Adaptive AI-based prediction and optimization of milling parameters using ANFIS, ANN, and PSO for 316L stainless steel

This study presents the development of an adaptive, artificial intelligence-assisted module designed to enhance productivity and improve cutting conditions in milling operations. The proposed system integrates Artificial Neural Networks (ANN), an Adaptive Neuro-Fuzzy Inference System (ANFIS), and Particle Swarm Optimization (PSO) within a unified framework to enable real-time adaptation to variable machining conditions. Experimental studies conducted on 316L stainless steel yielded a dataset comprising 241 training and 29 testing samples, including cutting forces, tool wear, and surface roughness measurements. The feed-forward backpropagation ANN model, trained using the MATLAB Neural Network Toolbox, demonstrated high prediction accuracy, achieving an exceptionally low mean error of 0.031% in estimating cutting forces. Optimization of the key machining parameters—cutting speed, feed rate, and depth of cut—was performed using the PSO algorithm, considering constraints related to surface integrity, tool life, and power limitations. The developed module reliably predicts tool life, tool wear, and cutting forces, while dynamically optimizing the machining parameters to reduce production time, increase material removal efficiency, and enhance surface quality. Comparative analysis results indicate that the proposed AI-assisted module provides significantly superior performance over conventional catalog-based parameter selection. Specifically, the system delivers a 28.5% increase in material removal rate, a 14.8% improvement in tool life, an 18.3% reduction in surface roughness, and a 36.3% decrease in total production time. These outcomes demonstrate that the presented approach serves as a robust and practical solution for real-time process control and productivity enhancement in industrial milling operations.