Research on the prediction method for paint film quality in auto-mated robotic spraying based on the improved DEWOA-ANFIS model

Replacing traditional manual spraying with robotic systems reduces inefficiencies, improves quality consistency, and mitigates occupational health risks, making it a significant advancement in industrial modernization. Achieving human-level expertise in automated systems requires the integration of fuzzy logic and neural networks to replicate expert decision-making processes. An augmented adaptive neuro-fuzzy inference system (ANFIS) has been developed and optimized using the Differential Evolutionary Enhanced Whale Optimization Algorithm (DEWOA). The developed model establishes correlations between key spraying parameters and quantitative quality metrics, demonstrating excellent predictive performance for automated waterborne paint spraying. Quantitative validation further confirms the model’s superior predictive performance. Experimental results demonstrate that the DEWOA-ANFIS model predicts the average film thickness and cross-sectional thickness difference with MAEs of 0.431 and 0.445 μm, respectively, with errors within ±1 μm. This performance is significantly better than that of other comparative models. The model exhibits not only excellent prediction accuracy but also achieves a training time of just 1.05 s and an inference time of 6.85 ms, with minimal memory usage, highlighting its high computational efficiency and resource utilization. Robustness validation demonstrates that DEWOA-ANFIS remains stable under abnormal data perturbations ranging from 5% to 10%, with consistent results across multiple iterations, ensuring strong adaptability. Additionally, the model demonstrates robust resistance to real disturbances, such as high-intensity noise, nonlinear noise, and local anomalies, effectively addressing common disturbances and uncertainties in industrial processes. Furthermore, based on spraying speed, height, pressure, and the prediction model, a heatmap analysis of the nonlinear coupling effects is created, aligning with actual influence patterns and offering valuable guidance for spray process optimization. Therefore, the DEWOA-ANFIS model enhances the prediction accuracy and operational reliability of robotic spraying, enabling intelligent adaptive process control in complex manufacturing environments. Future research will incorporate dynamically changing environmental factors, explore hybrid gray-box modeling techniques, integrate real-time monitoring and feedback systems, and enhance the model’s applicability and stability across diverse industrial settings.