Film Thickness Prediction Based on Digital Twin and Neural Network

With the widespread adoption of industrial robots in the spray coating industry, spraying production efficiency has significantly improved, and related robotic spraying technologies are constantly evolving. To improve economic efficiency, high-quality robotic spraying production has emerged as a result. However, due to the unique structure of the sprayed workpiece and the multiple coupled process parameters, studying the relationship between process parameters that influence spray quality and the spraying process performance is challenging. To address this, this paper proposes a digital twin method for characterizing spray film thickness based on a predictive model. Under defined operating conditions, a mathematical model of spraying is constructed and combined with a neural network-based spray film thickness prediction model. This prediction model derives the mapping relationship between different robotic process parameters and film thickness. Finally, a spraying experimental platform was constructed to verify the accuracy of the prediction model. Regular workpieces were sprayed using the selected process parameters. Film thickness sampling points were measured after spraying, and the maximum error was $5.35 \mu \mathrm{m}$. This method can predict spraying performance with only a small number of spraying data samples, enabling prediction of process inputs. This method reduces workpiece waste while providing accurate results.