ANN and Machine Learning based predictions of MRR in AWSJ Machining of CFRP composites

Abstract

The study assesses the efficacy of Abrasive Water Suspension Jet (AWSJ) Machining, a non-conventional technique based on erosion principles, with a specific emphasis on its use in machining carbon fiber-reinforced plastics (CFRP) composites. The analysis examines critical process variables, including Speed, Feed, and Standoff distance, to evaluate their influence on Material Removal Rate (MRR), during underwater cutting operations. The results unambiguously support the superiority of underwater cutting. Expanding the diameter of the jet in underwater cutting improves both the width of the cut and the roughness of the surface. This also helps reduce vibrations in the nozzle when operating at high pressures, resulting in a smaller cut and a smoother surface. This highlights the effectiveness of underwater cutting in generating accurate machining results. In addition, the study utilizes machine learning (ML) models such as Random Forest and XGBoost to enhance the optimization of MRR, a crucial parameter in composite machining. The results demonstrate exceptional performance across all models, with XGBoost exhibiting outstanding accuracy and efficiency on both the training and test datasets. The comparative analysis reveals the competitive performance of Random Forest XGBoost and Artificial Neural Network (ANN) in optimizing MRR. These models achieve notable accuracy scores in both training and test sets, surpassing the regular statistical methods such as the Response Surface Methodology (RSM).