A camera welding control system based on vision
By introducing welding quality analysis of visual acquisition and deep learning models into the welding control system, weld defects are monitored and compensated in real time, and the problem of welding failures cannot be handled in time is solved, and the welding quality and efficiency are improved.
Patent Information
- Application Number
- CN202411191741.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-28
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2044-08-28
AI Technical Summary
The existing welding robot control system cannot monitor the welding quality of camera accessories in real time, resulting in the inability to re-weld the welding unqualified products in time, affecting the welding quality and efficiency.
A vision-based camera welding control system is designed, including welding guns, trajectory monitoring modules, vision acquisition modules, image processing modules, decision-making modules and control modules. The welding quality images are captured in real time through industrial cameras, and the welding defects are analyzed using deep learning models. , and perform rewelding or remelting operations based on the analysis results, displaying monitoring results and alarm information in real time.
Real-time monitoring of welding quality and re-welding or remelting operations are achieved, welding yield is improved, dependence on additional quality inspection equipment is reduced, and welding efficiency is improved.
Smart Images

Figure CN118699670B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of camera welding control systems, and in particular relates to a vision-based camera welding control system. Background Art
[0002] When producing cameras, welding equipment is required to connect different electrical components or parts of the camera together to ensure that the electrical and mechanical connections between different electrical components or parts are firm and reliable, while also ensuring the transmission of signals and power, which can increase the service life and reliability of the camera. In order to improve the accuracy and efficiency of welding of camera accessories, automated welding equipment or welding robots are usually used for welding operations. In order to enable automated welding equipment or welding robots to perform automated production, a complete welding control system is required to automatically control the automated welding equipment or welding robots.
[0003] As shown in the patent document, application number 202210378049.5, a vision-based welding robot control system is disclosed, which is composed of a main control module, a sub-control module, a teaching module, a power module, a drive module, a sensor module and a welding module. Compared with the prior art, the invention provides a vision-based welding robot control system, which can perform major operations and controls on the welding robot through the main control module and the sub-control module in cooperation with the teaching module, and can monitor the welding trajectory and welding working status of the welding robot through the sensor module in cooperation with the main control module, and can adjust the parameter information in the drive module through the sub-control module to make real-time adjustments to the welding robot, thereby further reducing welding errors.
[0004] However, the above-mentioned welding robot control system monitors the welding trajectory of the welding robot, but does not monitor the welding quality of the product. When the welding of the camera accessories is unqualified, it is generally necessary to use quality inspection equipment to first screen out the unqualified products, and then use welding equipment to repair them. The unqualified products cannot be repaired in time, which affects the welding quality of the camera accessories. Therefore, we need to propose a vision-based camera welding control system to solve the above-mentioned problems. Summary of the invention
[0005] In view of the above problems, the present invention provides a vision-based camera welding control system, comprising: a welding robot including a welding gun and used to perform welding operations, a trajectory monitoring module used to collect welding trajectories and welding states of the welding gun of the welding robot, and a visual acquisition module used to collect welding quality image information of camera accessories according to the welding trajectory, wherein the visual acquisition module comprises an industrial camera installed above the welding gun of the welding robot and capturing the welding quality of the camera accessories in real time;
[0006] An image processing module for processing and analyzing the welding quality image information of the camera accessories collected by the visual acquisition module, wherein the image processing module uses a deep learning model to analyze welding quality defects and evaluate whether the welding quality meets the welding standards;
[0007] A decision-making module for formulating a repair welding or remelting plan according to the welding quality evaluation result of the image processing module, wherein one end of the decision-making module is connected to a maintenance path formulation module for formulating a maintenance path according to the repair welding or remelting requirements;
[0008] A control module for controlling the welding robot to perform repair welding operations according to the repair welding path, and the control module is respectively connected to the trajectory monitoring module, the visual acquisition module, the image processing module, the decision-making module and the maintenance path formulation module;
[0009] Human-computer interaction module for real-time display of visual monitoring results, welding parameters and alarm information.
[0010] Furthermore, the trajectory monitoring module includes an optical sensor for capturing the relative position between the welding gun and the weld of the welding accessory in real time, an arc sensor for indirectly judging the position and shape of the weld by utilizing the changes in arc characteristics during the welding process, a data processor for digitally processing the data collected by the optical sensor and the arc sensor, and a feedback control unit for feedback and control.
[0011] Furthermore, the visual acquisition module includes an industrial camera installed above the welding robot's welding gun and capturing the welding quality of the camera accessories in real time. The industrial camera is installed above the welding gun, and then the parameters of the industrial camera are set. During the welding process, the industrial camera captures an image of the welding area through the lens, and the image includes the shape of the weld and the appearance of the weld.
[0012] Furthermore, the steps of analyzing welding quality defects by the image processing module are as follows:
[0013] A1. Select a batch of welding defect images and manually mark the welding quality defects to obtain a welding defect image set and a verification set;
[0014] A2. Perform image denoising, resizing, and grayscale adjustment on the welding defect image set to facilitate subsequent deep learning model processing;
[0015] A3. Use the processed welding defect image set to train a deep learning model. The deep learning model uses an existing pre-trained model or a self-designed and trained model.
[0016] A4. Use the validation set to evaluate the trained deep learning model and optimize the model based on the evaluation results.
[0017] A5. Perform image enhancement processing on the welding quality images collected in real time;
[0018] A6. Input the enhanced welding quality image into the trained deep learning model for quality defect analysis, so that the deep learning model can automatically identify defects in the welding image and output the location and type of the defects;
[0019] A7. Based on the identified welding defects and in combination with the welding standards, evaluate whether the welding quality meets the requirements. If there are welding defects that can be repaired or remelted, the repair information needs to be transmitted to the decision-making module for repair or remelting operations. If there are welding defects that cannot be repaired, the defective products will be transferred to the unqualified products through the robotic arm.
[0020] Furthermore, in step A5, when performing image enhancement processing on the collected welding quality image, the original image is recorded as X, and its pixel probability density distribution is recorded as , where x represents the pixel value variable, the converted image is recorded as Y, and its pixel probability density distribution is recorded as , where y represents the pixel value variable, X is transformed to obtain Y, and is a constant, and the selected conversion function is the cumulative distribution function, that is,
[0021] , (1)
[0022] Among them, L is ,
[0023] At the same time, according to probability knowledge, we can know that:
[0024] , (2)
[0025] Differentiating formula 1 yields:
[0026] , (3)
[0027] According to formula 2 and formula 3, we can get:
[0028] , (4).
[0029] Furthermore, the repair welding methods formulated by the decision-making module include wire repair welding, filler repair welding and local re-welding. When formulating the repair welding plan, the influence of welding defects on welding strength and quality is evaluated according to the size, depth and location of the welding defects as well as the welding standards and requirements, and a repair welding plan using one of the methods of wire repair welding, filler repair welding and local re-welding is formulated.
[0030] Furthermore, the remelting plans formulated by the decision-making module include local remelting, overall remelting, trimming and re-welding. According to the size, shape and location of the welding defects, as well as the welding standards and requirements, the impact of welding defects on welding strength and quality is evaluated, and then according to the nature and location of the welding defects, one of the remelting plans among local remelting, overall remelting, trimming and re-welding is formulated.
[0031] Furthermore, the maintenance path formulation module includes the following steps when formulating the maintenance path:
[0032] B1. Conduct a comprehensive analysis of the need for repair welding or remelting to understand the specific welding defects that need to be repaired, the location and nature of the defects, and the goals and requirements of the repair;
[0033] B2. Plan the working space of the welding robot, determine the range that the welding robot can reach and operate, and take into account safety and operational feasibility;
[0034] B3. Based on the workspace planning, the moving path of the welding robot is planned to determine the specific path and steps of the maintenance. The specific path and steps of the maintenance include the moving trajectory of the welding robot and the sequence of welding points. When planning the path, the efficiency, safety and working space limitations of the maintenance should be taken into consideration.
[0035] B4. Based on the specific paths and steps of maintenance, the motion control design of the welding robot is carried out. The motion control includes the robot's trajectory planning, motion planning and trajectory tracking;
[0036] B5. Move and perform maintenance operations on the welding robot according to the established maintenance path to ensure that the welding robot can move along the planned path and complete maintenance tasks.
[0037] Furthermore, when the control module performs the repair welding operation, the control module includes the following steps:
[0038] C1. The control module needs to receive input information of the repair welding path, and the input information of the repair welding path includes the position coordinates of the repair welding and the shape and size of the weld to be repaired;
[0039] C2. After receiving the input information of the repair welding path, the control module converts the repair welding path into a specific motion path that the welding robot can execute; the motion path includes converting the target path into a path in the robot coordinate system, and considering the robot kinematics and obstacle avoidance factors;
[0040] C3, the control module generates motion control instructions for the welding robot according to the results of path planning; the motion control instructions involve the trajectory, speed and acceleration control of each joint of the robot, as well as the posture control of the welding gun or welding head. The motion control instructions are sent to the welding robot through the interface of the robot control system;
[0041] C4. During the repair welding operation, the control module needs to monitor the movement and status of the welding robot in real time, including detecting the actual position and posture of the robot and the feedback information during the welding process. Through sensors and feedback control systems, the various parameters of the welding process can be obtained in real time, and the robot can be adjusted and controlled in real time.
[0042] C5. When the repair welding operation is completed, the control module needs to send an instruction to end the action so that the welding robot can safely exit the working state.
[0043] Furthermore, the human-computer interaction module includes a visualization interface that displays monitoring results, welding parameters and alarm information in the form of graphics or text. The visualization interface uses charts, tables and images to enable operators to intuitively understand and analyze data.
[0044] The beneficial effects of the present invention are:
[0045] 1. The present invention collects welding quality image information of camera accessories through a visual acquisition module, and transmits the image information to an image processing module in real time for analysis and processing, so as to determine whether the welded accessories have welding defects. If welding defects exist, the image processing module makes a repair welding decision or a remelting decision according to the welding defects. The maintenance path formulation module processes and analyzes the welding defects and formulates a moving path of the welding robot's welding gun during repair welding or remelting. The control module controls the welding robot's welding gun to perform repair welding or remelting operations according to the moving path, so that the camera accessories can monitor the product welding quality in real time during welding, and timely repair welding or remelting operations on some unqualified products according to the monitoring results, thereby improving the welding yield rate of accessories to a certain extent, and no additional quality inspection equipment is required for welding quality inspection, which improves the welding efficiency of camera accessories to a certain extent;
[0046] 2. The present invention can provide effective identification of welding defects through the setting of the image processing module, combine the welding standards, and evaluate whether the welding quality meets the requirements, thereby improving the welding quality of camera accessories.
[0047] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0049] Figure 1 A system block diagram according to an embodiment of the present invention is shown;
[0050] Figure 2 A flow chart according to an embodiment of the present invention is shown;
[0051] Figure 3 A flowchart of welding quality defect analysis according to an embodiment of the present invention is shown;
[0052] Figure 4 A flowchart of a maintenance path formulation module according to an embodiment of the present invention is shown;
[0053] Figure 5 A flow chart of the repair welding operation according to an embodiment of the present invention is shown. DETAILED DESCRIPTION
[0054] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0055] The embodiment of the present invention provides a camera welding control system based on vision, such as Figure 1-5 As shown, it includes a welding robot, a trajectory monitoring module, a visual acquisition module, an image processing module, a decision-making module, a maintenance path formulation module, a control module and a human-computer interaction module, wherein the welding robot includes a welding gun for performing welding operations. The welding robot uses the current existing technology (such as the welding robot disclosed in the patent document application number 201910311456.2). The welding robot has six degrees of freedom and can perform automated welding, which will not be described in detail here.
[0056] The trajectory monitoring module is used to collect the welding trajectory and welding status of the welding robot's welding gun; the trajectory monitoring module includes an optical sensor for capturing the relative position between the welding gun and the weld of the welding accessory in real time, an arc sensor that uses the changes in arc characteristics during the welding process to indirectly judge the position and shape of the weld, a data processor that digitally processes the data collected by the optical sensor and the arc sensor, and a feedback control unit that performs feedback and control. Through the mutual coordination of the optical sensor and the arc sensor, the data processor and the feedback control unit, the welding trajectory and welding status can be effectively monitored in real time, and accurate trajectory correction and process parameter adjustment can be performed based on the monitoring results to ensure high-quality and efficient automated welding operations.
[0057] Among them, the optical sensor is installed in front of the welding gun. The optical sensor can use an infrared camera to capture the relative position information between the welding gun and the weld of the camera accessory in real time. Through image processing, it can accurately measure the geometric features such as the position, width and depth of the weld and the deviation of the weld edge to ensure that the welding gun always moves along the weld path.
[0058] The arc sensor uses the changes in arc characteristics (such as current, voltage, arc length, etc.) during the welding process to indirectly determine the position and shape of the weld. When the welding gun deviates from the center of the weld, these parameters will change accordingly. By analyzing these changes and feeding back to the control system, real-time trajectory correction can be performed.
[0059] The data processor collects welding current, voltage, wire feed speed, gas flow, welding speed and other process parameters in real time through optical sensors and arc sensors integrated in the welding power supply or welding circuit. After the collected data is digitized, it is analyzed in real time through an embedded processor or a dedicated controller to identify changes in the weld position.
[0060] The feedback control unit calculates the corrected trajectory of the welding gun in real time based on the information fed back by the optical sensor and the arc sensor, and ensures the precise movement of the welding gun along the weld by adjusting the joint angle of the welding robot or the position of the linear axis, including dynamic adjustment of the welding gun's posture (such as swing amplitude and angle), welding speed, and arc starting and ending positions. It then uses visual systems, infrared thermal imagers and other equipment to monitor the thermal deformation of the workpiece during the welding process, and predicts its impact on the subsequent welding trajectory by analyzing the thermal deformation data. It then performs compensation calculations in advance to ensure that the welding robot can accurately track changes in the weld position caused by thermal expansion or contraction.
[0061] The visual acquisition module is used to collect image information of the welding quality of camera accessories according to the welding trajectory. The visual acquisition module includes an industrial camera installed above the welding gun of the welding robot and capturing the welding quality of the camera accessories in real time; wherein the industrial camera is equipped with a high-resolution, low-noise, and high-sensitivity image sensor to ensure clear and accurate images under various lighting conditions; it adopts industrial standard interfaces, such as GigE Vision, USB3 Vision, Camera Link, etc., to ensure stable data transmission and compatibility with industrial equipment; it supports high-speed continuous acquisition and fast image transmission, and is suitable for real-time monitoring and control on industrial automation and high-speed production lines.
[0062] Install the industrial camera above the welding gun to ensure that the industrial camera can effectively capture the image of the welding area, and then set the parameters of the industrial camera to obtain a clear and accurate welding image; the parameters include adjusting the exposure time, gain, and focus parameters. During the welding process, the industrial camera captures the image of the welding area through the lens, and the image includes the shape of the weld and the appearance of the weld.
[0063] The image processing module is used to process and analyze the welding quality image information of the camera accessories collected by the visual acquisition module. The image processing module uses a deep learning model to analyze the welding quality defects and evaluate whether the welding quality meets the welding standards, such as Figure 3 As shown, the steps of welding quality defect analysis are as follows:
[0064] A1. Select a batch of welding defect images, and manually mark the welding quality defects to obtain a welding defect image set and a verification set, wherein the welding defect image set includes a welding defect image set that can be repaired, a welding defect image set that can be remelted, and a welding defect image set that cannot be repaired, and the verification set includes welding defect images that can be repaired, a welding defect image that can be remelted, and a welding defect image that cannot be repaired;
[0065] A2. Perform image denoising, resizing, and grayscale adjustment on the welding defect image set to facilitate subsequent deep learning model processing;
[0066] A3. Use the processed welding defect image set to train a deep learning model. The deep learning model uses an existing pre-trained model or a self-designed and trained model. Existing pre-trained models include VGG, ResNet, etc.
[0067] A4. Use the validation set to evaluate the trained deep learning model and perform model optimization based on the evaluation results, such as adjusting model parameters and increasing the amount of data.
[0068] A5. Perform image enhancement processing on the welding quality images collected in real time to increase the accuracy of welding quality image processing;
[0069] A6. Input the enhanced welding quality image into the trained deep learning model for quality defect analysis, so that the deep learning model can automatically identify defects in the welding image and output the location and type of the defects;
[0070] A7. Based on the identified welding defects and in combination with the welding standards, evaluate whether the welding quality meets the requirements. If there are welding defects that can be repaired or remelted, the repair information needs to be transmitted to the decision-making module for repair or remelting operations. If there are welding defects that cannot be repaired, the defective products will be transferred to the unqualified products through the robotic arm.
[0071] In step A5, when performing image enhancement processing on the collected welding quality image, the original image is recorded as X, and its pixel probability density distribution is recorded as , where x represents the pixel value variable. The converted image is recorded as Y, and its pixel probability density distribution is recorded as , where y represents the pixel value variable, X is transformed to obtain Y, and is a constant, and the selected conversion function is the cumulative distribution function, that is,
[0072] , (1)
[0073] Among them, L is ,
[0074] At the same time, according to probability knowledge, we can know that:
[0075] , (2)
[0076] Differentiating formula 1 yields:
[0077] , (3)
[0078] According to formula 2 and formula 3, we can get:
[0079] , (4)
[0080] It can be seen from Formula 4 that when the original image X is converted into a new image Y through the cumulative distribution function, the probability density of Y is a constant.
[0081] The decision-making module is used to formulate a repair welding or remelting plan based on the welding quality evaluation result of the image processing module, wherein the repair welding plan includes wire repair welding, filler repair welding and local re-welding. When formulating the repair welding plan, the influence of welding defects on welding strength and quality is evaluated according to the size, depth and position of the welding defects as well as the welding standards and requirements, and a repair welding plan is formulated using one of the methods of wire repair welding, filler repair welding and local re-welding. The wire repair welding plan: use welding wire of the same or similar material for welding supplementation, and perform repair welding operations on defective parts to ensure that the welding defects are filled.
[0082] Filler repair welding solution: Use filler materials such as welding glue to fill the defective parts, and then heat the filler to weld with the workpiece to achieve the repair welding effect.
[0083] Partial re-welding solution: If the welding defect is larger than one-third of the weld width, the entire welding area needs to be re-welded to ensure that the welding quality meets the requirements.
[0084] Remelting plans include local remelting, overall remelting, trimming and re-welding. According to the size, shape and location of the welding defects, as well as the welding standards and requirements, the impact of welding defects on welding strength and quality is evaluated, and then according to the nature and location of the welding defects, one of the remelting plans among local remelting, overall remelting, trimming and re-welding is formulated. The local remelting plan is to locally remelt the welding defect area, set the welding parameters of the welding robot, and re-melt the solder at the welding defect to form a suitable welding shape and connection.
[0085] Overall remelting solution: If the width of the welding defect is greater than one-third of the weld width, the entire welding area needs to be remelted as a whole. By heating the entire welding area, the solder is re-melted and the welding connection is re-formed to repair and improve the welding defect.
[0086] Trimming and re-soldering solution: Trim the soldering defects with thorns or bridging, remove the excess solder, and then re-solder to ensure that the soldering quality meets the requirements.
[0087] One end of the decision-making module is connected to a maintenance path formulation module for formulating a maintenance path according to the repair welding or remelting requirements; when the maintenance path formulation module is formulating the maintenance path, Figure 4 As shown, the following steps are included:
[0088] B1. Conduct a comprehensive analysis of the need for repair welding or remelting to understand the specific welding defects that need to be repaired, the location and nature of the defects, and the goals and requirements of the repair;
[0089] B2. Plan the working space of the welding robot, determine the range that the welding robot can reach and operate, and take into account safety and operational feasibility;
[0090] B3. Based on the workspace planning, the moving path of the welding robot is planned to determine the specific path and steps of the maintenance. The specific path and steps of the maintenance include the moving trajectory of the welding robot, the sequence of welding points, etc. When planning the path, the efficiency, safety and working space limitations of the maintenance should be taken into consideration;
[0091] B4. Based on the specific paths and steps of maintenance, the motion control design of the welding robot is carried out. The motion control includes the robot's trajectory planning, motion planning and trajectory tracking;
[0092] B5. Move and perform maintenance operations on the welding robot according to the established maintenance path to ensure that the welding robot can move along the planned path and complete maintenance tasks.
[0093] The control module controls the welding robot to perform the repair welding operation according to the repair welding path, and the control module is respectively connected with the trajectory monitoring module, the visual acquisition module, the image processing module, the decision-making module and the maintenance path formulation module;
[0094] When the control module is performing repair welding, Figure 5 As shown, the following steps are included:
[0095] C1. The control module needs to receive input information of the repair welding path, and the input information of the repair welding path includes the position coordinates of the repair welding and the shape and size of the weld to be repaired;
[0096] C2. After receiving the input information of the repair welding path, the control module converts the repair welding path into a specific motion path that the welding robot can execute; the motion path includes converting the target path into a path in the robot coordinate system, and considering the robot kinematics and obstacle avoidance factors;
[0097] C3, the control module generates motion control instructions for the welding robot according to the results of path planning; the motion control instructions involve the trajectory, speed and acceleration control of each joint of the robot, as well as the posture control of the welding gun or welding head. The motion control instructions are sent to the welding robot through the interface of the robot control system;
[0098] C4. During the repair welding operation, the control module needs to monitor the movement and status of the welding robot in real time, including detecting the actual position and posture of the robot and the feedback information during the welding process. Through sensors and feedback control systems, the various parameters of the welding process can be obtained in real time, and the robot can be adjusted and controlled in real time to ensure the accuracy and stability of the repair welding operation.
[0099] C5. When the repair welding operation is completed, the control module needs to send an instruction to end the action so that the welding robot can safely exit the working state.
[0100] The human-computer interaction module is used to display visual monitoring results, welding parameters and alarm information in real time.
[0101] The human-computer interaction module includes a visual interface that displays monitoring results, welding parameters and alarm information in graphical or textual form. The visual interface uses charts, tables and images to enable operators to intuitively understand and analyze data.
[0102] The human-machine interaction module transmits data through the interface with the visual monitoring system and the welding control system, and needs to receive data, welding parameters and alarm information from the visual monitoring system in real time, and update them to the visual interface in real time. For example, the latest monitoring data and parameters can be periodically obtained through the network or local connection, and updated to the visual interface.
[0103] After receiving data from the visual monitoring system, the human-computer interaction module displays the data on a visual interface. The displayed information includes the inspection results of the weld, the location and type of welding defects, and then uses image processing technology to superimpose the monitoring results with the weld image, or indicate the location of the welding defect on the image for the operator to observe and analyze.
[0104] The human-computer interaction module displays the current welding parameters (such as current, voltage, speed, etc.) and the set target parameters on a visual interface, allowing operators to understand the parameter adjustments during the welding process at any time and make necessary interventions and adjustments.
[0105] If an abnormal situation or an alarm message occurs during the welding process, the human-computer interaction module can display and remind the operator in time. This can be an alarm message in the form of text prompts, sound prompts or pop-up windows. At the same time, the module can provide corresponding solutions or suggestions to help operators deal with problems.
[0106] The trajectory monitoring module is used to monitor the movement trajectory of the welding robot's welding gun. When the welding robot's welding gun performs welding operation, the visual acquisition module is triggered to collect the welding quality image information of the camera accessories, and the image information is transmitted to the image processing module in real time for analysis and processing to determine whether the welded accessories have welding defects. Welding defects include voids, insufficient welding, welding burrs, and tin bridging. If there are no welding defects, the welded products are directly transferred by the robotic arm, and the welded accessories are transferred to the next workstation for the next step. If there are welding defects, the image processing module makes a repair welding decision based on the welding defects. If the welding defect is voids or insufficient welding, repair welding is required. If the welding defect is tin bridging or welding burrs, remelting is required. If the defect is not one or more of voids, insufficient soldering, welding burrs and tin bridging, the welded accessories need to be transferred to the unqualified products through the robotic arm, and the welding defect data of the repair and remelting operations are transmitted to the maintenance path formulation module. The maintenance path formulation module processes and analyzes the welding defects and formulates the movement path of the welding robot's welding gun during repair or remelting. The control module controls the welding robot's welding gun to perform repair or remelting operations according to the movement path, so that the camera accessories can monitor the product welding quality in real time during welding, and repair or remelt some unqualified products in time according to the monitoring results, thereby improving the welding yield of accessories to a certain extent, and does not require additional quality inspection equipment for welding quality inspection, which improves the welding efficiency of camera accessories to a certain extent.
[0107] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent substitutions for some of the technical features therein; and these modifications or substitutions do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A vision-based camera welding control system, characterized in that: include: A welding robot including a welding gun and used to perform welding operations, a trajectory monitoring module for collecting welding trajectories and welding states of the welding gun of the welding robot, and a visual acquisition module for collecting image information of welding quality of camera accessories according to the welding trajectory, wherein the visual acquisition module includes an industrial camera installed above the welding gun of the welding robot and capturing the welding quality of the camera accessories in real time; An image processing module for processing and analyzing the welding quality image information of the camera accessories collected by the visual acquisition module, wherein the image processing module uses a deep learning model to analyze welding quality defects and evaluate whether the welding quality meets the welding standards; A decision-making module for formulating a repair welding or remelting plan according to the welding quality evaluation result of the image processing module, wherein one end of the decision-making module is connected to a maintenance path formulation module for formulating a maintenance path according to the repair welding or remelting requirements; A control module for controlling the welding robot to perform repair welding operations according to the repair welding path, and the control module is respectively connected to the trajectory monitoring module, the visual acquisition module, the image processing module, the decision-making module and the maintenance path formulation module; Human-computer interaction module for real-time display of visual monitoring results, welding parameters and alarm information; The steps of the image processing module in analyzing welding quality defects are as follows: A1. Select a batch of welding defect images and manually mark the welding quality defects to obtain a welding defect image set and a verification set; A2. Perform image denoising, resizing, and grayscale adjustment on the welding defect image set to facilitate subsequent deep learning model processing; A3. Use the processed welding defect image set to train a deep learning model. The deep learning model uses an existing pre-trained model or a self-designed and trained model. A4. Use the validation set to evaluate the trained deep learning model and optimize the model based on the evaluation results. A5. Perform image enhancement processing on the welding quality images collected in real time; A6. Input the enhanced welding quality image into the trained deep learning model for quality defect analysis, so that the deep learning model can automatically identify defects in the welding image and output the location and type of the defects; A7. Based on the identified welding defects and welding standards, evaluate whether the welding quality meets the requirements. If there are welding defects that can be repaired or remelted, the repair information needs to be transmitted to the decision-making module for repair or remelting operations. If there are welding defects that cannot be repaired, the unqualified products are transferred to the unqualified products through the robot arm; In step A5, when performing image enhancement processing on the collected welding quality image, the original image is recorded as X, and its pixel probability density distribution is recorded as , where x represents the pixel value variable, the converted image is recorded as Y, and its pixel probability density distribution is recorded as , where y represents the pixel value variable, X is transformed to obtain Y, and is a constant, and the selected conversion function is the cumulative distribution function, that is, ,(1) Among them, L is , At the same time, according to probability knowledge, we can know that: ,(2) Differentiating formula 1 yields: ,(3) According to formula 2 and formula 3, we can get: ,(4)。 2. A vision-based camera welding control system according to claim 1, characterized in that: The trajectory monitoring module includes an optical sensor for capturing the relative position between the welding gun and the weld of the welding accessory in real time, an arc sensor for indirectly judging the position and shape of the weld by utilizing the changes in arc characteristics during the welding process, a data processor for digitally processing the data collected by the optical sensor and the arc sensor, and a feedback control unit for feedback and control.
3. A vision-based camera welding control system according to claim 2, characterized in that: The visual acquisition module includes an industrial camera installed above the welding robot's welding gun and capturing the welding quality of the camera accessories in real time. The industrial camera is installed above the welding gun, and then the parameters of the industrial camera are set. During the welding process, the industrial camera captures an image of the welding area through the lens, and the image includes the shape of the weld and the appearance of the weld.
4. A vision-based camera welding control system according to claim 3, characterized in that: The repair welding methods formulated by the decision-making module include wire repair welding, filler repair welding and local re-welding. When formulating the repair welding plan, the influence of welding defects on welding strength and quality is evaluated according to the size, depth and position of the welding defects as well as the welding standards and requirements, and a repair welding plan using one of the methods of wire repair welding, filler repair welding and local re-welding is formulated.
5. A vision-based camera welding control system according to claim 4, characterized in that: The remelting plans formulated by the decision-making module include local remelting, overall remelting, trimming and re-welding. According to the size, shape and location of the welding defects, as well as the welding standards and requirements, the impact of welding defects on welding strength and quality is evaluated, and then according to the nature and location of the welding defects, one of the remelting plans among local remelting, overall remelting, trimming and re-welding is formulated.
6. A vision-based camera welding control system according to claim 5, characterized in that: The maintenance path formulation module includes the following steps when formulating the maintenance path: B1. Conduct a comprehensive analysis of the need for repair welding or remelting to understand the specific welding defects that need to be repaired, the location and nature of the defects, and the goals and requirements of the repair; B2. Plan the working space of the welding robot, determine the range that the welding robot can reach and operate, and take into account safety and operational feasibility; B3. Based on the workspace planning, the moving path of the welding robot is planned to determine the specific path and steps of the maintenance. The specific path and steps of the maintenance include the moving trajectory of the welding robot and the sequence of welding points. When planning the path, the efficiency, safety and working space limitations of the maintenance should be taken into consideration. B4. Based on the specific paths and steps of maintenance, the motion control design of the welding robot is carried out. The motion control includes the robot's trajectory planning, motion planning and trajectory tracking; B5. Move and perform maintenance operations on the welding robot according to the established maintenance path to ensure that the welding robot can move along the planned path and complete maintenance tasks.
7. A vision-based camera welding control system according to claim 6, characterized in that: When the control module performs the repair welding operation, the control module includes the following steps: C1. The control module needs to receive input information of the repair welding path, and the input information of the repair welding path includes the position coordinates of the repair welding and the shape and size of the weld to be repaired; C2. After receiving the input information of the repair welding path, the control module converts the repair welding path into a specific motion path that the welding robot can execute; the motion path includes converting the target path into a path in the robot coordinate system, and considering the robot kinematics and obstacle avoidance factors; C3, the control module generates motion control instructions for the welding robot according to the results of path planning; the motion control instructions involve the trajectory, speed and acceleration control of each joint of the robot, as well as the posture control of the welding gun or welding head. The motion control instructions are sent to the welding robot through the interface of the robot control system; C4. During the repair welding operation, the control module needs to monitor the movement and status of the welding robot in real time, including detecting the actual position and posture of the robot and the feedback information during the welding process. Through sensors and feedback control systems, the various parameters of the welding process can be obtained in real time, and the robot can be adjusted and controlled in real time. C5. When the repair welding operation is completed, the control module needs to send an instruction to end the action so that the welding robot can safely exit the working state.
8. A vision-based camera welding control system according to claim 7, characterized in that: The human-computer interaction module includes a visual interface that displays monitoring results, welding parameters and alarm information in the form of graphics or text. The visual interface uses charts, tables and images to enable operators to intuitively understand and analyze data.
