Arc welding robot welding monitoring system based on visual sensing

Abstract

The invention discloses an arc welding robot welding monitoring system based on visual sensing, which belongs to the technical field of welding monitoring and comprises a molten pool state evaluation module, a welding defect risk prediction module, a welding parameter adjustment module and a monitoring strategy evolution module. The defect risk prediction module predicts the defect risk in real time by combining the molten pool state, the process parameters and the splash characteristics, the welding parameter adjustment module dynamically adjusts the welding parameters according to the risk result, and the monitoring strategy evolution module realizes strategy autonomous evolution based on the intervention effect, and continuously improves the system performance. According to the system, through cooperative work of multiple modules, all-around real-time monitoring and accurate regulation and control in the welding process are achieved, problems can be found in time in the welding process, welding parameters can be rapidly adjusted, the defect rate is remarkably reduced, the welding quality stability and the production efficiency are improved, and powerful support is provided for optimization of the welding technology.

Description

Technical Field

[0001] This invention relates to the field of welding monitoring technology, and in particular to a welding monitoring system for arc welding robots based on vision sensing. Background Technology

[0002] In the industrial manufacturing sector, arc welding robots have become key equipment in modern welding production due to their advantages of high efficiency, precision, and repeatability. They are widely used in many industries such as automobile manufacturing, aerospace, and shipbuilding. As welding technology continues to develop towards intelligence and automation, more stringent requirements are being placed on the monitoring and quality control of the arc welding robot welding process.

[0003] Traditional arc welding robot welding monitoring mainly relies on human experience and simple sensor feedback, such as monitoring changes in electrical parameters during the welding process through current and voltage sensors to determine whether the welding status is normal. However, the welding process is an extremely complex physicochemical process. Although existing technologies have made some progress in sensor fusion and data processing, there are still many shortcomings in the comprehensive real-time monitoring and precise control of the welding process.

[0004] During the welding process, although most systems can detect problems that occur, they can only take remedial measures after the welding is completed. It is difficult to make timely and rapid adjustments during the welding process. The subsequent welding parameters can only be modified after the current welding process is completed. This not only increases production costs, but may also cause irreversible defects in the welded parts, affecting the overall welding quality. Moreover, once the monitoring strategies and evaluation models of existing monitoring systems are set, it is difficult to dynamically optimize and adjust them according to actual working conditions and historical data. This makes it difficult to adapt to the complex and ever-changing welding production environment and meet the high requirements of modern industry for welding quality stability and production efficiency. Summary of the Invention

[0005] To address the shortcomings mentioned in the background technology, we propose a vision-sensing-based arc welding robot welding monitoring system.

[0006] The technical solution mainly consists of: a vision-sensing-based arc welding robot welding monitoring system, including:

[0007] The molten pool condition assessment module is configured to acquire molten pool morphology parameters through a visual sensing device, acquire molten pool temperature parameters through an infrared thermometer, and combine them with the theoretical melting point temperature of the welding material to construct a comprehensive molten pool condition assessment model, thereby achieving a quantitative assessment of the molten pool welding condition and providing basic data for subsequent defect risk prediction.

[0008] The welding defect risk prediction module is configured to receive the comprehensive evaluation results of the molten pool state output by the molten pool state evaluation module, obtain welding process parameters and spatter particle characteristic parameters, construct a welding defect risk prediction model, and realize real-time prediction of welding defect risks.

[0009] The welding parameter adjustment module is configured to acquire the current welding parameters, the theoretical melting point temperature of the welding material, and the actual temperature of the molten pool. Combined with the defect risk prediction results output by the welding defect risk prediction module and the set defect risk threshold, a welding parameter adjustment model is constructed to realize the real-time adjustment of welding parameters.

[0010] The monitoring strategy evolution module is configured to obtain the defect risk prediction results, melt pool status assessment results and successful intervention counts before and after the adjustment, build a monitoring strategy evolution model, realize the autonomous evolution of the monitoring strategy, and improve system performance.

[0011] Preferably, the molten pool condition assessment module includes a molten pool morphology acquisition unit, a molten pool thermal state acquisition unit, and a molten pool condition comprehensive assessment unit:

[0012] The molten pool morphology acquisition unit is configured to acquire geometric morphology data of the molten pool through a visual sensing device, preprocess the acquired data, extract the molten pool area, shape feature parameters and brightness parameters, and output the processed molten pool morphology feature data to the molten pool state comprehensive evaluation unit.

[0013] The molten pool thermal state acquisition unit is configured to acquire temperature data of the molten pool through an infrared thermometer, preprocess the acquired data, extract the temperature characteristic parameters of the molten pool, and output the processed molten pool thermal state characteristic data to the molten pool state comprehensive evaluation unit to provide thermal state data for molten pool state evaluation.

[0014] The molten pool condition comprehensive evaluation unit is configured to receive molten pool morphology feature data output by the molten pool morphology acquisition unit and molten pool thermal state feature data output by the molten pool thermal state acquisition unit. By weighted fusion of molten pool morphology feature parameters and thermal state feature parameters, and combined with the theoretical melting point temperature of welding materials, a comprehensive evaluation model of molten pool condition is constructed to achieve quantitative evaluation of molten pool welding condition. By performing nonlinear mapping between the spatial characteristics of molten pool morphology parameters and the thermal characteristics of temperature parameters, multi-dimensional parameters are fused into a single evaluation index, providing standardized input data for defect risk prediction.

[0015] Preferably, the molten pool morphology acquisition unit specifically comprises:

[0016] Image data of the molten pool is acquired by a visual sensing device. Image preprocessing technology is used to denoise, enhance and segment the acquired image data to extract the area, shape feature parameters and brightness parameters of the molten pool. The processed molten pool morphological feature data is output to the molten pool state comprehensive evaluation unit to provide accurate morphological feature data for molten pool state evaluation.

[0017] Preferably, the welding defect risk prediction module includes a molten pool state correlation unit, a process parameter correlation unit, a spatter characteristic correlation unit, and a comprehensive defect risk prediction unit:

[0018] The molten pool state correlation unit is configured to receive the molten pool state comprehensive evaluation result output by the molten pool state comprehensive evaluation unit, construct the correlation model between molten pool state and welding defects by transforming the inverse of the molten pool state comprehensive evaluation index, realize the nonlinear impact assessment of molten pool state on welding defect risk, and output the molten pool state correlation result to the defect risk comprehensive prediction unit.

[0019] The process parameter association unit is configured to acquire welding process parameter data, construct a correlation model between welding process parameters and welding defects, and realize the impact assessment of welding process parameters on welding defect risk.

[0020] The spatter feature association unit is configured to acquire spatter particle feature parameter data, construct a linear function model by the ratio of the number of spatter particles to the maximum allowable number, realize the linearized impact assessment of spatter features on welding defect risk, and output the spatter feature association results to the defect risk comprehensive prediction unit.

[0021] The comprehensive defect risk prediction unit is configured to multiply and fuse the results of the molten pool state association, the process parameter association, and the spatter characteristic association to construct a comprehensive welding defect risk prediction model, realize the quantitative assessment of welding defect risk, and output the prediction results to the welding parameter adjustment module to provide a basis for welding parameter adjustment.

[0022] Preferably, the process parameter association unit specifically comprises:

[0023] Welding process parameter data is acquired, and an exponential function model is constructed by comparing the welding process parameters with the theoretical optimal parameters to assess the impact of welding process parameters on welding defect risk. The correlation results of the process parameters are then output to the comprehensive defect risk prediction unit, providing accurate data on the impact of process parameters for welding defect risk prediction.

[0024] Preferably, the welding parameter adjustment module includes a defect risk analysis unit, a molten pool state analysis unit, a parameter adjustment strategy generation unit, and a parameter adjustment execution unit.

[0025] The defect risk analysis unit is configured to receive the defect risk prediction results output by the welding defect risk prediction module, construct a defect risk analysis model by the ratio of the defect risk prediction index to a set threshold, realize the classification of welding defect risk levels and root cause analysis, and output the defect risk analysis results to the parameter adjustment strategy generation unit.

[0026] The molten pool condition analysis unit is configured to receive the comprehensive evaluation results of the molten pool condition output by the molten pool condition assessment module, construct a molten pool condition analysis model by the difference between the actual temperature of the molten pool and the theoretical melting point temperature of the welding material, realize the identification of abnormalities in the molten pool condition and the analysis of causes, and output the molten pool condition analysis results to the parameter adjustment strategy generation unit.

[0027] Preferably, the parameter adjustment strategy generation unit is configured to receive the defect risk analysis results and the molten pool state analysis results, construct a welding parameter adjustment strategy model by weighted fusion of the defect risk analysis results, the molten pool state analysis results and the current welding parameters, realize the quantitative calculation of the welding parameter adjustment amount, and output the welding parameter adjustment strategy to the parameter adjustment execution unit;

[0028] The parameter adjustment execution unit is configured to receive the welding parameter adjustment strategy output by the parameter adjustment strategy generation unit, output welding parameter adjustment instructions, realize real-time adjustment of welding parameters, and output the welding parameter adjustment effect data to the monitoring strategy evolution module to provide data support for the monitoring strategy evolution.

[0029] Preferably, the parameter adjustment execution unit specifically comprises:

[0030] The system receives the welding parameter adjustment strategy output by the parameter adjustment strategy generation unit, outputs the welding parameter adjustment command, uses real-time control technology to realize the real-time adjustment of welding parameters, and outputs the welding parameter adjustment effect data to the monitoring strategy evolution module to provide accurate adjustment effect data for the monitoring strategy evolution.

[0031] Preferably, the monitoring strategy evolution module includes:

[0032] The intervention effect evaluation unit is configured to receive the welding parameter adjustment effect data and welding quality data, construct an intervention effect evaluation model by transforming the ratio of the defect risk prediction index before and after adjustment, the ratio of the comprehensive evaluation index of the molten pool state before and after adjustment, and the logarithmic transformation of the number of successful interventions, so as to realize the quantitative evaluation of the welding parameter adjustment effect, and output the intervention effect evaluation results to the strategy evolution analysis unit.

[0033] The strategy evolution analysis unit is configured to receive the intervention effect evaluation results, defect risk prediction results, and melt pool status comprehensive evaluation results. It constructs a monitoring strategy evolution analysis model by comparing the intervention effect evaluation results with the set threshold, realizes the evolutionary requirements analysis of the monitoring strategy, and outputs the monitoring strategy evolutionary requirements analysis results to the strategy evolution execution unit.

[0034] The strategy evolution execution unit is configured to receive the monitoring strategy evolution requirement analysis results output by the strategy evolution analysis unit, update the parameters of the molten pool state assessment model, defect risk prediction model, and welding parameter adjustment model, realize the autonomous evolution of the monitoring strategy, improve system performance, and feed back the updated model parameters to the molten pool state assessment module, welding defect risk prediction module, and welding parameter adjustment module to achieve continuous improvement in system performance.

[0035] Compared with the prior art, the beneficial effects of the present invention are:

[0036] This invention constructs a complete closed-loop control system for welding quality. Through the collaborative work of four core modules—molten pool state assessment, defect risk prediction, parameter adjustment, and strategy evolution—it achieves comprehensive real-time monitoring and precise control of the welding process. The data interaction between the modules is smooth, forming a complete closed loop from data acquisition to strategy optimization, which effectively improves welding quality and stability. This closed-loop control mechanism can promptly detect and correct abnormalities in the welding process, significantly reduce the welding defect rate, and improve production efficiency, providing strong support for the optimization of welding processes.

[0037] In this invention, the welding defect risk prediction module uses a multi-factor correlation and fusion prediction algorithm to comprehensively consider the influence of molten pool state, process parameters, and spatter characteristics on welding defects, achieving real-time and accurate risk prediction. This prediction result provides an accurate basis for the welding parameter adjustment module, enabling it to dynamically adjust parameters such as welding current, speed, and gas flow rate according to the risk level and the abnormality level of the molten pool state. The parameter adjustment strategy is generated based on a weighted fusion algorithm, which can quantify the adjustment amount, ensuring the scientific nature and effectiveness of parameter optimization. Through this closed-loop control mechanism, deviations in the welding process can be corrected in a timely manner, effectively reducing the welding defect rate and improving the stability of welding quality.

[0038] In this invention, an adaptive optimization capability is achieved through a monitoring strategy evolution module, constructing a continuously improving technical closed loop. The intervention effect evaluation unit constructs a quantitative evaluation model by comparing the defect risk and molten pool state index before and after adjustment, combined with the number of successful interventions, providing data support for strategy evolution. The strategy evolution analysis unit drives the parameters of the three core models to update autonomously based on the evolution coefficient calculation results, enabling the system to dynamically adjust the evaluation threshold and prediction algorithm according to actual working conditions. This closed-loop evolution mechanism not only improves the stability of welding quality but also continuously optimizes the control strategy through historical data accumulation, ultimately achieving dual optimization of welding efficiency and defect rate. The system as a whole exhibits high reliability, strong adaptability, and continuous evolution characteristics, providing a complete technical solution for intelligent welding production. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the overall operation of the present invention.

[0040] Figure 2 This is a flowchart illustrating the operation between modules in this invention. Detailed Implementation

[0041] The present invention will now be described in further detail with reference to the accompanying drawings and preferred embodiments.

[0042] Example 1, refer to Figure 1-2 As shown, the arc welding robot welding monitoring system based on vision sensing includes a molten pool condition assessment module, a welding defect risk prediction module, a welding parameter adjustment module, and a monitoring strategy evolution module.

[0043] Specifically, the molten pool condition assessment module collects molten pool morphology parameters and molten pool temperature parameters. By combining the theoretical melting point temperature of the welding material, it constructs a comprehensive assessment model of the molten pool condition to achieve a quantitative assessment of the welding condition of the molten pool. The welding defect risk prediction module receives the comprehensive assessment results of the molten pool condition output by the molten pool condition assessment module, and collects welding process parameters and spatter particle characteristic parameters to construct a welding defect risk prediction model. It comprehensively considers the impact of molten pool condition, process parameters and spatter characteristics on welding defect risk to achieve real-time prediction of welding defect risk.

[0044] The welding parameter adjustment module collects current welding parameters, theoretical melting point temperature of welding materials, and actual temperature of the molten pool. Combining the defect risk prediction results and the set defect risk threshold, it constructs a welding parameter adjustment model to adjust the welding parameters in real time. The monitoring strategy evolution module collects defect risk prediction results, molten pool status assessment results, and the number of successful interventions before and after the adjustment, constructs a monitoring strategy evolution model, realizes the autonomous evolution of the monitoring strategy, and improves system performance.

[0045] This embodiment constructs a complete closed-loop control system for welding quality. Through the collaborative work of four core modules—molten pool state assessment, defect risk prediction, parameter adjustment, and strategy evolution—real-time monitoring of the welding process, defect risk prediction, automatic parameter adjustment, and autonomous strategy evolution are achieved. The coordinated operation of each module forms a complete closed loop from data acquisition to strategy optimization, effectively improving welding quality and stability, reducing welding defect rate, and increasing production efficiency.

[0046] Example 2, refer to Figure 1-2 As shown, the molten pool condition assessment module adopts a three-level architecture of multi-source data acquisition, feature extraction, and fusion assessment to achieve accurate quantification of the molten pool welding condition. The molten pool morphology acquisition unit acquires dynamic image data of the molten pool through a high-speed visual sensing device, and uses an adaptive threshold segmentation algorithm to denoise, enhance, and segment the image to extract morphological feature parameters such as molten pool area, roundness, weld width, and average brightness.

[0047] The molten pool thermal state acquisition unit collects temperature distribution data of the molten pool using an infrared thermometer. It preprocesses the temperature data using a Gaussian filtering algorithm to extract thermal state characteristic parameters such as average molten pool temperature, temperature gradient, and temperature uniformity. The processed molten pool thermal state characteristic data is then normalized and output to the molten pool state comprehensive evaluation unit. This unit uses a weighted fusion algorithm to fuse the molten pool morphology characteristic parameters with the thermal state characteristic parameters, and combines this with the theoretical melting point temperature of the welding material to construct a comprehensive evaluation model of the molten pool state. This achieves a quantitative evaluation of the molten pool welding state. The specific calculation formula is as follows:

[0048] ;

[0049] In the formula, S is the calculated comprehensive evaluation index of the molten pool condition, which reflects the quality of the molten pool welding condition; A is the actual area of ​​the molten pool; C is the roundness of the molten pool; L is the actual width of the weld; B is the average brightness of the molten pool; T is the actual temperature of the molten pool; and T0 is the theoretical melting point temperature of the welding material.

[0050] The molten pool condition comprehensive assessment unit outputs the molten pool condition comprehensive assessment results to the welding defect risk prediction module, providing standardized input data for defect risk prediction;

[0051] This embodiment achieves accurate quantitative assessment of the welding state of the molten pool through the collaborative work of the molten pool morphology acquisition unit, the molten pool thermal state acquisition unit, and the molten pool state comprehensive evaluation unit. The system uses multi-source data acquisition technology to acquire the morphology and temperature data of the molten pool, extracts feature parameters through advanced image processing and signal processing algorithms, and constructs an evaluation model using a weighted fusion algorithm, thus achieving accurate assessment of the molten pool state. The system's evaluation results can provide accurate basic data for subsequent defect risk prediction, thereby improving the overall performance of the system.

[0052] In this embodiment, the molten pool condition assessment module can collect molten pool morphology and temperature data in real time. Through multi-parameter fusion and nonlinear mapping, it achieves accurate assessment of the molten pool condition. By comparing preset thresholds, it can promptly detect abnormalities in the molten pool condition, providing a basis for intervention in the welding process. The standardized output of the comprehensive molten pool condition assessment results facilitates data interaction with subsequent modules, improving the overall performance of the system. The adaptive threshold segmentation algorithm and Gaussian filtering algorithm can effectively improve the accuracy and reliability of data processing, while the weighted fusion algorithm can fully utilize various feature parameters to improve the accuracy and robustness of the assessment model.

[0053] Example 3, refer to Figure 1-2 As shown, the welding defect risk prediction module adopts a three-level architecture of multi-factor correlation, fusion prediction, and risk classification to achieve real-time and accurate prediction of welding defect risk. The molten pool state correlation unit uses a nonlinear mapping algorithm to convert the comprehensive evaluation index of molten pool state into defect risk influence factors, realizing the nonlinear influence assessment of molten pool state on welding defect risk, and outputs the molten pool state correlation results to the defect risk comprehensive prediction unit. The process parameter correlation unit uses an exponential function model to construct the correlation between welding process parameters and welding defects, realizing the exponential influence assessment of welding process parameters on welding defect risk, and outputs the process parameter correlation results to the defect risk comprehensive prediction unit.

[0054] The spatter feature correlation unit uses a linear function model to construct the correlation between the number of spatter particles and welding defects, realizing a linearized assessment of the impact of spatter features on welding defect risk. The defect risk comprehensive prediction unit uses a product fusion algorithm to fuse the correlation results of molten pool state, process parameters, and spatter features to construct a comprehensive welding defect risk prediction model, realizing a quantitative assessment of welding defect risk. The specific calculation formula is as follows:

[0055] ;

[0056] In the formula, R is the welding defect risk prediction index, which reflects the degree of welding defect risk; S is the comprehensive evaluation index of the molten pool state; V is the actual welding speed, which is directly collected by the robot control system; V0 is the theoretical optimal welding speed, which is calculated based on the welding material and thickness; and F0 is the maximum allowable number of spatter particles.

[0057] In this embodiment, the real-time and accurate prediction of welding defect risk is achieved through the collaborative work of the molten pool state association unit, process parameter association unit, spatter feature association unit, and defect risk comprehensive prediction unit. By comprehensively considering the influence of molten pool state, welding process parameters, and spatter features on welding defect risk, a prediction model is constructed through multi-factor association and fusion prediction algorithms, which realizes accurate prediction of welding defect risk. The prediction results can provide an accurate basis for subsequent parameter adjustment and improve the overall performance of the system.

[0058] After calculating the welding defect risk prediction index, the comprehensive defect risk prediction unit outputs the defect risk prediction results to the welding parameter adjustment module, providing a basis for adjusting the welding parameters.

[0059] Example 4, refer to Figure 1-2 As shown, the welding parameter adjustment module includes a defect risk analysis unit, a molten pool state analysis unit, a parameter adjustment strategy generation unit, and a parameter adjustment execution unit. The defect risk analysis unit uses a risk grading algorithm to convert the defect risk prediction index into a risk level and root cause analysis results. The molten pool state analysis unit uses an anomaly detection algorithm to convert the comprehensive molten pool state evaluation index into a molten pool state anomaly level and cause analysis results, and outputs the molten pool state analysis results to the parameter adjustment strategy generation unit. The parameter adjustment strategy generation unit uses a weighted fusion algorithm to fuse the defect risk analysis results, molten pool state analysis results, and current welding parameters to construct a welding parameter adjustment strategy model, realizing the quantitative calculation of the welding parameter adjustment amount. The specific calculation formula is as follows:

[0060] ;

[0061] In the formula, △I is the welding current adjustment amount, I0 is the current welding current, R is the welding defect risk prediction index, R0 is the defect risk threshold, which is determined through historical data statistical analysis, T0 is the theoretical melting point temperature of the welding material, and T is the actual temperature of the molten pool.

[0062] In the formula, The risk levels output by the defect risk analysis unit are specifically classified as follows:

[0063] when > When the risk threshold is high, the risk level is high, which means that the current welding defect risk prediction index is much higher than the defect risk threshold, which is likely to lead to serious welding defects. At the same time, the root cause analysis result is that the defect risk is too high, which may be due to unreasonable welding parameters, unstable molten pool state or excessive spatter.

[0064] When the risk level is high, the emergency intervention mechanism is immediately triggered to perform significant parameter adjustments, including calculating the maximum adjustment amount using the formula for adjusting the welding current, adjusting the welding speed and gas flow accordingly, and triggering the audible and visual alarm system to notify the operator to conduct on-site inspection and confirmation.

[0065] When the low-risk lower limit is ≤ When the risk level is ≤ high risk threshold, the risk level is low risk, indicating that the current welding defect risk prediction index is close to the defect risk threshold and the welding process is stable.

[0066] When the risk level is low, perform minor parameter adjustments, including calculating the maximum adjustment amount of welding current, making corresponding adjustments to welding speed and gas flow rate, and triggering system prompts to notify operators to pay attention and track the process.

[0067] when When the risk level is below the low risk threshold, the risk level is extremely low, indicating that the current welding defect risk prediction index is far below the defect risk threshold and the welding process is very stable.

[0068] When the risk level is extremely low, the intervention mechanism is not triggered, and the current welding parameters remain unchanged. Specifically, this includes continuing to collect data on the molten pool state, welding process parameters, and spatter characteristics, updating the welding defect risk prediction index in real time, and recording the current welding parameters, molten pool state, and defect risk prediction results in the system database to provide data support for subsequent analysis and optimization.

[0069] Furthermore, The molten pool state anomaly level output by the molten pool state analysis unit is classified as follows:

[0070] when > When the preset low temperature is reached, the abnormality level of the molten pool is severe, indicating that the actual temperature of the molten pool is much lower than the theoretical melting point of the welding material, which can easily lead to welding defects such as incomplete fusion and incomplete penetration. At the same time, the cause analysis result is that the molten pool temperature is too low, which may be due to insufficient welding current, excessive welding speed, or insufficient shielding gas flow. It is necessary to increase the welding current, increase the molten pool temperature, and decrease the welding speed to increase the heat input of the molten pool.

[0071] When the lower limit of high temperature threshold is ≤ When the temperature is ≤ the preset low temperature value, the abnormal level of the molten pool is normal, which means that the actual temperature of the molten pool is close to the theoretical melting point temperature of the welding material and the welding process is stable. At this time, keep the current welding parameters unchanged and continue to monitor the molten pool status to ensure that the molten pool status is stable.

[0072] when When the temperature is below the lower limit of the high temperature, the abnormality level of the molten pool is severe, indicating that the actual temperature of the molten pool is much higher than the theoretical melting point of the welding material. This can easily lead to welding defects such as burn-through and excessive spatter. The cause analysis results indicate that the molten pool temperature is too high, which may be due to excessive welding current, slow welding speed, or excessive shielding gas flow. In this case, it is necessary to reduce the welding current, lower the molten pool temperature, and increase the welding speed to reduce the heat input to the molten pool.

[0073] Among them, the high-risk threshold, low-risk lower limit, high-temperature lower limit threshold and low-temperature preset value need to be set and adjusted according to the actual material's melting point, thermal conductivity, linear expansion coefficient and other characteristics;

[0074] Furthermore, when severe abnormalities and high risks occur simultaneously in the molten pool, the primary goal is to prevent welding defects. Priority is given to ensuring that the molten pool temperature returns to the normal range. Under the premise of ensuring welding quality, welding production efficiency is improved through parameter optimization at extremely low risk levels. After each current adjustment, the molten pool temperature and defect risk are quickly reassessed to form a closed-loop control.

[0075] The parameter adjustment execution unit adjusts the welding parameters in real time using real-time control technology and outputs the welding parameter adjustment effect data to the monitoring strategy evolution module.

[0076] Example 5, refer to Figure 1-2 As shown, the monitoring strategy evolution module includes an intervention effect evaluation unit, a strategy evolution analysis unit, and a strategy evolution execution unit. The intervention effect evaluation unit constructs an intervention effect evaluation model by combining parameters such as the ratio of the defect risk prediction index R before and after adjustment and the ratio of the comprehensive evaluation index S of the molten pool state before and after adjustment, thereby achieving a quantitative evaluation of the welding parameter adjustment effect. The strategy evolution analysis unit constructs a monitoring strategy evolution analysis model by comparing the data output by the intervention effect evaluation unit with a set threshold, thereby achieving the evolutionary demand analysis of the monitoring strategy. The specific calculation formula is as follows:

[0077] ;

[0078] In the formula, E is the monitoring strategy evolution coefficient, reflecting the degree of evolution of the monitoring strategy, R1 is the adjusted defect risk prediction index, R2 is the original defect risk prediction index, S1 is the adjusted melt pool state comprehensive evaluation index, S2 is the original melt pool state comprehensive evaluation index, and N is the number of successful interventions.

[0079] By calculating the evolution coefficient E of the monitoring strategy, the evolution level and optimization effect of the monitoring strategy can be intuitively reflected. The magnitude of the strategy evolution is judged by the size of the E value, and the larger the E value, the higher the degree of evolution.

[0080] Meanwhile, by integrating before-and-after comparison data of the defect risk prediction index and the comprehensive evaluation index of the molten pool condition, as well as the number of successful interventions, the system provides accurate decision-making basis for the strategy evolution execution unit. This drives the autonomous updating of parameters in the three core models of molten pool condition assessment, defect risk prediction, and welding parameter adjustment, ultimately forming a closed loop of data collection, effect evaluation, strategy evolution, and parameter optimization. This achieves a dynamic balance between welding quality stability and production efficiency, and promotes the continuous improvement of welding system performance.

[0081] It should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should also be within the scope of protection of this invention.

Claims

1. A vision-sensing-based arc welding robot welding monitoring system, characterized in that: include: The molten pool condition assessment module is configured to acquire molten pool morphology parameters through a visual sensing device, acquire molten pool temperature parameters through an infrared thermometer, and combine them with the theoretical melting point temperature of the welding material to construct a comprehensive molten pool condition assessment model, thereby achieving a quantitative assessment of the molten pool welding condition and providing basic data for subsequent defect risk prediction. The welding defect risk prediction module is configured to receive the comprehensive evaluation results of the molten pool state output by the molten pool state evaluation module, obtain welding process parameters and spatter particle characteristic parameters, construct a welding defect risk prediction model, and realize real-time prediction of welding defect risks. The welding parameter adjustment module is configured to acquire the current welding parameters, the theoretical melting point temperature of the welding material, and the actual temperature of the molten pool. Combined with the defect risk prediction results output by the welding defect risk prediction module and the set defect risk threshold, a welding parameter adjustment model is constructed to realize the real-time adjustment of welding parameters. The monitoring strategy evolution module is configured to obtain the defect risk prediction results, melt pool status assessment results and the number of successful interventions before and after the adjustment, construct the monitoring strategy evolution model, and realize the autonomous evolution of the monitoring strategy.

2. The arc welding robot welding monitoring system based on vision sensing according to claim 1, characterized in that: The molten pool condition assessment module includes a molten pool morphology acquisition unit, a molten pool thermal condition acquisition unit, and a molten pool condition comprehensive assessment unit. The molten pool morphology acquisition unit is configured to acquire geometric morphology data of the molten pool through a visual sensing device, preprocess the acquired data, extract the molten pool area, shape feature parameters and brightness parameters, and output the processed molten pool morphology feature data to the molten pool state comprehensive evaluation unit. The molten pool thermal state acquisition unit is configured to acquire temperature data of the molten pool through an infrared thermometer, preprocess the acquired data, extract the temperature characteristic parameters of the molten pool, and output the processed molten pool thermal state characteristic data to the molten pool state comprehensive evaluation unit to provide thermal state data for molten pool state evaluation. The molten pool condition comprehensive evaluation unit is configured to receive molten pool morphology feature data output by the molten pool morphology acquisition unit and molten pool thermal state feature data output by the molten pool thermal state acquisition unit. By weighted fusion of molten pool morphology feature parameters and thermal state feature parameters, and combined with the theoretical melting point temperature of welding materials, a comprehensive evaluation model of molten pool condition is constructed to achieve quantitative evaluation of molten pool welding condition. By performing nonlinear mapping between the spatial characteristics of molten pool morphology parameters and the thermal characteristics of temperature parameters, multi-dimensional parameters are fused into a single evaluation index, providing standardized input data for defect risk prediction.

3. The arc welding robot welding monitoring system based on vision sensing according to claim 2, characterized in that: The molten pool morphology acquisition unit is specifically: Image data of the molten pool is acquired by a visual sensing device. Image preprocessing technology is used to denoise, enhance and segment the acquired image data to extract the area, shape feature parameters and brightness parameters of the molten pool. The processed molten pool morphological feature data is output to the molten pool state comprehensive evaluation unit to provide accurate morphological feature data for molten pool state evaluation.

4. The arc welding robot welding monitoring system based on vision sensing according to claim 3, characterized in that: The welding defect risk prediction module includes a molten pool state correlation unit, a process parameter correlation unit, a spatter characteristic correlation unit, and a comprehensive defect risk prediction unit. The molten pool state correlation unit is configured to receive the molten pool state comprehensive evaluation result output by the molten pool state comprehensive evaluation unit, construct the correlation model between molten pool state and welding defects by transforming the inverse of the molten pool state comprehensive evaluation index, realize the nonlinear impact assessment of molten pool state on welding defect risk, and output the molten pool state correlation result to the defect risk comprehensive prediction unit. The process parameter association unit is configured to acquire welding process parameter data, construct a correlation model between welding process parameters and welding defects, and realize the impact assessment of welding process parameters on welding defect risk. The spatter feature association unit is configured to acquire spatter particle feature parameter data, construct a linear function model by the ratio of the number of spatter particles to the maximum allowable number, realize the linearized impact assessment of spatter features on welding defect risk, and output the spatter feature association results to the defect risk comprehensive prediction unit. The comprehensive defect risk prediction unit is configured to multiply and fuse the results of the molten pool state association, the process parameter association, and the spatter characteristic association to construct a comprehensive welding defect risk prediction model, realize the quantitative assessment of welding defect risk, and output the prediction results to the welding parameter adjustment module to provide a basis for welding parameter adjustment.

5. The arc welding robot welding monitoring system based on vision sensing according to claim 4, characterized in that: The process parameter association unit is specifically: Welding process parameter data is acquired, and an exponential function model is constructed by comparing the welding process parameters with the theoretical optimal parameters to assess the impact of welding process parameters on welding defect risk. The correlation results of the process parameters are then output to the comprehensive defect risk prediction unit, providing accurate data on the impact of process parameters for welding defect risk prediction.

6. The arc welding robot welding monitoring system based on vision sensing according to claim 5, characterized in that: The welding parameter adjustment module includes a defect risk analysis unit, a molten pool state analysis unit, a parameter adjustment strategy generation unit, and a parameter adjustment execution unit. The defect risk analysis unit is configured to receive the defect risk prediction results output by the welding defect risk prediction module, construct a defect risk analysis model by the ratio of the defect risk prediction index to a set threshold, realize the classification of welding defect risk levels and root cause analysis, and output the defect risk analysis results to the parameter adjustment strategy generation unit. The molten pool condition analysis unit is configured to receive the comprehensive evaluation results of the molten pool condition output by the molten pool condition assessment module, construct a molten pool condition analysis model by the difference between the actual temperature of the molten pool and the theoretical melting point temperature of the welding material, realize the identification of abnormalities in the molten pool condition and the analysis of causes, and output the molten pool condition analysis results to the parameter adjustment strategy generation unit.

7. The arc welding robot welding monitoring system based on vision sensing according to claim 6, characterized in that: The parameter adjustment strategy generation unit is configured to receive the defect risk analysis results and the molten pool state analysis results, construct a welding parameter adjustment strategy model by weighted fusion of the defect risk analysis results, the molten pool state analysis results and the current welding parameters, realize the quantitative calculation of the welding parameter adjustment amount, and output the welding parameter adjustment strategy to the parameter adjustment execution unit. The parameter adjustment execution unit is configured to receive the welding parameter adjustment strategy output by the parameter adjustment strategy generation unit, output welding parameter adjustment instructions, realize real-time adjustment of welding parameters, and output the welding parameter adjustment effect data to the monitoring strategy evolution module to provide data support for the monitoring strategy evolution.

8. The arc welding robot welding monitoring system based on vision sensing according to claim 7, characterized in that: The parameter adjustment execution unit is specifically: The system receives the welding parameter adjustment strategy output by the parameter adjustment strategy generation unit, outputs the welding parameter adjustment command, uses real-time control technology to realize the real-time adjustment of welding parameters, and outputs the welding parameter adjustment effect data to the monitoring strategy evolution module to provide accurate adjustment effect data for the monitoring strategy evolution.

9. The arc welding robot welding monitoring system based on vision sensing according to claim 8, characterized in that: The monitoring strategy evolution module includes: The intervention effect evaluation unit is configured to receive the welding parameter adjustment effect data and welding quality data, construct an intervention effect evaluation model by transforming the ratio of the defect risk prediction index before and after adjustment, the ratio of the comprehensive evaluation index of the molten pool state before and after adjustment, and the logarithmic transformation of the number of successful interventions, so as to realize the quantitative evaluation of the welding parameter adjustment effect, and output the intervention effect evaluation results to the strategy evolution analysis unit. The strategy evolution analysis unit is configured to receive the intervention effect evaluation results, defect risk prediction results, and melt pool status comprehensive evaluation results. It constructs a monitoring strategy evolution analysis model by comparing the intervention effect evaluation results with the set threshold, realizes the evolutionary requirements analysis of the monitoring strategy, and outputs the monitoring strategy evolutionary requirements analysis results to the strategy evolution execution unit. The strategy evolution execution unit is configured to receive the monitoring strategy evolution requirement analysis results output by the strategy evolution analysis unit, update the parameters of the molten pool state assessment model, defect risk prediction model, and welding parameter adjustment model, realize the autonomous evolution of the monitoring strategy, improve system performance, and feed back the updated model parameters to the molten pool state assessment module, welding defect risk prediction module, and welding parameter adjustment module to achieve continuous improvement in system performance.

Images