A plate processing control system based on visual detection and a method thereof

Abstract

This application discloses a vision-based sheet metal processing control system and method, relating to the field of vision inspection. The system includes: predicting stress distribution during processing based on initial processing path data to obtain predicted stress distribution data; adjusting and evaluating processing parameters based on the predicted stress distribution data to obtain a preliminary processing control parameter set; acquiring real-time processing environment data; identifying environmental factors affecting the predicted stress distribution data to obtain temperature and humidity-related material deformation influencing factors; performing environmental adaptability correction on the preliminary processing control parameter set to obtain environmental processing control parameter data; constructing a real-time processing control model based on the environmental processing control parameter data to obtain a dynamic optimization control model for sheet metal processing; and deploying the dynamic optimization control model to the machine tool control equipment to execute sheet metal processing control. This application improves the efficiency of sheet metal processing control.

Description

Technical Field

[0001] This application relates to the field of visual inspection technology, and in particular to a visual inspection-based control system and method for sheet metal processing. Background Technology

[0002] In furniture manufacturing, building decoration, and other fields, the precision of board processing directly determines product quality. With the increasing demand for refined and efficient board processing, traditional board processing control relies heavily on manual experience to set parameters such as cutting speed and feed rate, or simply uses basic sensors to collect single data points for coarse control. This approach is ill-suited to the processing needs of boards with different materials and surface conditions. In recent years, visual inspection technology has been gradually applied to board processing, primarily to identify obvious surface defects such as cracks and scabs to prevent increased scrap rates due to processing in defective areas. However, current technologies often limit visual inspection to defect identification, failing to further analyze material uniformity by combining the detected surface texture features, or to predict stress distribution during processing. Furthermore, they neglect the impact of environmental temperature and humidity changes on the physical and mechanical properties of the board and processing precision. This results in insufficient adaptability of processing control parameters to the actual characteristics of the board and environmental conditions, making it difficult to meet the demands for high-precision and high-stability board processing. Therefore, an integrated processing control method that combines visual inspection data, stress prediction, and environmental adaptation is urgently needed.

[0003] Among related technologies, existing sheet metal processing control technologies have several shortcomings: I. Traditional control methods rely too much on human experience, and the judgment of the uniformity of the board material and potential defects is highly subjective. It is easy for the processing stress to concentrate due to parameter setting deviation, which can cause the board to deform or crack, resulting in poor processing accuracy and stability. Second, the application of visual inspection in sheet metal processing control is limited to surface defect identification. It does not convert visual data into quantitative information such as material uniformity distribution and defect distribution density, and cannot provide comprehensive data support for processing path planning and stress prediction. As a result, the path planning does not avoid the weak areas of the material, and the processing risk is high. Third, the control of sheet material processing did not take into account the influence of changes in temperature and humidity of the processing environment on the linear expansion coefficient and elastic modulus of the sheet material, and ignored the deformation of the sheet material caused by environmental factors, resulting in a disconnect between the preset processing parameters and the actual processing state. Fourth, most existing control models are fixed parameter models, which cannot dynamically optimize control strategies based on real-time board status and environmental data. They are difficult to cope with sudden situations during processing, and are prone to fluctuations in processing quality, leading to an increase in scrap rate. They are difficult to adapt to the production needs of modern high-precision board processing and have room for improvement. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides a vision-based inspection-based control system and method for sheet metal processing.

[0005] In a first aspect, this application provides a method for controlling the processing of sheet metal based on visual inspection, comprising the following steps: Step S1: Acquire surface images of the board to be processed to obtain board surface image data. Analyze the surface texture and structural features of the board surface image data to obtain board surface texture feature data and potential defect feature data. Based on the board surface texture feature data and potential defect feature data, evaluate the board material uniformity and structural integrity to obtain board material uniformity distribution data and defect distribution density data. Step S2: Based on the material uniformity distribution data and defect distribution density data of the plate, a preliminary planning of the machining tool path is performed to obtain initial machining path data. Based on the initial machining path data, the stress distribution during the machining process is predicted to obtain plate machining stress prediction distribution data. Based on the plate machining stress prediction distribution data, the machining parameters are adjusted and evaluated to obtain a preliminary machining control parameter set. Step S3: Obtain real-time processing environment data, which includes ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, identify the environmental factors affecting the predicted distribution data of the plate processing stress to obtain the temperature and humidity-related material deformation influence factor. Based on the temperature and humidity-related material deformation influence factor, perform processing environment adaptability correction on the preliminary processing control parameter set to obtain environmental processing control parameter data. Step S4: Construct and optimize a real-time processing control model for the environmental processing control parameter data to obtain a dynamic optimization control model for sheet metal processing. Deploy the dynamic optimization control model for sheet metal processing to the machine tool control equipment to execute sheet metal processing control.

[0006] Preferably, step S1 includes the following steps: Step S11: Using a high-resolution industrial camera group arranged above and to the side of the plate processing table, surface images of the plate to be processed are acquired under different lighting conditions to obtain plate surface image data. Step S12: Perform image preprocessing on the surface image data of the board to obtain standard image data of the board surface; Step S13: Perform multi-scale feature extraction on the standard image data of the board surface, and perform surface texture feature analysis and potential defect detection to obtain surface texture feature data and potential defect feature data of the board. Step S14: Based on the surface texture feature data of the board, analyze the uniformity and directional consistency of the surface texture to generate board material uniformity distribution data; based on the potential defect feature data, calculate the degree of defect aggregation and distribution location to generate defect distribution density data.

[0007] Preferably, step S2 includes the following steps: Step S21: Based on the material uniformity distribution data and defect distribution density data of the plate, and using an obstacle avoidance path planning algorithm, obtain the initial processing path data; Step S22: Based on the initial processing path data and the physical and mechanical property parameters of the plate, simulate the stress response inside the plate when the tool applies force at different path points, predict the corresponding stress concentration area and deformation during the processing, and obtain the plate processing stress prediction distribution data. Step S23: Identify stress over-limit areas and locate deformation-sensitive areas in the stress prediction distribution data of the plate processing to obtain processing risk area identification data; Step S24: Based on the processing risk area identification data, adjust and evaluate the processing parameters of the corresponding segments in the initial processing path data to generate a preliminary processing control parameter set.

[0008] Preferably, step S22 includes the following steps: Step S221: Assign mechanical property mapping values ​​to the uniformity distribution data of the board material to obtain the spatial mechanical property distribution data of the board. Step S222: Calculate the theoretical cutting force and torque at the path points of the initial machining path data based on the spatial mechanical property distribution data of the sheet metal, and obtain the theoretical cutting load data at the path points; Step S223: Based on the initial machining path data, predict the tool movement speed and acceleration during the machining process to obtain tool motion dynamics data; Step S224: Perform coupled simulation of the dynamic cutting process based on the tool motion dynamics data and the path point theoretical cutting load data to simulate the stress response of the plate under dynamic load; Step S225: Combine the defect distribution density data to reduce and correct the material strength in the high defect density area, and perform stress response simulation to obtain the predicted distribution data of plate processing stress.

[0009] Preferably, step S3 includes the following steps: Step S31: Real-time processing environment data is obtained by collecting temperature and humidity data of the processing environment in real time through environmental sensors deployed in the processing area; Step S32: Based on the real-time processing environment data, query the preset material temperature and humidity expansion coefficient database to obtain the linear expansion coefficient and elastic modulus correction coefficient of the board under the current environment; Step S33: Correct the predicted stress distribution data of the plate processing according to the linear expansion coefficient and elastic modulus correction coefficient to obtain the environmentally corrected predicted stress distribution data; Step S34: Identify processing risk areas based on the environmentally corrected stress prediction distribution data, and optimize the preliminary processing control parameter set to generate environmental processing control parameter data.

[0010] Preferably, step S33 includes the following steps: Step S331: Perform time series analysis on the real-time processing environment data to obtain environmental change trend data; Step S332: Based on the environmental change trend data, predict the thermal expansion deformation and moisture absorption deformation of the plate during the future processing time period to obtain the predicted deformation data of the plate; Step S333: Use the predicted deformation data of the plate as the initial boundary condition, superimpose it on the predicted stress distribution data of the plate processing, and perform coupled simulation. Based on the results of the coupled simulation, confirm the environmentally corrected predicted stress distribution data.

[0011] Preferably, step S4 includes the following steps: Step S41: Construct a reward function with processing quality, processing efficiency, and processing stability as optimization objectives; Step S42: Use the environmental processing control parameter data as the initial strategy, and use the plate surface image data, the plate processing stress prediction distribution data, and the real-time processing environment data as the state input; Step S43: Train an agent model in a virtual processing simulation environment using a deep reinforcement learning algorithm. The agent model learns the optimal control strategy under different states by trying different processing parameters to adjust actions and receiving reward values ​​calculated according to the reward function. Step S44: The trained intelligent agent model is encapsulated into the dynamic optimization control model for sheet metal processing and deployed to the CNC equipment of the machine tool to realize dynamic optimization control of the processing process.

[0012] Secondly, this application provides a vision-based sheet metal processing control system, comprising: The data acquisition module is used to acquire surface images of the board to be processed, obtain board surface image data, perform surface texture and structural feature analysis on the board surface image data, obtain board surface texture feature data and potential defect feature data, and evaluate the board material uniformity and structural integrity based on the board surface texture feature data and potential defect feature data, obtain board material uniformity distribution data and defect distribution density data. The analysis module is used to perform preliminary planning of the machining tool path based on the uniformity distribution data and defect distribution density data of the plate material to obtain initial machining path data, predict the stress distribution during the machining process based on the initial machining path data to obtain plate machining stress prediction distribution data, and evaluate the adjustment of machining parameters based on the plate machining stress prediction distribution data to obtain a preliminary machining control parameter set. The correction module is used to acquire real-time processing environment data, including ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, the module identifies the environmental factors affecting the predicted distribution data of the plate processing stress, obtains the temperature and humidity-related material deformation influence factor, and performs processing environment adaptability correction on the preliminary processing control parameter set based on the temperature and humidity-related material deformation influence factor to obtain environmental processing control parameter data. The optimization control module is used to construct and optimize the real-time processing control model based on the environmental processing control parameter data, thereby obtaining a dynamic optimization control model for sheet metal processing. The dynamic optimization control model for sheet metal processing is then deployed to the machine tool control equipment to execute sheet metal processing control.

[0013] Thirdly, this application provides a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform any of the above-described vision-based sheet metal processing control methods.

[0014] In summary, this application includes the following beneficial technical effects: This application provides a vision-based control method for sheet metal processing. High-resolution vision inspection is used to deeply mine the surface texture features and potential defects of the sheet metal, generating data on material uniformity distribution and defect density. This provides comprehensive data support for processing control, overcoming the limitations of traditional vision inspection which only identifies defects. This reduces processing risks caused by material misjudgment from the source. Combined with stress prediction and obstacle avoidance path planning, it accurately avoids areas of stress concentration and high defect density, reducing sheet metal deformation and cracking, and significantly improving processing accuracy. Simultaneously, real-time environmental temperature and humidity data are incorporated to correct the sheet metal's mechanical parameters and stress prediction results, eliminating the interference of environmental factors on processing accuracy and ensuring processing stability. A dynamic optimization control model is constructed through deep reinforcement learning, enabling adaptive adjustment of processing parameters according to changes in sheet metal state and environment. This eliminates excessive reliance on human experience, reducing operating costs and scrap rates, effectively meeting the demands of high-precision and high-stability modern sheet metal processing, thus significantly improving the efficiency of sheet metal processing control. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart of a method for controlling sheet metal processing based on visual inspection, according to an embodiment of this application.

[0017] Figure 2 This is a schematic diagram of a visual inspection-based sheet metal processing control system according to an embodiment of this application. Detailed Implementation

[0018] The following is in conjunction with the appendix Figure 1-2 This application will be described in further detail.

[0019] Example 1 This application discloses a method for controlling the processing of sheet metal based on visual inspection.

[0020] Reference Figure 1 A method for controlling sheet metal processing based on vision inspection includes the following steps: Step S1: Acquire surface images of the board to be processed to obtain board surface image data. Analyze the surface texture and structural features of the board surface image data to obtain board surface texture feature data and potential defect feature data. Based on the board surface texture feature data and potential defect feature data, evaluate the board material uniformity and structural integrity to obtain board material uniformity distribution data and defect distribution density data. Step S2: Based on the material uniformity distribution data and defect distribution density data of the plate, a preliminary planning of the machining tool path is performed to obtain initial machining path data. Based on the initial machining path data, the stress distribution during the machining process is predicted to obtain plate machining stress prediction distribution data. Based on the plate machining stress prediction distribution data, the machining parameters are adjusted and evaluated to obtain a preliminary machining control parameter set. Step S3: Obtain real-time processing environment data, which includes ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, identify the environmental factors affecting the predicted distribution data of the plate processing stress to obtain the temperature and humidity-related material deformation influence factor. Based on the temperature and humidity-related material deformation influence factor, perform processing environment adaptability correction on the preliminary processing control parameter set to obtain environmental processing control parameter data. Step S4: Construct and optimize a real-time processing control model for the environmental processing control parameter data to obtain a dynamic optimization control model for sheet metal processing. Deploy the dynamic optimization control model for sheet metal processing to the machine tool control equipment to execute sheet metal processing control.

[0021] Specifically, the process begins with surface image acquisition of the sheet metal. Appropriate image acquisition equipment is used to obtain surface image data, which is then analyzed in depth to extract texture features and potential defect characteristics. Based on these features, the material uniformity and structural integrity of the sheet metal are assessed, yielding material uniformity distribution data and defect density data. Next, based on these data, a preliminary tool path is planned, forming initial machining path data. This initial path data is then used to predict stress distribution during machining, allowing for the evaluation and adjustment of machining parameters, resulting in a preliminary machining control parameter set. Subsequently, real-time temperature and humidity data from the machining environment are collected, and the impact of these environmental factors on the machining stress is analyzed. A temperature- and humidity-related material deformation influencing factor is identified, and this factor is used to correct the preliminary machining control parameter set, yielding environmental machining control parameter data. Finally, based on the environmental machining control parameter data, a dynamic optimization control model for sheet metal machining is constructed and optimized, and then deployed to the machine tool control equipment to achieve effective control of the sheet metal machining process.

[0022] By adopting the above technical solution, visual inspection technology can comprehensively and accurately acquire texture and defect information of the board surface, providing a reliable basis for subsequent processing path planning and parameter adjustment. This avoids processing quality problems caused by insufficient understanding of the board characteristics. Combined with stress distribution prediction and environmental factor influence analysis, processing parameters can be dynamically adjusted to make the processing process more adaptable to board characteristics and environmental changes. This effectively reduces stress concentration and material deformation during processing, improves processing accuracy and board utilization. At the same time, the deployment of the dynamic optimization control model realizes intelligent control of the processing process, improves processing efficiency and stability, and reduces the cost and error of manual intervention.

[0023] It should be noted that step S1 includes the following steps: Step S11: Using a high-resolution industrial camera group arranged above and to the side of the plate processing table, surface images of the plate to be processed are acquired under different lighting conditions to obtain plate surface image data. Step S12: Perform image preprocessing on the surface image data of the board to obtain standard image data of the board surface; Step S13: Perform multi-scale feature extraction on the standard image data of the board surface, and perform surface texture feature analysis and potential defect detection to obtain surface texture feature data and potential defect feature data of the board. Step S14: Based on the surface texture feature data of the board, analyze the uniformity and directional consistency of the surface texture to generate board material uniformity distribution data; based on the potential defect feature data, calculate the degree of defect aggregation and distribution location to generate defect distribution density data.

[0024] Specifically, firstly, a high-resolution industrial camera group is arranged above and to the side of the sheet metal processing table to ensure that the surface of the sheet metal to be processed can be covered from multiple angles. Under different lighting conditions, the camera group is controlled to acquire surface images of the sheet metal to be processed. By adjusting parameters such as light intensity and angle, image data that can comprehensively reflect the surface condition of the sheet metal is obtained, forming sheet metal surface image data. Next, the acquired sheet metal surface image data is preprocessed, including removing noise from the image, correcting geometric distortion of the image, and adjusting the contrast and brightness of the image to improve the image quality and obtain standard image data of the sheet metal surface. Then, a multi-scale feature extraction algorithm is used to process the standard image data of the sheet metal surface to extract feature information from different scales. At the same time, combined with texture analysis algorithm and defect detection algorithm, the texture features and potential defect features of the sheet metal surface are identified respectively, resulting in sheet metal surface texture feature data and potential defect feature data. Finally, based on the surface texture feature data of the board, the uniformity and directional consistency of the surface texture are analyzed, and the uniformity distribution data of the board material is generated through quantitative processing; based on the potential defect feature data, the number, size, location and other information of defects are statistically analyzed, the degree of defect aggregation and distribution density are calculated, and defect distribution density data is generated.

[0025] By employing the above technical solution, high-resolution industrial cameras deployed at multiple locations and image acquisition under different lighting conditions can capture the detailed information of the board surface to the greatest extent, avoiding feature omissions caused by acquisition angle or lighting issues. The image preprocessing step effectively improves image quality, providing a high-quality data foundation for subsequent feature extraction and analysis, and improving the accuracy of feature recognition. Multi-scale feature extraction ensures that textures and defects within different size ranges can be effectively identified, making the obtained texture feature data and defect feature data more comprehensive and accurate. The material uniformity distribution data and defect distribution density data generated based on the above feature data can objectively reflect the material condition and structural integrity of the board, providing key reference for subsequent processing path planning and parameter adjustment, and helping to improve the scientificity and rationality of processing decisions.

[0026] It should be noted that step S2 includes the following steps: Step S21: Based on the material uniformity distribution data and defect distribution density data of the plate, and using an obstacle avoidance path planning algorithm, obtain the initial processing path data; Step S22: Based on the initial processing path data and the physical and mechanical property parameters of the plate, simulate the stress response inside the plate when the tool applies force at different path points, predict the corresponding stress concentration area and deformation during the processing, and obtain the plate processing stress prediction distribution data. Step S23: Identify stress over-limit areas and locate deformation-sensitive areas in the stress prediction distribution data of the plate processing to obtain processing risk area identification data; Step S24: Based on the processing risk area identification data, adjust and evaluate the processing parameters of the corresponding segments in the initial processing path data to generate a preliminary processing control parameter set.

[0027] Specifically, firstly, combining the material uniformity distribution data and defect density data of the sheet metal, an obstacle avoidance path planning algorithm is used to initially design the machining tool path. During the planning process, regions with uniform material and areas with concentrated defects are fully considered to ensure the path avoids areas with high defect density and areas where material inhomogeneity may lead to machining risks, thus forming initial machining path data. Next, the physical and mechanical properties of the sheet metal, such as elastic modulus, Poisson's ratio, and yield strength, are collected. Based on the initial machining path data, finite element simulation and other methods are used to simulate the stress changes within the sheet metal when the tool applies force at different path points, analyzing the stress transmission and distribution patterns to predict machining... The stress concentration areas and deformation of the sheet metal that may occur during the process are analyzed to obtain the predicted distribution data of sheet metal processing stress. Then, the predicted distribution data of sheet metal processing stress is analyzed in depth, stress thresholds and deformation thresholds are set, and areas with stress exceeding the thresholds and areas sensitive to deformation are identified. These areas are marked to obtain processing risk area identification data. Finally, based on the processing risk area identification data, for the corresponding segments of the initial processing path data that pass through the risk areas, the adjustment schemes of processing parameters such as cutting speed, feed rate, and depth of cut are evaluated. The effectiveness of different adjustment schemes in reducing processing risks is verified by simulation, and the optimal scheme is selected to generate a preliminary processing control parameter set.

[0028] By employing the above technical solution and using an obstacle avoidance path planning algorithm for initial path planning, high-risk areas can be avoided in the early stages of processing, reducing processing defects from the source and improving processing safety. Combining the physical and mechanical properties of the board material for stress distribution prediction allows for advance understanding of the stress state and deformation trend of the board material during processing, providing a scientific basis for identifying processing risk areas and avoiding blind spots in the processing process. Identifying risk areas makes the adjustment of processing parameters more targeted. By optimizing the processing parameters of corresponding segments, the possibility of stress exceeding limits and excessive deformation can be effectively reduced, improving the stability of processing quality. The generation of the preliminary processing control parameter set provides basic parameter guidance for subsequent processing processes, making processing operations more standardized and precise, and helping to improve processing efficiency and product qualification rate.

[0029] Furthermore, step S22 includes the following steps: Step S221: Assign mechanical property mapping values ​​to the uniformity distribution data of the board material to obtain the spatial mechanical property distribution data of the board. Step S222: Calculate the theoretical cutting force and torque at the path points of the initial machining path data based on the spatial mechanical property distribution data of the sheet metal, and obtain the theoretical cutting load data at the path points; Step S223: Based on the initial machining path data, predict the tool movement speed and acceleration during the machining process to obtain tool motion dynamics data; Step S224: Perform coupled simulation of the dynamic cutting process based on the tool motion dynamics data and the path point theoretical cutting load data to simulate the stress response of the plate under dynamic load; Step S225: Combine the defect distribution density data to reduce and correct the material strength in the high defect density area, and perform stress response simulation to obtain the predicted distribution data of plate processing stress.

[0030] Specifically, firstly, based on the material uniformity distribution data of the sheet metal, a mapping relationship between material uniformity and mechanical properties is established. The material uniformity indexes of different regions are converted into corresponding mechanical property parameters, such as hardness and strength. Mechanical property values ​​are assigned to each region of the sheet metal to obtain the spatial mechanical property distribution data of the sheet metal. Next, based on the spatial mechanical property distribution data of the sheet metal, combined with the position information of each path point in the initial machining path data, the theoretical cutting force and torque of the tool when performing cutting operations at each path point are calculated using cutting mechanics theory. Considering factors such as the contact state between the tool and the sheet metal and the cutting method, the theoretical cutting load data of the path points is obtained. Then, based on the tool's motion trajectory in the initial machining path data, the tool is analyzed in different paths. The motion law of the segment is analyzed to predict the speed and acceleration of the tool at each path point. The influence of factors such as the curvature and length of the path on the motion parameters is considered to obtain the tool motion dynamics data. Then, the tool motion dynamics data and the theoretical cutting load data of the path points are input into the dynamic cutting process simulation model for coupled simulation analysis. The stress generation, transmission and distribution inside the plate under dynamic load are simulated to obtain dynamic stress response data. Finally, the material strength in the high defect density area is reduced and corrected by combining the defect distribution density data. Considering the weakening effect of defects on the material's load-bearing capacity, the corrected strength parameters are incorporated into the stress response simulation and the simulation calculation is carried out again to obtain the final plate machining stress prediction distribution data.

[0031] It should be noted that step S3 includes the following steps: Step S31: Real-time processing environment data is obtained by collecting temperature and humidity data of the processing environment in real time through environmental sensors deployed in the processing area; Step S32: Based on the real-time processing environment data, query the preset material temperature and humidity expansion coefficient database to obtain the linear expansion coefficient and elastic modulus correction coefficient of the board under the current environment; Step S33: Correct the predicted stress distribution data of the plate processing according to the linear expansion coefficient and elastic modulus correction coefficient to obtain the environmentally corrected predicted stress distribution data; Step S34: Identify processing risk areas based on the environmentally corrected stress prediction distribution data, and optimize the preliminary processing control parameter set to generate environmental processing control parameter data.

[0032] Specifically, firstly, environmental sensors are strategically placed in the sheet metal processing area. These sensors monitor real-time temperature and humidity changes in the processing environment, collecting and storing the monitored temperature and humidity data to form real-time processing environment data. Next, a pre-defined database of material temperature and humidity expansion coefficients is established, storing the linear expansion coefficients and elastic modulus correction coefficients for different types of sheet metal under varying temperature and humidity conditions. Based on the temperature and humidity values ​​in the real-time processing environment data, this database is queried to obtain the corresponding linear expansion coefficients and elastic modulus correction coefficients for the sheet metal to be processed under the current environmental conditions. Then, using these coefficients, the previously obtained sheet metal processing stress prediction distribution data is corrected. The impact of material expansion and contraction due to temperature and humidity changes, as well as changes in elastic modulus, on the stress state of the sheet metal is considered, and the stress distribution is recalculated to obtain environmentally corrected stress prediction distribution data. Finally, based on the environmentally corrected stress prediction distribution data, potential stress over-limit and other risk areas during processing are re-identified. According to the characteristics of these risk areas, parameters such as cutting speed and feed rate in the initial processing control parameter set are optimized and adjusted. The applicability of the adjusted parameters in the current environment is verified through simulation, generating environmental processing control parameter data.

[0033] By adopting the above technical solution, the temperature and humidity data of the processing environment can be collected in real time, enabling timely monitoring of changes in environmental factors and providing a real-time basis for subsequent parameter correction. The preset material temperature and humidity expansion coefficient database facilitates the rapid acquisition of material characteristic parameters under different environments, ensuring the scientific nature and accuracy of parameter correction. Environmental correction is applied to the stress prediction distribution data of the sheet metal processing, making the stress prediction results more consistent with the sheet metal state under actual processing conditions, avoiding stress prediction errors caused by ignoring environmental factors. Based on the corrected stress data, the processing control parameters are optimized, enabling the processing parameters to adapt to environmental changes, effectively reducing sheet metal deformation and processing errors caused by temperature and humidity changes, improving the stability of the processing process and the consistency of processing quality, and is especially suitable for sheet metal processing scenarios that are sensitive to the environment.

[0034] Furthermore, step S33 includes the following steps: Step S331: Perform time series analysis on the real-time processing environment data to obtain environmental change trend data; Step S332: Based on the environmental change trend data, predict the thermal expansion deformation and moisture absorption deformation of the plate during the future processing time period to obtain the predicted deformation data of the plate; Step S333: Use the predicted deformation data of the plate as the initial boundary condition, superimpose it on the predicted stress distribution data of the plate processing, and perform coupled simulation. Based on the results of the coupled simulation, confirm the environmentally corrected predicted stress distribution data.

[0035] Specifically, firstly, time series analysis is performed on real-time processing environment data. Temperature and humidity data over a period of time are collected, and time series analysis methods such as moving average and exponential smoothing are used to analyze the changing trends of temperature and humidity data, including the rate of change and fluctuation amplitude, to obtain environmental change trend data. Next, based on the environmental change trend data and combined with the thermal expansion and moisture absorption characteristics of the sheet material, a sheet deformation prediction model is established to predict the magnitude, direction, and distribution of thermal expansion and moisture absorption deformation that the sheet material may experience during the entire processing period, resulting in predicted sheet deformation data. Then, the predicted sheet deformation data is used as initial boundary conditions and input into the calculation model for the predicted distribution of sheet material processing stress. Coupled simulation analysis is performed with the stress generated during processing to simulate the stress distribution state of the sheet material under the combined action of deformation caused by environmental factors and processing stress. Based on the results of the coupled simulation, the superposition effect and distribution law of stress are analyzed to confirm the predicted stress distribution data after considering the influence of environmental changes, i.e., the environmentally corrected predicted stress distribution data.

[0036] It should be noted that step S4 includes the following steps: Step S41: Construct a reward function with processing quality, processing efficiency, and processing stability as optimization objectives; Step S42: Use the environmental processing control parameter data as the initial strategy, and use the plate surface image data, the plate processing stress prediction distribution data, and the real-time processing environment data as the state input; Step S43: Train an agent model in a virtual processing simulation environment using a deep reinforcement learning algorithm. The agent model learns the optimal control strategy under different states by trying different processing parameters to adjust actions and receiving reward values ​​calculated according to the reward function. Step S44: The trained intelligent agent model is encapsulated into the dynamic optimization control model for sheet metal processing and deployed to the CNC equipment of the machine tool to realize dynamic optimization control of the processing process.

[0037] Specifically, firstly, a reward function is constructed, with processing quality, processing efficiency, and processing stability as optimization objectives. The precision and surface roughness of the processed sheet metal are used as evaluation indicators for processing quality; processing time and material utilization are used as evaluation indicators for processing efficiency; and stress fluctuations and tool vibrations during processing are used as evaluation indicators for processing stability. These indicators are integrated into a comprehensive reward value through weighted allocation. Next, environmental processing control parameter data is used as the initial strategy for the agent model, determining the initial parameter settings of the model. Simultaneously, sheet metal surface image data, sheet metal processing stress prediction distribution data, and real-time processing environment data are used as the model's state inputs. This data comprehensively reflects various state information during the processing. Then, a deep reinforcement learning algorithm is used to train the agent model in a constructed virtual processing simulation environment. During training, the agent model continuously tries different processing parameters to adjust actions, such as changing cutting speed and feed rate, applying these actions to the virtual processing environment. The reward value of the reward function is then calculated based on the processing results. By learning the relationship between the reward value and actions / states, the agent model continuously optimizes its decision-making strategy to learn the optimal control strategy that maximizes the reward value under different states. Finally, once the intelligent agent model reaches the preset performance indicators, it is encapsulated as a dynamic optimization control model for sheet metal processing. This model is then deployed to the CNC equipment of the machine tool through interface development and other methods, enabling the CNC equipment to call the optimal control strategy output by the model based on the real-time collected processing status data, thereby achieving dynamic optimization control of the processing process.

[0038] By adopting the above technical solution, the constructed reward function comprehensively considers processing quality, efficiency, and stability, making the optimization objective more comprehensive and avoiding the problem of declining other performance indicators caused by single-objective optimization. Using environmental processing control parameter data as the initial strategy provides a good starting point for model training, accelerates the model's convergence speed, and improves training efficiency. The application of deep reinforcement learning algorithms enables the model to autonomously learn the optimal control strategy in complex processing environments, exhibiting strong adaptability and generalization ability. It can cope with various uncertainties in the processing process. Deploying the trained model to CNC equipment realizes intelligent dynamic control of the processing process, which can automatically adjust processing parameters according to real-time status, ensuring that the processing process is always in the optimal state, effectively improving the consistency and stability of processing quality, while also increasing processing efficiency and reducing dependence on operator experience.

[0039] Example 2 This application also discloses a visual inspection-based sheet metal processing control system.

[0040] Reference Figure 2A vision-based inspection-based sheet metal processing control system includes: The data acquisition module is used to acquire surface images of the board to be processed, obtain board surface image data, perform surface texture and structural feature analysis on the board surface image data, obtain board surface texture feature data and potential defect feature data, and evaluate the board material uniformity and structural integrity based on the board surface texture feature data and potential defect feature data, obtain board material uniformity distribution data and defect distribution density data. The analysis module is used to perform preliminary planning of the machining tool path based on the uniformity distribution data and defect distribution density data of the plate material to obtain initial machining path data, predict the stress distribution during the machining process based on the initial machining path data to obtain plate machining stress prediction distribution data, and evaluate the adjustment of machining parameters based on the plate machining stress prediction distribution data to obtain a preliminary machining control parameter set. The correction module is used to acquire real-time processing environment data, including ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, the module identifies the environmental factors affecting the predicted distribution data of the plate processing stress, obtains the temperature and humidity-related material deformation influence factor, and performs processing environment adaptability correction on the preliminary processing control parameter set based on the temperature and humidity-related material deformation influence factor to obtain environmental processing control parameter data. The optimization control module is used to construct and optimize the real-time processing control model based on the environmental processing control parameter data, thereby obtaining a dynamic optimization control model for sheet metal processing. The dynamic optimization control model for sheet metal processing is then deployed to the machine tool control equipment to execute sheet metal processing control.

[0041] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention, they should all fall within the protection scope of the present invention.

[0042] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0043] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A method for controlling sheet metal processing based on vision inspection, characterized in that, Includes the following steps: Step S1: Acquire surface images of the board to be processed to obtain board surface image data. Analyze the surface texture and structural features of the board surface image data to obtain board surface texture feature data and potential defect feature data. Based on the board surface texture feature data and potential defect feature data, evaluate the board material uniformity and structural integrity to obtain board material uniformity distribution data and defect distribution density data. Step S2: Based on the material uniformity distribution data and defect distribution density data of the plate, a preliminary planning of the machining tool path is performed to obtain initial machining path data. Based on the initial machining path data, the stress distribution during the machining process is predicted to obtain plate machining stress prediction distribution data. Based on the plate machining stress prediction distribution data, the machining parameters are adjusted and evaluated to obtain a preliminary machining control parameter set. Step S3: Obtain real-time processing environment data, which includes ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, identify the environmental factors affecting the predicted distribution data of the plate processing stress to obtain the temperature and humidity-related material deformation influence factor. Based on the temperature and humidity-related material deformation influence factor, perform processing environment adaptability correction on the preliminary processing control parameter set to obtain environmental processing control parameter data. Step S4: Construct and optimize a real-time processing control model for the environmental processing control parameter data to obtain a dynamic optimization control model for sheet metal processing. Deploy the dynamic optimization control model for sheet metal processing to the machine tool control equipment to execute sheet metal processing control.

2. The method for controlling sheet metal processing based on vision inspection according to claim 1, characterized in that, Step S1 includes the following steps: Step S11: Using a high-resolution industrial camera group arranged above and to the side of the plate processing table, surface images of the plate to be processed are acquired under different lighting conditions to obtain plate surface image data. Step S12: Perform image preprocessing on the surface image data of the board to obtain standard image data of the board surface; Step S13: Perform multi-scale feature extraction on the standard image data of the board surface, and perform surface texture feature analysis and potential defect detection to obtain surface texture feature data and potential defect feature data of the board. Step S14: Based on the surface texture feature data of the board, analyze the uniformity and directional consistency of the surface texture to generate board material uniformity distribution data; based on the potential defect feature data, calculate the degree of defect aggregation and distribution location to generate defect distribution density data.

3. The method for controlling sheet metal processing based on vision inspection according to claim 1, characterized in that, Step S2 includes the following steps: Step S21: Based on the material uniformity distribution data and defect distribution density data of the plate, and using an obstacle avoidance path planning algorithm, obtain the initial processing path data; Step S22: Based on the initial processing path data and the physical and mechanical property parameters of the plate, simulate the stress response inside the plate when the tool applies force at different path points, predict the corresponding stress concentration area and deformation during the processing, and obtain the plate processing stress prediction distribution data. Step S23: Identify stress over-limit areas and locate deformation-sensitive areas in the stress prediction distribution data of the plate processing to obtain processing risk area identification data; Step S24: Based on the processing risk area identification data, adjust and evaluate the processing parameters of the corresponding segments in the initial processing path data to generate a preliminary processing control parameter set.

4. The method for controlling sheet metal processing based on vision inspection according to claim 3, characterized in that, Step S22 includes the following steps: Step S221: Assign mechanical property mapping values ​​to the uniformity distribution data of the board material to obtain the spatial mechanical property distribution data of the board. Step S222: Calculate the theoretical cutting force and torque at the path points of the initial machining path data based on the spatial mechanical property distribution data of the sheet metal, and obtain the theoretical cutting load data at the path points; Step S223: Based on the initial machining path data, predict the tool movement speed and acceleration during the machining process to obtain tool motion dynamics data; Step S224: Perform coupled simulation of the dynamic cutting process based on the tool motion dynamics data and the path point theoretical cutting load data to simulate the stress response of the plate under dynamic load; Step S225: Combine the defect distribution density data to reduce and correct the material strength in the high defect density area, and perform stress response simulation to obtain the predicted distribution data of plate processing stress.

5. The method for controlling sheet metal processing based on vision inspection according to claim 1, characterized in that, Step S3 includes the following steps: Step S31: Real-time processing environment data is obtained by collecting temperature and humidity data of the processing environment in real time through environmental sensors deployed in the processing area; Step S32: Based on the real-time processing environment data, query the preset material temperature and humidity expansion coefficient database to obtain the linear expansion coefficient and elastic modulus correction coefficient of the board under the current environment; Step S33: Correct the predicted stress distribution data of the plate processing according to the linear expansion coefficient and elastic modulus correction coefficient to obtain the environmentally corrected predicted stress distribution data; Step S34: Identify processing risk areas based on the environmentally corrected stress prediction distribution data, and optimize the preliminary processing control parameter set to generate environmental processing control parameter data.

6. The method for controlling sheet metal processing based on vision inspection according to claim 5, characterized in that, Step S33 includes the following steps: Step S331: Perform time series analysis on the real-time processing environment data to obtain environmental change trend data; Step S332: Based on the environmental change trend data, predict the thermal expansion deformation and moisture absorption deformation of the plate during the future processing time period to obtain the predicted deformation data of the plate; Step S333: Use the predicted deformation data of the plate as the initial boundary condition, superimpose it on the predicted stress distribution data of the plate processing, and perform coupled simulation. Based on the results of the coupled simulation, confirm the environmentally corrected predicted stress distribution data.

7. The method for controlling sheet metal processing based on vision inspection according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: Construct a reward function with processing quality, processing efficiency, and processing stability as optimization objectives; Step S42: Use the environmental processing control parameter data as the initial strategy, and use the plate surface image data, the plate processing stress prediction distribution data, and the real-time processing environment data as the state input; Step S43: Train an agent model in a virtual processing simulation environment using a deep reinforcement learning algorithm. The agent model learns the optimal control strategy under different states by trying different processing parameters to adjust actions and receiving reward values ​​calculated according to the reward function. Step S44: The trained intelligent agent model is encapsulated into the dynamic optimization control model for sheet metal processing and deployed to the CNC equipment of the machine tool to realize dynamic optimization control of the processing process.

8. A vision-based sheet metal processing control system, applied to the vision-based sheet metal processing control method described in any one of claims 1-7, characterized in that, include: The data acquisition module is used to acquire surface images of the board to be processed, obtain board surface image data, perform surface texture and structural feature analysis on the board surface image data, obtain board surface texture feature data and potential defect feature data, and evaluate the board material uniformity and structural integrity based on the board surface texture feature data and potential defect feature data, obtain board material uniformity distribution data and defect distribution density data. The analysis module is used to perform preliminary planning of the machining tool path based on the uniformity distribution data and defect distribution density data of the plate material to obtain initial machining path data, predict the stress distribution during the machining process based on the initial machining path data to obtain plate machining stress prediction distribution data, and evaluate the adjustment of machining parameters based on the plate machining stress prediction distribution data to obtain a preliminary machining control parameter set. The correction module is used to acquire real-time processing environment data, including ambient temperature data and humidity data. Based on the ambient temperature data and humidity data, the module identifies the environmental factors affecting the predicted distribution data of the plate processing stress, obtains the temperature and humidity-related material deformation influence factor, and performs processing environment adaptability correction on the preliminary processing control parameter set based on the temperature and humidity-related material deformation influence factor to obtain environmental processing control parameter data. The optimization control module is used to construct and optimize the real-time processing control model based on the environmental processing control parameter data, thereby obtaining a dynamic optimization control model for sheet metal processing. The dynamic optimization control model for sheet metal processing is then deployed to the machine tool control equipment to execute sheet metal processing control.

9. A computer-readable storage medium, characterized in that: The device stores instructions that, when executed on a computer, cause the computer to perform a vision-based sheet metal processing control method as described in any one of claims 1 to 7.

Images