Prediction of Workpiece Film Thickness via Multi-Region Segmented Model of Painting Process Parameters

This study proposes a method to solve the problem of predicting the coating film thickness of the workpiece, which is called the multi-region segmentation model. First, four types of sensing data contribute 125 features, including Clean, Oven, Painting, and Environment. Data preprocessing covers several aspects, such as missing values, outliers, and scales. Then, the key features are extracted and given to the machine learning algorithm to build a model, verify, and test. In addition, the multi-region segmentation model is the main idea and aims to reduce the model from falling into the trap of overfitting when modeling. At the same time, the classifier guides the test data to a more suitable regressor. The experimental results show that the multi-region segmentation model based on the Pearson correlation coefficient has obtained a relatively ideal performance. The RMSE is 15.3724. This is superior to the results submitted for IMBD Competition 2022 and the official standards. In future research work, we will devote ourselves to strengthening the strategy of data pre-processing, and it is expected to improve the model's error.