Butt weld quality control method based on fuzzy control

Abstract

A butt weld quality control method based on fuzzy control is used for analyzing and controlling welding quality in real time in the MIG / MAG butt welding process and comprises the steps that firstly, quality change and current adjustment are fuzzified, then fuzzified input is sent into a fuzzy model, then an operation result is defuzzified, and finally a current adjustment value is output. In this way, a neural network and fuzzy control method is adopted to replace manual detection and adjustment in the welding process. According to the method, a fuzzy control algorithm is effectively combined with another fuzzy control algorithm, real-time control over the welding process in the true sense is achieved, due to the strong learning ability of the neural network and the advantages of model control, the adaptive capacity is high, the welding conditions in most scenes can be dealt with, and the expected economic benefits are remarkable.

Description

Technical Field

[0001] The invention relates to the technical field of welding, in particular to a butt weld quality control method based on fuzzy control. Background Art

[0002] The robot MIG / MAG welding system uses the robot as a walking and posture control mechanism, and controls the arc start and end of welding, while the unified MIG / MAG welder is used for the modification and setting of welding parameters by manual settings. During the welding process, the robot sends an arc start command to the welder, and the robot starts to execute the walking command. The welder outputs the set current, heats the workpiece, and fuses the welding wire, fills it into the workpiece groove, and connects the workpieces. At the end of welding, the robot moves to the specified position and sends an arc extinguishing command to the welder to end the welding and form a stable welded joint.

[0003] During the welding process, due to factors such as processing and equipment errors in the workpiece and changes in the welding environment, the welder will adjust the welding current according to changes in arc length. However, such adjustments often rely on the internal welding procedures, and the decision-making information is single. When faced with complex changes in working conditions during the welding process, good welding quality cannot be obtained. The traditional manual control method has a single applicable scenario. First, it is necessary to analyze the welding signal related to the quality, establish an analysis model, and then adjust the current to achieve the purpose of quality control. This method has poor robustness and transferability and does not have learning ability. Different models can only be developed for different scenarios. Therefore, more weld information is needed to perform online quality control of MIG / MAG welding. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a butt weld quality control method based on fuzzy control by adopting neural network and fuzzy control methods to replace manual detection and adjustment in the welding process to obtain better welding quality.

[0005] The technical problem to be solved by the present invention is achieved by the following technical solution. The present invention is a butt weld quality control method based on fuzzy control, which first fuzzifies the quality change and current adjustment, then sends the fuzzified input to the fuzzy model, then defuzzifies the result of the operation, and finally outputs the value of the current adjustment, thereby using a neural network and fuzzy control method to replace manual detection and adjustment during the welding process.

[0006] The technical problem to be solved by the present invention can also be further achieved by the following technical solutions. For the above-mentioned butt weld quality control method based on fuzzy control, the specific operation steps of the method are as follows: (1) Conduct basic experiments in advance, and establish the relationship between weld groove information, welding current change and welding quality change index by training the neural network model, so as to obtain a better quality analysis model; (2) By manually adjusting the current change in the welding process, summarizing the experience of manual parameter adjustment, the current change input is fuzzified and calculated through the fuzzy control algorithm, and the fuzzy value is output. Then, the fuzzification is defuzzified and the specific current control value is output; (3) Preparation before welding First, prepare the sample with grooved and cleaned edges for welding, adjust the laser sensor and welding gun to be on the same axis, and connect the shielding gas required for welding in the normal way; (4) Before welding, check the import of the analysis model and the control model, the working condition of the current sensor, and ensure the normal operation of the system; (5) The arc is started normally according to the MIG / MAG welding method, and normal welding begins. At the same time, the laser sensor and current sensor begin to collect data and send the collected data to the analysis model. The analysis model outputs the current welding quality according to the calculation results of the neural network, and inputs the current adjustment amount to the control model. After the control model receives the data of the current adjustment amount, it outputs the change parameter of the current to the welding machine according to the fuzzy control algorithm. The welding machine outputs the changed current to achieve real-time control of the quality; (6) Step (5) is repeated from the beginning of welding until the welding is completed.

[0007] The technical problem to be solved by the present invention can also be further achieved by the following technical solutions. For the above-mentioned butt weld quality control method based on fuzzy control, the method uses a robot MIG / MAG welding system, which includes a robot body, a unified MIG / MAG welding machine, a laser sensor for collecting weld groove information, an online welding quality analysis system, an online welding quality control system, an industrial computer, a welding current sensor and a display. The laser sensor for collecting weld groove information is installed on the robot body mechanical arm, and its position is on the same axis as the welding gun position; During the welding process, the information collected by the laser sensor and welding current sensor that collect weld groove information is displayed in real time on the monitor of the industrial computer and sent to the welding quality online analysis system. The real-time situation of the current welding quality is obtained through the pre-built-in analysis model, and suggestions for correction are given. The output current change value is sent to the welding quality online control system in real time. The built-in control algorithm outputs the current change in real time to the unified MIG / MAG welding machine to control the current in real time.

[0008] The technical problem to be solved by the present invention can also be further achieved through the following technical solutions. For the butt weld quality control method based on fuzzy control described above, the laser sensor is also used to collect the change in weld surface quality after welding is completed, so as to calculate the specified quality change index.

[0009] The technical problem to be solved by the present invention can also be further achieved through the following technical solutions. For the butt weld quality control method based on fuzzy control described above, the collected weld groove information includes the width of the groove, the height difference on both sides of the groove, and the area calculated from the height difference.

[0010] Compared with the prior art, the present invention has the following beneficial effects: (1) The present invention utilizes the powerful learning ability and robustness of neural networks, has low training cost and strong migration ability, effectively establishes an effective connection between the groove information of the weld during the welding process and the change of welding current, and can analyze the welding quality in real time; (2) The present invention utilizes fuzzy control algorithm to effectively apply human experience to welding control. The model establishment cost is low and the subsequent expansion and upgrading capability is strong. At the same time, the input parameters of the model are the output parameters of the neural network. No human intervention and guidance are required throughout the process, which truly realizes the real-time intelligent control of the MIG / MAG welding process. (3) The present invention effectively combines the fuzzy control algorithm with the fuzzy control algorithm to achieve real-time control of the welding process in a true sense. Due to the powerful learning ability of the neural network and the advantages of model control, the system has strong adaptability and can cope with welding situations in a wide range of scenarios. The expected economic benefits are significant. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 The present invention is a control flow chart. DETAILED DESCRIPTION

[0012] 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 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.

[0013] Reference Figure 1, a butt weld quality control method based on fuzzy control, is a robot MIG / MAG butt welding quality online analysis and control method based on neural network and fuzzy control. The method uses a robot MIG / MAG welding system, including a robot body, a unified MIG / MAG welder, a laser sensor for collecting weld groove information, a welding quality online analysis system, a welding quality online control system, an industrial computer, a welding current sensor, and a display. Among them, the laser sensor for collecting weld groove information is installed on the robot body arm, and its position is on the same axis as the welding gun position, ensuring that the laser sensor can accurately collect the weld groove information during welding. The main groove information includes: the width of the groove, the height difference on both sides of the groove, and the area calculated from the height difference. Through the laser sensor, the surface quality change of the weld after welding can also be collected and the specified quality change index can be calculated; During the welding process, the information collected by the laser sensor and welding current sensor that collect weld groove information is displayed in real time on the monitor of the industrial computer and sent to the welding quality online analysis system. The real-time situation of the current welding quality is obtained through the pre-built-in analysis model, and suggestions for correction are given. The output current change value is sent to the welding quality online control system in real time. The built-in control algorithm outputs the current change in real time to the unified MIG / MAG welding machine, thereby realizing real-time control of the current and achieving the purpose of real-time quality control.

[0014] The specific steps of this method are: (1) Conduct basic experiments in advance, and establish the relationship between weld groove information, welding current change and welding quality change index by training the neural network model, so as to obtain a better quality analysis model; (2) By manually adjusting the current change in the welding process, summarizing the experience of manual parameter adjustment, the current change input is fuzzified and calculated through the fuzzy control algorithm, and the fuzzy value is output. Then, the fuzzification is defuzzified and the specific current control value is output; (3) Preparation before welding First, prepare the sample with grooved and cleaned edges for welding, adjust the laser sensor and welding gun to be on the same axis, and connect the shielding gas required for welding in the normal way; (4) Before welding, check the import of the analysis model and the control model, the working condition of the current sensor, and ensure the normal operation of the system; (5) The arc is started normally according to the MIG / MAG welding method, and normal welding begins. At the same time, the laser sensor and current sensor begin to collect data and send the collected data to the analysis model. The analysis model outputs the current welding quality according to the calculation results of the neural network, and inputs the current adjustment amount to the control model. After the control model receives the data of the current adjustment amount, it outputs the change parameter of the current to the welding machine according to the fuzzy control algorithm. The welding machine outputs the changed current to achieve real-time control of the quality; (6) Step 5 is repeated from the beginning of welding until the welding is completed.

[0015] The present invention provides a robot MIG / MAG butt welding quality online control method based on neural network and fuzzy control, which is used to analyze and control the welding quality in real time during the MIG / MAG butt welding process. According to the previous experiments, by collecting the changes in the weld groove information during the MIG / MAG welding process, such as the groove width, the height difference between the two grooves and the area difference of the upper surface misalignment of the groove, and obtaining the current and speed information at the same time, it is found that the changes in the above-mentioned collected information can directly affect the final weld surface formation. Through the neural network technology, the above-mentioned relationship can be well fitted to achieve online quality evaluation. Through experiments, it is found that the surface quality change of the weld is closely related to the welding current, so it is necessary to control the surface formation by changing the current change; the current adjustment is often done manually, which is difficult to achieve in the robot welding system. The fuzzy control algorithm can be used to directly fuzzify the manual adjustment strategy and output it to the welding machine to achieve quality control. Based on this, the robot MIG / MAG butt welding quality online analysis and control can be achieved.

[0016] The MIG / MAG butt welding quality online control system collects weld surface information related to welding quality during the welding process, establishes a neural network, outputs welding quality deviation status, and establishes the relationship between the quality deviation status and welding current. By converting manual adjustment experience into a fuzzy control model, the current deviation is used as the input of the model control model, and the current adjustment amount is used as the output, so as to achieve real-time control of the welding current and then real-time control of the welding quality.

[0017] With the help of MIG / MAG butt welding quality online analysis and control system, the real-time welding quality of MIG / MAG welding can be effectively improved, which can greatly save labor costs. At the same time, due to the learning and upgrading ability of neural networks, adaptation to different scenarios can be achieved at a lower cost.

[0018] The advantages of the butt weld quality control method based on fuzzy control provided by the present invention are: 1. The present invention replaces the manual MIG / MAG welding quality analysis method with a neural network method, uses a laser sensor to accurately collect the groove information of the weld, and collects the current change during the welding process. After the welding is completed, the surface quality of the weld is collected, and the indicators are specified. Through the neural network method, the groove information and current information are effectively associated with the quality situation, thereby realizing the real-time analysis of the welding quality during the robot MIG / MAG welding process. Compared with the method of collecting image information during the welding process, the pre-information of the welding process is fully utilized to analyze the welding quality, which effectively reduces the lag of information analysis. The association between groove information and quality is often associated with image images before. 2. The present invention replaces the manual MIG / MAG welding quality control method with a fuzzy control method, records the quality change situation analyzed by the welder and the idea of ​​current adjustment, fuzzifies the quality change situation and the current adjustment method, and then establishes a correlation through fuzzy reasoning, thereby replacing the fuzzy control method and realizing the quality control of robot MIG / MAG welding; 3. Connecting the above two methods. Traditional quality analysis and control methods often require human participation in the analysis or control process. The present invention effectively connects the above neural network and fuzzy control methods, associates the quality changes with the current changes, and finally realizes intelligent control analysis and control of the entire welding quality process, truly solving the problem of automatic quality control in the robot MIG / MAG welding process without human intervention.

Claims

1. A butt weld quality control method based on fuzzy control, characterized in that: First, the quality change and current adjustment are fuzzified, and then the fuzzified input is sent to the fuzzy model. Then the result of the operation is defuzzified, and finally the current adjustment value is output. In this way, neural network and fuzzy control methods are used to replace manual detection and adjustment in the welding process.

2. The butt weld quality control method based on fuzzy control according to claim 1 is characterized in that: The specific steps of this method are as follows: (1) Conduct basic experiments in advance, and establish the relationship between weld groove information, welding current change, and welding quality change indicators by training the neural network model, so as to obtain a better quality analysis model; (2) By manually adjusting the current change in the welding process, summarizing the experience of manual parameter adjustment, the current change input is fuzzified and calculated through the fuzzy control algorithm, and the fuzzy value is output. Then, the fuzzification is defuzzified and the specific current control value is output; (3) Preparation before welding First, prepare the sample with grooved and cleaned edges for welding, adjust the laser sensor and welding gun to be on the same axis, and connect the shielding gas required for the welding process in the normal way; (4) Before welding, check the import of the analysis model and the control model, the working condition of the current sensor, and ensure the normal operation of the system; (5) The arc is started normally according to the MIG / MAG welding method, and normal welding begins. At the same time, the laser sensor and current sensor begin to collect data and send the collected data to the analysis model. The analysis model outputs the current welding quality according to the calculation results of the neural network, and inputs the current adjustment amount to the control model. After the control model receives the data of the current adjustment amount, it outputs the change parameter of the current to the welding machine according to the fuzzy control algorithm. The welding machine outputs the changed current to achieve real-time control of the quality; (6) Step (5) is repeated from the beginning of welding until the welding is completed.

3. The butt weld quality control method based on fuzzy control according to claim 1 or 2, characterized in that: The method uses a robot MIG / MAG welding system, which includes a robot body, a unified MIG / MAG welder, a laser sensor for collecting weld groove information, an online welding quality analysis system, an online welding quality control system, an industrial computer, a welding current sensor and a display. The laser sensor for collecting weld groove information is installed on the robot body mechanical arm, and its position is on the same axis as the welding gun position. During the welding process, the information collected by the laser sensor and welding current sensor that collect weld groove information is displayed in real time on the monitor of the industrial computer and sent to the welding quality online analysis system. The real-time situation of the current welding quality is obtained through the pre-built-in analysis model, and suggestions for correction are given. The output current change value is sent to the welding quality online control system in real time. The built-in control algorithm outputs the current change in real time to the unified MIG / MAG welding machine to control the current in real time.

4. The butt weld quality control method based on fuzzy control according to claim 3 is characterized in that: The laser sensor is also used to collect the change in weld surface quality after welding is completed, so as to calculate a specified quality change index.

5. The butt weld quality control method based on fuzzy control according to claim 3 is characterized in that: The collected weld groove information includes the groove width, the height difference on both sides of the groove, and the area calculated from the height difference.

Images