supervised production line are obtained. The production line monitoring strategy of the target supervised production line is determined based on the production alternation frequency and production synchronization coefficient.

[0058] This invention utilizes several production line monitoring records. Each production line monitoring record records at least one node-related category coefficient, category interference index, reflection difference coefficient, abnormal reflection parameter, abnormal reflection distribution index, collection quality coefficient, production continuity index, monitoring difference parameter, difference concentration index, and execution difference coefficient during the monitoring and management of the target production line. Each production line monitoring record also has a corresponding qualification mark. The qualification mark records whether the accuracy of the quality inspection results for the target welded parts meets the user's requirements. It can be understood that the user can determine whether the accuracy of the quality inspection results for the target welded parts meets the requirements based on self-defined indicators. For example, self-defined indicators can be, but are not limited to, warning confidence level. Warning confidence level = the number of correct welding defect judgments for the target welded parts / the number of welding defect judgments for the target welded parts.

[0059] Specifically, if the production evaluation condition responded by the production analysis module is that the production alternation frequency is greater than the preset production alternation frequency or the production synchronization coefficient is greater than the preset production synchronization coefficient, then the node monitoring module will perform monitoring node analysis on the target supervised production line, including:

[0060] The node monitoring module responds to node analysis conditions and determines the monitoring interference status of each monitoring execution node within the target supervised production line based on the node association category coefficient and category interference index.

[0061] The node analysis method for each monitoring execution node is determined based on the monitoring interference status;

[0062] The node analysis conditions are determined by the production analysis module to enable the node monitoring module to perform monitoring node analysis on the target supervised production line.

[0063] Wherein, the production alternation frequency is the average number of alternations of the target welded parts processed by each product processing device within the current production line evaluation cycle, and the production synchronization coefficient is the number of target welded parts that need to be processed within the production line evaluation cycle. The values ​​of the preset production alternation frequency and the preset production synchronization coefficient can be determined by the user according to the actual working scenario. For example, the user can set them according to the production line supervision records. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset production alternation frequency and the smaller the value of the preset production synchronization coefficient. One preset production alternation frequency value is provided, which is recorded as the evaluation reference record for the production line monitoring node analysis for the target supervised production line. The average production alternation frequency of the target supervised production line in the evaluation reference record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts is recorded as the preset production alternation frequency. Another preset production synchronization coefficient value is provided, which is recorded as the average production synchronization coefficient of the target supervised production line in the evaluation reference record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts.

[0064] Specifically, the node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is greater than the preset node association category coefficient or the category interference index is greater than the preset category interference index, then the monitored execution node is determined to be a type of monitored execution node.

[0065] The node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is less than or equal to the preset node association category coefficient and the category interference index is less than or equal to the preset category interference index, then the monitored execution node is determined to be a Class II monitored execution node.

[0066] Specifically, if a monitoring execution node has a node association category coefficient greater than the preset node association category coefficient or a category interference index greater than the preset category interference index, then the monitoring execution node with the node association category coefficient greater than the preset node association category coefficient or a category interference index greater than the preset category interference index is determined to be a Class I monitoring execution node. If a monitoring execution node has a node association category coefficient less than or equal to the preset node association category coefficient and a category interference index less than or equal to the preset category interference index, then the monitoring execution node with the node association category coefficient less than or equal to the preset node association category coefficient and a category interference index less than or equal to the preset category interference index is determined to be a Class II monitoring execution node.

[0067] For a single monitoring execution node, the node-related category coefficient is the number of categories of related monitoring products of the monitoring execution node within the current production line evaluation cycle. For a single target welded part, if the target welded part needs to be processed by each product processing device corresponding to the monitoring execution node, then the target welded part is recorded as the related monitoring product of the monitoring execution node. The category interference index is determined based on the material complexity parameter and quality inspection conflict coefficient of the related monitoring product of the monitoring execution node. Category interference index = ln(material complexity parameter × quality inspection conflict coefficient). The material complexity parameter is the number of different material types of each related monitoring product corresponding to the monitoring execution node. The quality inspection conflict coefficient is the absolute value of the difference between the maximum and minimum values ​​of the gloss of each target welded part monitored within the current production line supervision cycle when the welding processing stage is at the monitoring execution node.

[0068] The values ​​of the preset node association category coefficient and the preset category interference index can be determined by the user according to the actual working scenario. For example, the user can set them according to the production line monitoring records. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset node association category coefficient and the smaller the value of the preset category interference index. A method for determining the value of the preset node association category coefficient is provided, which records the maximum value of the node association category coefficient of the second-class monitoring execution node in the production line monitoring records that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts as the preset node association category coefficient. A method for determining the value of the preset category interference index is provided, which records the maximum value of the category interference index of the second-class monitoring execution node in the production line monitoring records that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts as the preset category interference index.

[0069] Specifically, the analysis setting condition for the node monitoring module response is the existence of a type of monitoring execution node. The determination interference analysis module performs reflection analysis on the node monitoring images of each type of monitoring execution node, including:

[0070] The interference analysis module responds to the reflection analysis conditions and determines the abnormal reflection points based on the reflection difference coefficient corresponding to the point analysis range.

[0071] The interferometric analysis module determines the analysis method of the monitoring images of each node based on the abnormal reflection parameters and the abnormal reflection distribution index.

[0072] The reflection analysis condition is that the node monitoring module determines that the interference analysis module collects and analyzes the reflection of the node monitoring images of each type of monitoring execution node.

[0073] The node monitoring images are images collected by each monitoring execution node for the target welded part. By performing image analysis on the node monitoring images, it is determined whether there are abnormal products in the production process. Abnormal products are target welded parts with production quality defects. The production quality defects in this invention include, but are not limited to: cracks and scratches, oxide spots, shrinkage marks, porosity and deformation defects.

[0074] For any pixel within a single node's monitoring image, if the reflection difference coefficient of that pixel within the point analysis range is greater than a preset reflection difference coefficient, then that similar point is recorded as an abnormal reflection point. The point analysis range is a circular area centered on that pixel and with a preset analysis distance as its radius. The reflection difference coefficient... m is the number of pixels included in the position analysis range of this pixel, ft is the brightness value of the t-th pixel other than this pixel in the position analysis range of this pixel, and f0 is the brightness value of this pixel.

[0075] The user can determine the values ​​of the preset reflection difference coefficient and the preset analysis distance according to the actual working scenario. For example, the user can set them according to the production line monitoring record. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset reflection difference coefficient and the larger the value of the preset analysis distance. A method for determining the value of the preset reflection difference coefficient is provided, which is the average value of the reflection difference coefficient of abnormal reflection points in the production line monitoring record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts. A method for determining the value of the preset analysis distance is provided, which is 3% of the short side dimension of the acquired node monitoring image.

[0076] For a node monitoring image where the abnormal reflection points have been determined, the abnormal reflection parameter is the number of abnormal reflection points in the node monitoring image. The node monitoring image is divided into several rectangular areas of equal area and denoted as monitoring sub-images. The abnormal reflection distribution index is the number of monitoring sub-images containing abnormal reflection points in the node monitoring image.

[0077] Specifically, the interference distribution condition of the interference analysis module is that there is a type of monitoring execution node whose node monitoring image has an abnormal reflection parameter greater than a preset abnormal reflection parameter or an abnormal reflection distribution index greater than a preset abnormal reflection index. Then, based on the abnormal reflection distribution index, the acquisition execution angle of the type of monitoring execution node is increased and adjusted.

[0078] The increase in the acquisition execution angle is positively correlated with the abnormal reflection distribution index.

[0079] Specifically, the interference distribution condition of the interference analysis module is that there is a type of monitoring execution node whose abnormal reflection parameter is less than or equal to a preset abnormal reflection parameter and whose abnormal reflection distribution index is less than or equal to a preset abnormal reflection distribution index. Then, based on the reference reflection difference coefficient, the light source richness parameter of the downstream monitoring node of this type of monitoring execution node is increased and adjusted.

[0080] The increase in the richness parameters of the light source is positively correlated with the reference reflection difference coefficient.

[0081] The values ​​of the preset abnormal reflection parameters and the preset abnormal reflection distribution index can be determined by the user according to the actual working scenario. For example, the user can set them according to the production line monitoring records. A method for determining the value of the preset abnormal reflection parameters is provided, in which the production line monitoring records that increase the acquisition execution angle of a certain type of monitoring execution node based on the abnormal reflection distribution index are recorded as adjustment reference records, and the average value of the abnormal reflection parameters in the adjustment reference records that meet the user's requirements for the accuracy of the quality inspection results of the target welded parts is recorded as the preset abnormal reflection parameters. A method for determining the value of the preset abnormal reflection distribution index is provided, in which the average value of the abnormal reflection distribution index in the adjustment reference records that meet the user's requirements for the accuracy of the quality inspection results of the target welded parts is recorded as the preset abnormal reflection distribution index.

[0082] If the abnormal reflection parameter of the monitoring image of a certain type of monitoring execution node is greater than the preset abnormal reflection parameter or the abnormal reflection distribution index is greater than the preset abnormal reflection distribution index, it indicates that there are many abnormal reflection locations. At this time, the abnormality is caused by the target welded part with a mixture of many other materials or simply improper monitoring parameter settings. By avoiding direct light, the abnormal reflection phenomenon in the image can be reduced. The acquisition execution angle is the maximum angle value of the acute angle formed by the light emitted by the set light source and the plane where the target monitoring product is located.

[0083] If the abnormal reflection parameter of the node monitoring image of a certain type of monitoring execution node is less than or equal to the preset abnormal reflection parameter and the abnormal reflection distribution index is less than or equal to the preset abnormal reflection distribution index, the cause of the abnormality is that there is a target welded part mixed with a small amount of other materials. By improving the lighting conditions of the downstream monitoring node of this type of monitoring execution node, a higher quality node monitoring image can be obtained, thereby improving the accuracy of the quality inspection results for this batch of target welded parts. The downstream monitoring node is the monitoring execution node that the target welded part passes through after passing through this type of monitoring execution node. The reference reflection difference coefficient is the average value of the reflection difference coefficients of each abnormal reflection point in the node monitoring image of this type of monitoring execution node. The light source richness parameter is the number of different light emission directions of the light source when acquiring node monitoring images.

[0084] Specifically, the analysis setting condition for the node monitoring module response is that there are two types of monitoring execution nodes. Then, the interference analysis module performs a quality analysis of the node monitoring images for each type of monitoring execution node, including:

[0085] The interferometric analysis module responds to quality analysis conditions and determines the acquisition quality coefficient of the monitoring image of each node based on abnormal reflection parameters and edge abnormal parameters.

[0086] The change setting condition for the response of the interference analysis module is that if the acquisition quality coefficient of the monitoring image of a node is less than the preset acquisition quality coefficient, then the light source distribution index of the second type of monitoring execution node corresponding to the monitoring image of that node is increased and adjusted according to the acquisition quality coefficient and the target reflection index.

[0087] The quality analysis condition node monitoring module determines that the interference analysis module performs quality analysis on the node monitoring images of each type II monitoring execution node.

[0088] Specifically, for a single node monitoring image, the acquisition quality coefficient is determined based on abnormal reflection parameters and edge anomaly parameters. The acquisition quality coefficient is the product of the abnormal reflection parameters and the edge anomaly parameters. The edge anomaly parameters are the number of abnormal edge regions determined after edge recognition of the node monitoring image. Edge recognition is performed on the node monitoring image to segment and obtain several image sub-regions. If the obtained image sub-regions cannot determine the category of the target welded part based on the region contour, then the image sub-regions are recorded as abnormal edge regions. How to segment the node monitoring image based on edge recognition and match the corresponding target welded part is a content that is easy for those skilled in the art to understand and will not be elaborated here.

[0089] If the acquisition quality coefficient of a node monitoring image is less than the preset acquisition quality coefficient, the light source distribution index of the second-class monitoring execution node corresponding to the node monitoring image with an acquisition quality coefficient less than the preset acquisition quality coefficient is increased according to the acquisition quality coefficient and the target reflectance index. Conversely, the light source distribution index of the second-class monitoring execution node corresponding to the node monitoring image with an acquisition quality coefficient greater than or equal to the preset acquisition quality coefficient is not adjusted. The value of the preset acquisition quality coefficient can be determined by the user according to the actual working scenario. For example, the user can set it according to the production line supervision record. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the larger the value of the preset acquisition quality coefficient is. A method for determining the value of the preset acquisition quality coefficient is provided. The production line supervision record that adjusts the monitoring parameters according to the environment and material status is recorded as the quality reference record. The maximum value of the acquisition quality coefficient of each second-class monitoring execution node in the quality reference record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts is recorded as the preset acquisition quality coefficient.

[0090] For a single Class II monitoring execution node, if the acquisition quality coefficient of the node monitoring image of the Class II monitoring execution node is less than the preset acquisition quality coefficient, it indicates that the Class II monitoring execution node cannot meet the environmental and material requirements of the target welded part when acquiring node monitoring images. It is necessary to adjust the monitoring parameters according to the current actual working scenario. Based on the acquisition quality coefficient and the target reflectance index, the light source distribution index of the Class II monitoring execution node when acquiring node monitoring images is increased. The increase in the light source distribution index is positively correlated with the light source adjustment index. The light source distribution index is the number of different angles formed by the light source arranged during node monitoring image acquisition and the center position corresponding to the node monitoring image. The light source adjustment index = target reflectance index / acquisition quality coefficient. The target reflectance index is the maximum value of the gloss of each target welded part monitored in the current production line supervision cycle at the welding processing stage of the Class II monitoring execution node. How to determine the gloss of the target welded part is a matter that is known to those skilled in the art and will not be elaborated here.

[0091] Specifically, if the production evaluation condition responded by the production analysis module is that the production alternation frequency is less than or equal to the preset production alternation frequency and the production synchronization coefficient is less than or equal to the preset production synchronization coefficient, then the monitoring and evaluation module will perform monitoring quality analysis on the target supervised production line, including:

[0092] The monitoring and evaluation module responds to monitoring and analysis conditions and determines the regulatory difference parameters for each monitoring execution node based on the node quality inspection coefficient and the production continuity index.

[0093] Determine whether to update monitoring parameters for each monitoring execution node based on regulatory difference parameters;

[0094] The monitoring and analysis conditions are determined by the production analysis module to enable the monitoring and evaluation module to perform monitoring quality analysis on the target supervised production line.

[0095] Among them, if the production alternation frequency of the target supervised production line is less than or equal to the preset production alternation frequency and the production synchronization coefficient is less than or equal to the preset production synchronization coefficient, it indicates that the target supervised production line has a low risk of mixed target welded parts within the current production line evaluation cycle. Therefore, by evaluating the difference between the quality inspection results of each monitoring execution node and other related monitoring execution nodes in real time, it is determined whether there are any abnormalities in each monitoring execution node, and the monitoring parameters of each monitoring execution node are adjusted according to the analysis results. For a single monitoring execution node, the node quality inspection coefficient is the number of abnormal products determined by the monitoring execution node during the quality inspection monitoring phase. The time when the node quality inspection coefficient is obtained is recorded as the end time of the quality inspection monitoring phase. The duration of the quality inspection monitoring phase can be determined by the user according to the actual work scenario. For example, the user can set it according to the production line supervision record. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the longer the duration of the quality inspection monitoring phase. One possible value for the duration of the quality inspection monitoring phase is 10 hours.

[0096] For any two monitoring execution nodes that have the same target welded part, the target welded part shared by the two monitoring execution nodes is denoted as the reference evaluation product for the two monitoring execution nodes. If the production continuity index of the reference evaluation product between the two monitoring execution nodes is less than a preset production continuity index, where the production continuity index is the number of welding processing steps between the welding processing steps of the reference evaluation product of the two monitoring execution nodes, then the two monitoring execution nodes are determined to be difference evaluation nodes. For any one monitoring execution node, the regulatory difference parameter... n is the number of difference evaluation nodes of the monitoring execution node, Yi is the node quality inspection difference coefficient of the i-th difference evaluation node of the monitoring execution node, the node quality inspection difference coefficient is the absolute value of the difference between the node quality inspection coefficient of the difference evaluation node and the node quality inspection coefficient of the monitoring execution node, Xi is the difference evaluation coefficient corresponding to the i-th difference evaluation node, and the difference evaluation coefficient is negatively correlated with the production continuity index of each difference evaluation node.

[0097] The value of the preset production continuity index can be determined by the user based on the actual work scenario. For example, the user can set it based on the production line monitoring records. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset production continuity index. A method for determining the value of the preset production continuity index is provided, which is the average value of the production continuity index among the monitoring execution nodes that are mutual difference evaluation nodes in the production line monitoring records that meet the user's requirements for the accuracy of the quality inspection results of the target welded parts.

[0098] Specifically, if the monitoring and evaluation module responds to the difference analysis condition that there is a regulatory execution node with a regulatory difference parameter greater than the preset regulatory difference parameter, then the difference compensation module will adjust the monitoring parameters of the difference analysis node and determine the difference compensation method for each difference analysis node based on the difference concentration index.

[0099] Regulatory execution nodes whose regulatory difference parameters are greater than the preset regulatory difference parameters are recorded as difference analysis nodes.

[0100] Specifically, the compensation setting condition for the difference compensation module is that if the difference concentration index of a difference analysis node is greater than the preset difference concentration index, then the monitoring parameters of that difference analysis node are updated according to the execution difference coefficient.

[0101] The compensation setting condition for the difference compensation module is that if the difference concentration index of a difference analysis node is less than or equal to a preset difference concentration index, then the light source richness parameter of that difference analysis node is adjusted according to the image confidence stability coefficient.

[0102] The value of the preset regulatory difference parameter can be determined by the user according to the actual working scenario. For example, the user can set it according to the production line monitoring record. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset regulatory difference parameter. A method for determining the value of the preset regulatory difference parameter is provided, which is the minimum value of the regulatory difference parameter of each monitoring execution node in the production line monitoring record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts.

[0103] For a single regulatory execution node, if the regulatory difference parameter of the regulatory execution node is greater than the preset regulatory difference parameter, it indicates that the quality inspection result of the regulatory execution node is different from other regulatory execution nodes in the target regulatory production line, and thus indicates that the regulatory execution node needs to adjust the monitoring parameters corresponding to the target welded part. The difference concentration index = 1 / the average value of each adjacent difference continuity index of the regulatory execution node. For any two difference assessment nodes, if there are no other difference assessment nodes in the production process of the reference assessment product, the number of welding processing steps between the welding processing steps of the reference assessment product corresponding to the two difference assessment nodes is recorded as the adjacent difference continuity index.

[0104] The value of the preset difference concentration index can be determined by the user according to the actual work scenario. For example, the user can set it according to the production line monitoring record. A method for determining the value of the preset difference concentration index is provided. The production line monitoring record that adjusts the monitoring parameters of the difference analysis node according to the execution difference coefficient is recorded as the centralized reference record. The average value of the difference concentration index of each difference analysis node in the centralized reference record that meets the user's requirements for the accuracy of the quality inspection results of the target welded parts is recorded as the preset difference concentration index.

[0105] The fact that the regulatory difference parameters of each difference analysis node are greater than the preset regulatory difference parameters indicates that the quality inspection results obtained by the difference analysis node differ from the quality inspection results of many monitoring execution nodes within the target regulatory production line. The reliability of the quality inspection results obtained by the difference analysis node is low, and therefore, its monitoring parameters need to be adjusted. For a single difference analysis node, if the difference concentration index of the difference analysis node is greater than the preset difference concentration index, it indicates that the monitoring execution nodes that differ from the quality inspection results of the difference analysis node are relatively concentrated. This suggests that the monitoring parameters used by the difference analysis node in acquiring node monitoring images are not suitable for the current working scenario. By analyzing whether the monitoring parameters of each difference assessment node are consistent, it can be determined whether the monitoring parameters can guarantee the reliability of the quality inspection results, and the monitoring parameters can be adjusted accordingly. If the difference concentration index of the difference analysis node is less than or equal to the preset difference concentration index, it indicates that the monitoring execution nodes that differ from the quality inspection results of the difference analysis node are relatively dispersed. That is, the quality inspection results of the difference analysis node and its existing difference assessment nodes cannot guarantee reliability, and the current monitoring parameters for node monitoring image acquisition for the target welded parts cannot meet the actual working scenario.

[0106] For a single difference analysis node, if the difference concentration index of that node is greater than a preset difference concentration index, the execution difference coefficient among the difference evaluation nodes of that node is detected. The execution difference coefficient is the sum of the products of the execution difference parameter of each category of monitoring parameters and the corresponding difference weight coefficient of that category of monitoring parameters. For a single category of monitoring parameters, the execution difference parameter is the absolute value of the difference between the maximum and minimum values ​​of the monitoring parameter of that category at each difference evaluation node of the difference analysis node. This is used to determine the values ​​of the monitoring parameters of each category and their corresponding difference weight coefficients for the execution difference coefficient. Users can then use this information to determine the values ​​of the execution difference coefficients. The system is configured for actual work scenarios, providing a method to determine the categories of monitoring parameters for execution difference coefficients and the values ​​of the difference weight coefficients corresponding to each category of monitoring parameters. The categories of monitoring parameters include: resolution, depth of field, aperture, and focal length. The value of the difference weight coefficient corresponding to each category of monitoring parameters is 0.25. If the execution difference coefficient of the difference analysis node is less than the preset execution difference coefficient, then the monitoring parameters of each category of the difference analysis node with the execution difference coefficient less than the preset execution difference coefficient are set according to the execution difference parameters of each category of monitoring parameters. The average value of the monitoring parameters of each category of the difference evaluation nodes of the difference analysis node is recorded as the value of the monitoring parameter corresponding to each category of the difference analysis node.

[0107] The value of the preset execution difference coefficient can be determined by the user according to the actual working scenario. For example, the user can set it according to the production line monitoring records. The higher the user's requirement for the accuracy of the quality inspection results of the target welded parts, the smaller the value of the preset execution difference coefficient. A method for determining the value of the preset execution difference coefficient is provided, in which the production line monitoring records that adjust the monitoring parameters of the difference analysis node according to the execution difference coefficient are recorded as a type of compensation record, and the maximum value of the execution difference coefficient of the difference analysis node in the type of compensation record that meets the user's requirement for the accuracy of the quality inspection results of the target welded parts is recorded as the preset execution difference coefficient.

[0108] For a single difference analysis node, if the difference concentration index of the difference analysis node is greater than the preset difference concentration index, the light source richness parameter of the difference analysis node is increased. The increase in the light source richness parameter is negatively correlated with the image confidence stability coefficient. The image confidence stability coefficient = 1 / the number of times the difference analysis node is identified as a difference analysis node during the confidence assessment phase.

[0109] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0110] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A machine vision-based intelligent welding system, characterized in that, include: The production analysis module is used to periodically respond to production assessment conditions in order to determine production line monitoring strategies and to perform monitoring node analysis or monitoring quality analysis for the target production line. The node monitoring module, which is connected to the production analysis module, is used to determine the category of each monitoring execution node in response to node status assessment conditions, and to determine the interference analysis method as either acquisition reflection analysis or acquisition quality analysis for node monitoring images in response to analysis setting conditions. An interference analysis module, connected to the node monitoring module, is used to respond to reflection analysis conditions, determine node adjustment strategies based on abnormal reflection parameters and abnormal reflection distribution index, and respond to quality analysis conditions, determine whether to adjust the light source distribution index of the second-class monitoring execution node based on abnormal reflection parameters and edge anomaly parameters. The monitoring and evaluation module, which is connected to the production analysis module, is used to identify the difference analysis nodes of the target monitored production line and, in response to the difference analysis conditions, to determine whether to update the monitoring parameters for each monitoring execution node. The difference compensation module, which is connected to the monitoring and evaluation module, is used to determine the difference compensation method for each difference analysis node in response to the compensation setting conditions. This method is either to update the monitoring parameters of the difference analysis node based on the difference coefficient, or to adjust the light source richness parameters of the difference analysis node based on the image confidence stability coefficient.

2. The intelligent welding system based on machine vision according to claim 1, characterized in that, If the production evaluation condition responded by the production analysis module is that the production alternation frequency is greater than the preset production alternation frequency or the production synchronization coefficient is greater than the preset production synchronization coefficient, then the node monitoring module will perform monitoring node analysis on the target supervised production line, including: The node monitoring module responds to node analysis conditions and determines the monitoring interference status of each monitoring execution node within the target supervised production line based on the node association category coefficient and category interference index. The node analysis method for each monitoring execution node is determined based on the monitoring interference status; The node analysis conditions are determined by the production analysis module to enable the node monitoring module to perform monitoring node analysis on the target supervised production line.

3. The intelligent welding system based on machine vision according to claim 2, characterized in that, The node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is greater than the preset node association category coefficient or the category interference index is greater than the preset category interference index, then the monitored execution node is determined to be a type of monitored execution node. The node status evaluation condition of the node monitoring module is that if the node association category coefficient of the monitored execution node is less than or equal to the preset node association category coefficient and the category interference index is less than or equal to the preset category interference index, then the monitored execution node is determined to be a Class II monitored execution node.

4. The intelligent welding system based on machine vision according to claim 3, characterized in that, The node monitoring module's response analysis setting condition is the existence of a type of monitoring execution node. The determination interference analysis module performs reflection analysis on the node monitoring images of each type of monitoring execution node, including: The interference analysis module responds to the reflection analysis conditions and determines the abnormal reflection points based on the reflection difference coefficient corresponding to the point analysis range. The interferometry analysis module determines the node adjustment strategy based on the abnormal reflection parameters and the abnormal reflection distribution index. The reflection analysis condition is that the node monitoring module determines that the interference analysis module collects and analyzes the reflection of the node monitoring images of each type of monitoring execution node.

5. The intelligent welding system based on machine vision according to claim 4, characterized in that, The interference distribution condition of the interference analysis module response is that there is a type of monitoring execution node whose node monitoring image has an abnormal reflection parameter greater than a preset abnormal reflection parameter or an abnormal reflection distribution index greater than a preset abnormal reflection index. Then, based on the abnormal reflection distribution index, the acquisition execution angle of this type of monitoring execution node is increased and adjusted. The increase in the acquisition execution angle is positively correlated with the abnormal reflection distribution index.

6. The intelligent welding system based on machine vision according to claim 5, characterized in that, The interference distribution condition of the interference analysis module response is that there is a type of monitoring execution node whose abnormal reflection parameter of the node monitoring image is less than or equal to the preset abnormal reflection parameter and whose abnormal reflection distribution index is less than or equal to the preset abnormal reflection distribution index. Then, based on the reference reflection difference coefficient, the light source richness parameter of the back monitoring node of this type of monitoring execution node is increased and adjusted. The increase in the richness parameters of the light source is positively correlated with the reference reflection difference coefficient.

7. The intelligent welding system based on machine vision according to claim 6, characterized in that, The node monitoring module's response analysis setting condition is the existence of two types of monitoring execution nodes. In this case, the interference analysis module performs a quality analysis of the node monitoring images for each type of monitoring execution node, including: The interferometric analysis module responds to quality analysis conditions and determines the acquisition quality coefficient of the monitoring image of each node based on abnormal reflection parameters and edge abnormal parameters. The change setting condition for the response of the interference analysis module is that if the acquisition quality coefficient of the monitoring image of a node is less than the preset acquisition quality coefficient, then the light source distribution index of the second type of monitoring execution node corresponding to the monitoring image of that node is increased and adjusted according to the acquisition quality coefficient and the target reflection index. The quality analysis condition is that the node monitoring module determines the quality analysis of the node monitoring images collected by the interference analysis module for each type II monitoring execution node.

8. The intelligent welding system based on machine vision according to claim 7, characterized in that, If the production evaluation condition responded by the production analysis module is that the production alternation frequency is less than or equal to the preset production alternation frequency and the production synchronization coefficient is less than or equal to the preset production synchronization coefficient, then the monitoring and evaluation module will perform monitoring quality analysis on the target supervised production line, including: The monitoring and evaluation module responds to monitoring and analysis conditions and determines the regulatory difference parameters for each monitoring execution node based on the node quality inspection coefficient and the production continuity index. Determine whether to update monitoring parameters for each monitoring execution node based on regulatory difference parameters; The monitoring and analysis conditions are determined by the production analysis module to enable the monitoring and evaluation module to perform monitoring quality analysis on the target supervised production line.

9. The intelligent welding system based on machine vision according to claim 8, characterized in that, If the monitoring and evaluation module responds to the difference analysis condition that there is a regulatory execution node with a regulatory difference parameter greater than the preset regulatory difference parameter, then the difference compensation module will adjust the monitoring parameters of the difference analysis node and determine the difference compensation method for each difference analysis node based on the difference concentration index. Regulatory execution nodes whose regulatory difference parameters are greater than the preset regulatory difference parameters are recorded as difference analysis nodes.

10. The intelligent welding system based on machine vision according to claim 9, characterized in that, The compensation setting condition for the difference compensation module is that if the difference concentration index of a difference analysis node is greater than the preset difference concentration index, then the monitoring parameters of the difference analysis node are updated according to the execution difference coefficient. The compensation setting condition for the difference compensation module is that if the difference concentration index of a difference analysis node is less than or equal to a preset difference concentration index, then the light source richness parameter of that difference analysis node is adjusted according to the image confidence stability coefficient.

Images