Automatic spraying process intelligent regulation and control and quality optimization method and system

Abstract

The invention relates to the technical field of industrial robot surface treatment, in particular to an intelligent control and quality optimization method and system for industrial robot automatic spraying in the automatic spraying process, and is suitable for scenes such as automobile body spraying, household appliance shell coating and aerospace part surface treatment. Dynamic parameter optimization, film thickness uniformity control, coating utilization rate improvement and defect online detection in the spraying process can be achieved, and the problems that in traditional spraying, the coating quality consistency is poor, coating waste is serious, and energy consumption is high are solved.

Description

Technical Field

[0001] This invention relates to the field of industrial robot surface treatment technology, and in particular to an automated spraying method and system for industrial robots, which is an intelligent control and quality optimization method for the automated spraying process. Background Technology

[0002] Automated spraying with industrial robots is a key process in modern manufacturing for achieving surface protection and decoration of products. Currently, there are four major technical bottlenecks in spraying process control: First, controlling film thickness uniformity is difficult; traditional fixed-parameter spraying cannot adapt to complex curved surfaces (curvature variation range ±30°) and masked areas, resulting in film thickness deviations of up to ±20%. Second, paint utilization is low; conventional spraying trajectory planning does not consider paint atomization characteristics, resulting in material utilization of only 40-60%, leading to significant waste. Third, energy consumption and environmental costs are high; the lack of optimized matching between spraying parameters and energy consumption results in excessive VOC emissions. Fourth, defect detection is lagging; reliance on manual visual inspection or offline detection makes it impossible to detect defects such as runs and pinholes in real time.

[0003] In existing technologies, some solutions improve coating quality by enhancing robot motion precision, but lack dynamic parameter adjustment mechanisms; some solutions employ simple flow control strategies, which can only adapt to a single workpiece; and some solutions rely on manual experience for adjustment, making it difficult to achieve flexible production of multiple product types. Therefore, there is an urgent need to construct a coating control system that integrates multi-source sensing and intelligent decision-making. Summary of the Invention

[0004] This invention provides a method and system for intelligent control and quality optimization of an automated spraying process. Through a four-layer architecture of multi-sensor fusion perception, precise film thickness control, process parameter optimization, and online defect detection, it achieves high-quality, high-efficiency, and low-consumption automated spraying. The core innovation lies in proposing four key algorithms: a multi-field coupling-based dynamic film thickness prediction and control algorithm, a coating atomization characteristic-driven spraying trajectory optimization algorithm, a multi-objective collaborative intelligent decision-making algorithm for spraying parameters, and a real-time identification and source analysis algorithm for coating defects. A complete mathematical model and efficiency-enhancing mechanism have also been established, which can significantly improve spraying quality and resource utilization efficiency.

[0005] A first aspect of this invention provides a method for intelligent control and quality optimization of an automated spraying process, comprising the following steps:

[0006] Multi-source information acquisition during the spraying process: Deploy a vision-infrared-flow sensing fusion system to collect real-time data on the workpiece's 3D model, surface temperature field, paint flow rate, atomization pressure, robot motion parameters, and environmental data;

[0007] Film thickness dynamic prediction and control: Based on the acquired data, the coating thickness is accurately predicted and closed-loop controlled by a film thickness dynamic prediction and control algorithm based on multi-field coupling.

[0008] Spraying trajectory and parameter optimization: Based on the film thickness control target, the spraying trajectory optimization algorithm driven by the coating atomization characteristics and the intelligent decision-making algorithm for multi-objective collaborative spraying parameters are activated to generate the optimal trajectory and parameter combination;

[0009] Coating Defect Detection and Process Correction: Surface defects are detected through real-time identification and source analysis algorithms for coating defects, the root causes are located, and parameter adjustment schemes are generated;

[0010] Full-process closed-loop optimization: Based on historical data, a quality-efficiency-cost correlation model is established, and algorithm parameters are continuously iterated and optimized to achieve continuous improvement of the spraying process.

[0011] A second aspect of the present invention provides a system for implementing the above method, comprising:

[0012] Multi-source sensing units: 3D scanner (accuracy ±0.1mm), high-speed camera (20 million pixels resolution, frame rate ≥120fps), infrared thermal imager (temperature measurement range 0-100℃), flow sensor (accuracy ±1%FS), pressure sensor, robot encoder;

[0013] Data processing unit: GPU-accelerated computing module (image recognition and defect detection), CFD computing module, industrial computer (data fusion and algorithm execution);

[0014] Intelligent decision-making unit: Deploys four core algorithms to achieve film thickness control, trajectory optimization, parameter decision-making and defect analysis;

[0015] Robot and spraying equipment control unit: six-axis spraying robot (repeat positioning accuracy ±0.05mm), high-precision spraying equipment (flow control accuracy ±2%), real-time controller (control cycle ≤2ms);

[0016] Human-computer interaction and management unit: monitoring terminal, process management software, and database server, to realize the monitoring and data management of the spraying process.

[0017] Beneficial effects

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The coating quality has been significantly improved: film thickness uniformity has increased from ±20% to ±3%, coating defect rate has decreased from 8% to 0.1%, and product qualification rate has increased to 99.9%.

[0020] Improved resource utilization efficiency: Coating utilization rate increased from 50% to 85%, saving more than 35% of annual coating costs and reducing VOC emissions by 40%;

[0021] Production efficiency and energy consumption optimization: Spraying time is reduced by 12%, overall energy consumption is reduced by 25%, and process debugging time is reduced from 4 hours to 15 minutes;

[0022] Enhanced flexibility: It can automatically adapt to complex curved workpieces and 50+ types of coatings, reducing changeover time by 90% and manual intervention rate by 95%. Attached Figure Description

[0023] Figure 1 Flowchart of the system method algorithm implementation principle. Detailed Implementation

[0024] Example 1: Algorithm for Dynamic Prediction and Control of Film Thickness Based on Multi-Field Coupling

[0025] Algorithm principle: A multi-field coupled model of "spraying parameters-flow field-temperature field-film thickness" is established. The trajectory of coating particles and the deposition process are simulated by computational fluid dynamics (CFD). The film thickness is predicted and controlled in real time by combining a deep learning model.

[0026] Key innovations:

[0027] A hybrid model for film thickness prediction is proposed:

[0028] h(x,y,z,t)=fCFD(u,v,p,T)+fDL(Δhhist), where fCFD is the physical model and fDL is the data-driven correction model;

[0029] Design an adaptive film thickness closed-loop control strategy to dynamically adjust the spraying parameters based on the prediction deviation Δh = htarget - hpredicted;

[0030] A surface partitioning algorithm was developed to divide the workpiece surface into regions with different curvatures (κ<0.5m-1 is a smooth region, and κ≥0.5m-1 is a curved region) and implement differentiated control.

[0031] Modeling efficiency enhancement principle:

[0032] Establish a film thickness uniformity-control accuracy model: Q1=1-σ(h) / htarget, where σ(h) is the standard deviation of film thickness;

[0033] By increasing Q1 (from 0.8 to 0.98), film thickness deviation can be reduced, thus decreasing rework rates.

[0034] Mathematical verification: When the film thickness control accuracy is improved from ±20% to ±3%, material waste is reduced by 35% and rework costs are reduced by 80%.

[0035] Overall benefits: Film thickness uniformity improved by 22.5%, coating performance stability (adhesion, corrosion resistance) improved by 40%, and rework rate reduced from 15% to 1%.

[0036] Innovation Point 2: Spraying Trajectory Optimization Algorithm Driven by Coating Atomization Characteristics

[0037] Algorithm principle: Based on the atomization characteristic parameters of the coating (mist cone angle θ, particle diameter distribution D32, velocity distribution v(r)), a mapping relationship between the spraying trajectory and the coating deposition efficiency is established, and an improved genetic algorithm is used to optimize the path spacing, spraying speed and overlap rate.

[0038] Key innovations:

[0039] A coating utilization rate assessment model is proposed:

[0040] η=∫∫Aworkpieceρ(x,y)dxdy / ∫∫Asprayρ(x,y)dxdy, which quantifies the effective utilization rate of the coating;

[0041] Design a variable spacing trajectory planning strategy, and dynamically adjust the path spacing s = k·D32·cos(θ / 2) / κα according to the workpiece curvature κ;

[0042] Develop an intelligent processing algorithm for occlusion areas, which automatically generates detour trajectories and re-spraying strategies based on visual recognition.

[0043] Modeling efficiency enhancement principle:

[0044] Establish a utilization rate-trajectory parameter model: η=f(s,v,γ), where s is the path spacing, v is the spraying speed, and γ is the overlap rate;

[0045] Reduce paint waste and lower raw material costs by optimizing η (from 50% to 85%);

[0046] Mathematical verification: When η increases by 35 percentage points, for a production line with an annual output of 100,000 cars, the annual saving in paint costs is approximately 2.8 million yuan;

[0047] Overall benefits: 70% increase in paint utilization, 40% reduction in VOC emissions, 15% optimization of trajectory length, and 12% reduction in spraying time.

[0048] Innovation Point 3: Intelligent Decision-Making Algorithm for Multi-Objective Collaborative Spraying Parameters

[0049] Algorithm principle: A multi-objective optimization model is established with the goals of film thickness uniformity, coating utilization rate and energy consumption. Based on the workpiece material, shape and process requirements, key parameters such as spray flow rate q, atomization pressure p, robot speed v and distance d are dynamically optimized.

[0050] Key innovations:

[0051] The objective function for parameter optimization is proposed as follows: F = ω1(1-Q1) + ω2(1-η) + ω3(E / E0), where E is the real-time energy consumption and E0 is the baseline energy consumption;

[0052] Design an adaptive weight adjustment mechanism to dynamically adjust ω1, ω2, and ω3 according to production needs (quality priority / efficiency priority / cost priority);

[0053] We developed a fast parameter optimization algorithm and combined it with process knowledge graph pruning to optimize the space, compressing the optimization time to less than 100ms.

[0054] Modeling efficiency enhancement principle:

[0055] Establish a comprehensive benefit evaluation model: B = c1·Q1 + c2·η - c3·(E / E0);

[0056] Achieving a balance between quality, efficiency, and cost by maximizing B;

[0057] Mathematical verification: After parameter optimization, the unit area spraying cost is reduced by 32%, while the film thickness qualification rate remains above 99%;

[0058] Overall benefits: Overall energy consumption is reduced by 25%, process parameter debugging time is shortened from 4 hours to 15 minutes, and adaptability to multiple varieties is improved by 80%.

[0059] Innovation Point 4: Algorithm for Real-time Identification and Source Tracing of Coating Defects

[0060] Algorithm principle: High-speed vision system acquires images of wet and dry films, extracts defect features (area, shape, gray value), and uses an improved Transformer model to classify and identify defects such as drips, pinholes, and orange peel. Combined with time-series data of process parameters, the cause of defects is located.

[0061] Key innovations:

[0062] A defect severity assessment index is proposed: S = ∑wi·si, where si is the defect quantification parameter and wi is the weighting coefficient;

[0063] Design defect-parameter correlation analysis model, based on mutual information MI(f,d)=I(F;D) to locate key influencing parameters;

[0064] Develop a closed-loop correction mechanism to automatically generate parameter adjustment schemes based on the defect type (e.g., reduce flow rate by 10-15% for sag defects).

[0065] Modeling efficiency enhancement principle:

[0066] Establish a defect rate-correction speed model: R = r0·ek·t, where t is the defect response time;

[0067] Rapidly curb defect propagation by shortening t (from 2 hours to 10 seconds);

[0068] Mathematical verification: When the defect detection response time is reduced from 2 hours to 10 seconds, the batch defect incidence rate is reduced by 98%;

[0069] Overall benefits: Defect identification accuracy ≥99%, defect traceability time reduced by 99.7%, batch quality incidents reduced by 95%, and customer complaint rate reduced by 90%.

[0070] Example 2: Application of a Multi-Field Coupling-Based Dynamic Prediction and Control Algorithm for Film Thickness

[0071] To address the issues of poor film thickness uniformity and low control precision in traditional spraying processes, this embodiment innovatively applies a "film thickness dynamic prediction and control algorithm based on multi-field coupling." By constructing a prediction system that integrates physical and data models, and combining it with differentiated regional control strategies, it achieves precise control and stable improvement of sprayed film thickness.

[0072] (I) Construction Logic of Multi-Field Coupling Model

[0073] The core of the algorithm lies in establishing a multi-dimensional correlation between "spraying parameters, flow field, temperature field, and film thickness," breaking the limitations of single-parameter control.

[0074] First, a physical model was constructed using computational fluid dynamics (CFD) to simulate the trajectory and deposition process of coating particles after they are ejected from the nozzle. In the model, parameters such as spraying pressure, nozzle velocity, and coating viscosity were used as inputs. The flow field distribution was simulated by solving the Navier-Stokes equations, and the changes in particle velocity and concentration under different flow field conditions were analyzed. Simultaneously, a temperature field coupling analysis was introduced to consider the influence of ambient temperature and coating solids content on the particle drying rate, establishing the correlation between temperature distribution and particle deposition efficiency. The physical model can preliminarily predict the film thickness distribution trend under different combinations of spraying parameters, providing a basis for subsequent control.

[0075] Secondly, to address the issue of significant prediction deviations in physical models under complex working conditions (such as curved workpiece surfaces and airflow interference), a deep learning model is introduced as a data-driven correction module. Process parameters from historical spraying processes, multi-site monitoring data, and actual film thickness detection results are collected to construct a training dataset. A correction model is designed using a convolutional neural network (CNN) combined with a long short-term memory network (LSTM), dynamically adjusting the prediction results of the physical model using historical film thickness deviation data. Through the synergy between the physical model and the data model, a significant improvement in film thickness prediction accuracy is achieved, ensuring reliable prediction values ​​are output under various working conditions.

[0076] (II) Adaptive Closed-Loop Control and Surface Partitioning Strategy

[0077] Based on the film thickness prediction results, a full-process adaptive closed-loop control mechanism is designed: real-time acquisition of film thickness detection data during the spraying process (using an online laser thickness gauge), calculation of the deviation between the actual film thickness and the target film thickness, and dynamic adjustment of spraying parameters according to the magnitude and trend of the deviation. For example, when the predicted film thickness is lower than the target value, the spraying flow rate is appropriately increased or the nozzle movement speed is reduced while ensuring the coating atomization effect; when the film thickness deviation shows a continuous increasing trend, the atomization pressure and spraying distance are adjusted simultaneously to prevent further expansion of the deviation. The response period of the closed-loop control is controlled at the millisecond level to ensure that the film thickness deviation can be corrected in a timely manner.

[0078] To address the challenge of painting complex curved surfaces, a surface partitioning algorithm was developed: A 3D model of the workpiece surface is obtained through 3D visual scanning, the curvature value of each region is calculated, and the workpiece surface is divided into smooth zones (curvature value less than 0.5m). -1 ) and the curved region (curvature value greater than or equal to 0.5m) -1 For flat areas, conventional spraying parameters and path planning are used to ensure spraying efficiency; for curved areas, the spraying path spacing is appropriately reduced, the nozzle moving speed is decreased, and the spraying overlap rate is increased. Fine parameter control is used to avoid uneven film thickness caused by changes in surface curvature.

[0079] (III) Qualitative Explanation of Algorithm Enhancement

[0080] The application of a multi-field coupled prediction model significantly improves the accuracy of film thickness prediction, raising the film thickness uniformity index from 0.8 in traditional control to 0.98, with film thickness deviation controlled within ±3%, far exceeding the ±20% accuracy level of traditional processes. This improvement directly leads to a significant reduction in material waste and a substantial increase in coating utilization. Simultaneously, the rework rate due to unqualified film thickness has decreased from 15% to 1%, reducing rework costs by over 80%. Furthermore, the improved film thickness uniformity enhances the adhesion, corrosion resistance, and other performance stability of the coating by 40%, extending the service life of the workpiece and comprehensively improving the quality competitiveness of the sprayed products.

[0081] Example 2: Application of a spray trajectory optimization algorithm driven by paint atomization characteristics

[0082] To address the issues of paint waste and low spraying efficiency caused by the mismatch between traditional spraying trajectory planning and paint atomization characteristics, this embodiment innovatively applies a "spraying trajectory optimization algorithm driven by paint atomization characteristics." By establishing a correlation model between atomization parameters and trajectory parameters, combined with an intelligent path planning strategy, high efficiency and energy saving in the spraying process are achieved.

[0083] (I) Modeling the correlation between atomization characteristics and trajectory parameters

[0084] The algorithm first focuses on the precise characterization of coating atomization properties. It collects core parameters such as the fog cone angle, particle diameter distribution, and velocity distribution using specialized atomization detection equipment to establish a fog characteristic database. Based on this, a mapping model between the spray trajectory and coating deposition efficiency is constructed: using the fog cone angle and particle diameter distribution as key input variables, the algorithm analyzes the deposition distribution of coating particles on the workpiece surface under different path spacing, spraying speed, and overlap rates, quantifying the effective utilization rate of the coating (i.e., the ratio of the actual amount of coating deposited on the workpiece surface to the total amount of coating sprayed).

[0085] To achieve global optimization of trajectory parameters, an improved genetic algorithm is employed to optimize path spacing, spraying speed, and overlap rate. The algorithm uses paint utilization and spraying time as optimization objectives, designs a fitness function, and iteratively searches for the optimal parameter combination through selection, crossover, and mutation operations. Compared to traditional genetic algorithms, the improved algorithm introduces atomization characteristic constraints to avoid generating trajectory parameters that do not conform to atomization patterns. Simultaneously, it employs adaptive mutation probability to accelerate the algorithm's convergence speed, ensuring that the globally optimal solution is found in a shorter time.

[0086] (II) Strategies for Handling Variable Spacing Trajectories and Obscured Areas

[0087] Based on the workpiece's curved surface features and atomization characteristics, a variable-spacing trajectory planning strategy was developed: the spraying path spacing is dynamically adjusted according to the curvature values ​​of different regions of the workpiece. For regions with greater curvature (such as curved areas), the path spacing is smaller to ensure uniform paint coverage; for regions with less curvature (such as flat areas), the path spacing is appropriately increased to improve spraying efficiency. The adjustment formula for the path spacing is determined based on the mist cone angle and particle diameter distribution, ensuring effective overlap of paint deposition areas on adjacent spraying paths and avoiding problems such as missed spraying or excessive overlap.

[0088] To address situations where workpiece surfaces have obscured areas (such as holes, grooves, or uncoated areas requiring protection), an intelligent algorithm for handling obscured areas was developed. This algorithm uses a machine vision system to identify obscured areas on the workpiece surface and generate their boundary coordinates. Based on the shape and location of the obscured areas, the spraying trajectory is automatically adjusted to generate a detour path, preventing paint from being sprayed onto the obscured areas. For areas around the obscured areas that are prone to missed spraying, a respraying strategy is designed. By locally adjusting spraying parameters (such as reducing spraying speed and decreasing the fog cone angle), the film thickness in the surrounding areas is ensured to meet the standards.

[0089] (III) Qualitative Explanation of Algorithm Enhancement

[0090] The trajectory optimization algorithm driven by paint atomization characteristics increases paint utilization from the traditional 50% to 85%, significantly reducing paint waste and lowering raw material costs and VOC emissions (by 40%). The variable-pitch trajectory planning strategy optimizes the spray trajectory length by 15% and shortens spraying time by 12% while ensuring spraying quality, significantly improving production efficiency. The intelligent masking area processing algorithm avoids mis-spraying in masked areas and missed spraying in surrounding areas, reducing subsequent cleaning work and rework rates, further lowering production costs. Taking a production line with an annual output of 100,000 cars as an example, annual paint cost savings can reach over 2.8 million yuan, demonstrating significant economic benefits.

[0091] Example 3: Application of Intelligent Decision-Making Algorithm for Multi-Objective Collaborative Spraying Parameters

[0092] To address the challenge of balancing multiple objectives such as film thickness uniformity, paint utilization, and energy consumption in traditional spraying parameter adjustments, this embodiment innovatively applies a "multi-objective collaborative intelligent decision-making algorithm for spraying parameters." By establishing a multi-objective optimization model and an adaptive weighting mechanism, it achieves the global optimal decision for spraying parameters.

[0093] (iv) Construction of Multi-Objective Optimization Model

[0094] The algorithm focuses on film thickness uniformity, coating utilization, and energy consumption as core optimization objectives, constructing a multi-objective optimization model. Spray flow rate, atomization pressure, robot speed, and spraying distance are used as decision variables, and correlation functions are established between each decision variable and the optimization objectives: the film thickness uniformity function is based on a multi-field coupled film thickness prediction model, reflecting the impact of parameter changes on film thickness distribution; the coating utilization function is based on atomization characteristics and trajectory optimization model, quantifying the impact of parameter changes on effective coating deposition; and the energy consumption function establishes a mapping relationship between parameters and energy consumption by collecting real-time energy consumption data from the spraying equipment.

[0095] To achieve synergistic optimization of multiple objectives, a multi-objective optimization objective function is designed. After normalizing the three optimization objectives, a comprehensive objective function is obtained by weighted summation using weighted coefficients. The weighted coefficients are set to fully consider the priority of different production needs. For example, in a quality-priority scenario, film thickness uniformity has the largest weighted coefficient; in a cost-priority scenario, coating utilization rate has the largest weighted coefficient; and in an efficiency-priority scenario, the weighted coefficients related to energy consumption and spraying speed are appropriately adjusted.

[0096] (V) Adaptive Weight Adjustment and Fast Optimization Strategy

[0097] Develop an adaptive weight adjustment mechanism to dynamically adjust the weight coefficients of each objective based on real-time production needs and operational status: obtain the priority of the current production task through the production management system (e.g., efficiency priority for urgent orders, quality priority for high-end products), and automatically match the initial value of the corresponding weight coefficient; during the spraying process, monitor the achievement of each optimization objective in real time. If a certain objective deviates significantly from the expected value, automatically increase the weight coefficient of that objective to guide parameter adjustments toward that objective and ensure the overall balance of multiple objectives.

[0098] To shorten the parameter optimization time, a fast parameter optimization algorithm was developed: by combining the spraying process knowledge graph, constraints on the parameter optimization space were constructed, and parameter combinations that do not meet the process requirements (such as parameters that exceed the equipment's capacity or parameters that easily lead to poor atomization) were pruned to narrow down the optimization range; an improved particle swarm optimization algorithm was adopted, introducing chaotic initialization and adaptive inertial weights to improve the algorithm's global search capability and convergence speed, compressing the parameter optimization time to less than 100ms to meet the requirements of real-time control.

[0099] (VI) Qualitative Explanation of Algorithm Enhancement

[0100] The multi-objective collaborative intelligent decision-making algorithm for spraying parameters achieves synergistic optimization of film thickness uniformity, paint utilization, and energy consumption. In quality-priority scenarios, the film thickness pass rate remains above 99%; in cost-priority scenarios, the unit area spraying cost is reduced by 32%; and in efficiency-priority scenarios, overall energy consumption is reduced by 25%, further shortening the spraying cycle time. The adaptive weight adjustment mechanism allows the algorithm to flexibly adapt to different production needs, improving the flexibility of the production line. The rapid parameter optimization algorithm reduces process parameter debugging time from the traditional 4 hours to 15 minutes, significantly reducing changeover debugging time and improving production line utilization, making it particularly suitable for multi-variety, small-batch production modes.

[0101] Example 4: Application of Real-time Identification and Source Tracing Algorithm for Coating Defects

[0102] To address the issues of slow detection and difficulty in locating the root causes of traditional coating defects, this embodiment innovatively applies a "real-time identification and source tracing algorithm for coating defects." Through high-speed visual inspection and intelligent data analysis, it enables rapid identification, classification, and root cause tracing of defects, and automatically generates corrective solutions.

[0103] (vii) Real-time defect identification and severity assessment

[0104] The algorithm acquires real-time images of wet and dry films using a high-speed vision system, constructing a full-process recognition system of "image acquisition - preprocessing - feature extraction - defect classification": In the image preprocessing stage, adaptive threshold segmentation and denoising algorithms are used to enhance the contrast between the defect area and the background; in the feature extraction stage, feature parameters such as the area, shape, gray value, and edge contour of the defect are extracted to form a defect feature vector; in the defect classification stage, an improved Transformer model is used, which utilizes its powerful feature capture capability to achieve accurate classification of common defects such as drips, pinholes, and orange peel, with a classification accuracy of over 99%.

[0105] To quantify the impact of defects on coating quality, a defect severity assessment index was developed: weighting coefficients were set according to the severity of different defect types, and the severity index of each defect was calculated by combining quantitative parameters such as defect size and quantity; the severity indices of all defects on the workpiece surface were weighted and summed to obtain the overall defect severity assessment result of the workpiece, providing a basis for subsequent processing decisions (e.g., minor defects can be released, while severe defects require rework).

[0106] (viii) Defect source tracing analysis and closed-loop correction

[0107] Establish a defect-parameter correlation analysis model to quickly locate the root cause of defects: Collect time-series data on spraying parameters (such as flow rate, pressure, and speed), environmental parameters (such as temperature and humidity), and atomization characteristic parameters at the time of defect occurrence. Calculate the mutual information value between each parameter and the defect type. The larger the mutual information value, the more significant the parameter's impact on the defect, thus identifying key influencing parameters. For example, sagging defects are usually related to excessive spraying flow rate and insufficient atomization pressure, while pinhole defects may be related to paint viscosity and ambient humidity.

[0108] Develop a closed-loop correction mechanism that automatically generates parameter adjustment plans based on defect type and root cause analysis results: Establish a "defect type-correction strategy" mapping knowledge base, and preset corresponding parameter adjustment ranges for different defect types. For example, when a sagging defect is detected, automatically reduce the spray flow rate by 10-15% or increase the atomization pressure by 5-8%; when an orange peel defect is detected, adjust the paint viscosity or optimize the spraying distance. After the correction plan is issued, monitor the defect changes in real time. If the defect is not eliminated, further optimize and adjust the parameters to form a closed-loop control of "identification-source tracing-correction-verification".

[0109] (ix) Qualitative Explanation of Algorithm Enhancement

[0110] The real-time coating defect identification and traceability analysis algorithm reduces defect detection response time from the traditional 2 hours to 10 seconds, enabling immediate defect detection and handling, effectively curbing the spread of batch defects, and reducing the batch defect incidence rate by 98%. The defect identification accuracy reaches over 99%, avoiding misjudgments and omissions caused by manual inspection; defect traceability time is reduced by 99.7%, significantly decreasing the time spent investigating quality issues. The closed-loop correction mechanism significantly reduces the defect rework rate and customer complaint rate by over 90%, while parameter optimization reduces the recurrence of defects, further improving the quality stability of sprayed products and reducing quality costs.

[0111] In summary, through the synergistic application of the four innovative algorithms mentioned above, a complete intelligent control system for automated industrial robot spraying has been constructed. This system enables intelligent control of the entire spraying process, from film thickness prediction, trajectory planning, parameter decision-making to defect handling. It effectively solves the problems of unstable quality, low efficiency, and high cost in traditional spraying processes, and provides a practical and feasible technical solution for the intelligent upgrading of the spraying industry.

Claims

1. A method for intelligent control and quality optimization of an automated spraying process, characterized in that, include: Collect the workpiece's 3D model, surface temperature field, paint flow rate, atomization pressure, robot motion parameters, and environmental data; Accurate prediction and closed-loop control of coating thickness are achieved through a multi-field coupling-based dynamic prediction and control algorithm. Based on the coating atomization characteristics-driven spray trajectory optimization algorithm and the multi-objective collaborative intelligent decision-making algorithm for spray parameters, the optimal trajectory and parameter combination are generated. A real-time coating defect identification and source analysis algorithm is used to detect surface defects, locate the root cause, and generate parameter adjustment schemes. A quality-efficiency-cost correlation model is established based on historical data, and algorithm parameters are continuously iterated and optimized to form a closed-loop control.

2. The method according to claim 1, characterized in that, The multi-field coupling-based dynamic prediction and control algorithm for film thickness establishes a film thickness prediction hybrid model h(x,y,z,t)=fCFD(u,v,p,T)+fDL (Δhhist ), achieving a film thickness control accuracy of ±3% and a uniformity index Q1 ≥0.

98.

3. The method according to claim 1, characterized in that, The film thickness dynamic prediction and control algorithm based on multi-field coupling includes a surface partitioning processing algorithm, which divides the workpiece surface into different regions according to the curvature κ to implement differentiated control, thereby improving the coating performance stability by ≥40%.

4. The method according to claim 1, characterized in that, The spray trajectory optimization algorithm driven by the atomization characteristics of the coating establishes a coating utilization rate evaluation model η=∫∫Aworkpiece ρ(x,y)dxdy / ∫∫Aspray ρ(x,y)dxdy, which improves the coating utilization rate to ≥85%.

5. The method according to claim 1, characterized in that, The coating atomization characteristic-driven spray trajectory optimization algorithm adopts a variable spacing trajectory planning strategy, with path spacing s = k·D32 ·cos(θ / 2) / κα, resulting in a VOCs emission reduction of ≥40%.

6. The method according to claim 1, characterized in that, The multi-objective collaborative intelligent decision-making algorithm for spraying parameters includes an objective function. F = ω1 (1-Q1) + ω2(1-η) + ω3 (E / E0), using an adaptive weight adjustment mechanism, the overall energy consumption is reduced by ≥25%.

7. The method according to claim 1, characterized in that, The multi-objective collaborative intelligent decision-making algorithm for spraying parameters develops a rapid parameter optimization algorithm, which, combined with process knowledge graph pruning optimization space, achieves an optimization time of ≤100ms and reduces the process parameter debugging time to within 15 minutes.

8. The method according to claim 1, characterized in that, The proposed real-time coating defect identification and source tracing analysis algorithm proposes a defect severity assessment index S = ∑wi ·si, and adopts an improved Transformer model, achieving a defect identification accuracy of ≥99%.

9. The method according to claim 1, characterized in that, The real-time identification and tracing analysis algorithm for coating defects includes a defect-parameter correlation analysis model. Based on mutual information MI(f,d)=I(F;D), it locates key influencing parameters, reducing defect tracing time by ≥99.7%.

10. The method according to claim 1, characterized in that, The data collected in step 1) is transmitted in real time via an industrial bus with a transmission delay of ≤1ms. A spraying process database is established to store historical process parameters, sensor data, quality inspection results, and energy consumption indicators.

Images