Development of ANN and ANFIS Models for Prediction of Tool Wear in High-Speed Milling

Abstract

In precision machining, tool wear is one of the primary factors affecting machining quality and production efficiency. This study developed intelligent prediction models for tool wear in the high-speed milling (HSM) of AISI 1045 medium-carbon steel by integrating robust process design with backpropagation neural networks (BPNN) and adaptive neuro-fuzzy inference systems (ANFIS). Robust process design was employed to optimize the machining parameters, while the hyperparameters of both BPNN and ANFIS models were systematically optimized to improve prediction performance. Tool wear was measured after a fixed cutting length and used to establish the prediction models. The optimized machining parameters reduced tool wear by 53% compared with the worst experimental condition. The optimized BPNN model achieved a prediction accuracy of 96.68%, whereas the ANFIS model with Gaussian membership functions achieved 100%, demonstrating superior predictive performance. The proposed approach effectively combines robust process design and intelligent prediction models to accurately predict tool wear using a limited experimental dataset, providing an efficient methodology for tool wear prediction and machining parameter optimization in intelligent manufacturing applications.