Assessment of Machining Performance for Intelligent Tool Wear Prediction Using Hybrid Extreme Learning Machine

The increasing utilization of Ni-based superalloys in aerospace and power generation industries presents significant machining challenges due to their unique properties. This study introduces a hybrid extreme learning machine (ELM) to predict tool wear during face milling of Inconel X750 using multiple sensors, aiming to optimize machining processes and extend tool lifespan. Using two types of cutting tool inserts (Re =0.8 mm and Re =0.4 mm), the study employs a combination of stationary wavelet transform (SWT) and largest Lyapunov exponent (LLE) for signal processing and feature extraction. The primary performance characteristic, tool wear, is analyzed using ANOVA tests, and the predictive model’s accuracy is validated with real-world data. The multilayer perceptron ELM (MLELM) optimized with the slime mold algorithm (SMA) demonstrates low root-mean-square errors of 0.08 for T1 and 0.09 for T2, with minimum percentage errors of 0.82% for T1 and 0.68% for T2. These findings provide valuable insights for enhancing machining processes, contributing to more efficient, sustainable, and cost-effective manufacturing. This research underscores the importance of intelligent tool condition monitoring in advancing manufacturing technologies. The proposed model’s ability to optimize tool wear prediction can significantly improve manufacturing efficiency and reduce downtime in industrial applications. This research underscores the importance of intelligent tool condition monitoring in advancing manufacturing technologies and achieving sustainability goals.