An investigation on the prediction of thin alpha case removal in high pressure plain waterjet cleaning process using GA/PSO optimized neural networks and regression methods

摘要

High pressure plain waterjet (PWJ) has been considered a promising method for the removal of hard metallic coatings such as titanium alpha case layers due to its distinct advantages. The process modelling of alpha case removal using PWJ is of significance to the optimization of process parameters to improve productivity. However, the non-linear material properties of the alpha case layer place significant challenges on the process modelling. In the present study, evaluation algorithm optimized artificial neural networks and multilinear stepwise regression models were developed separately to predict the process outputs of PWJ alpha case removal, based on the data collected from a full factorial experimental design. As compared to regression models, the results presented from the proposed neural network models demonstrated a significant improvement in prediction accuracy. Thus, artificial neural networks can provide satisfactory models for the prediction of titanium alpha case removal.