Comparative analysis of predictive modeling techniques for micro-EDM of Ti-6Al-4V alloy: Unveiling the best approach

Abstract

Micro-Electrical Discharge Machining (Micro-EDM) has emerged as a crucial technique for fabricating intricate micro-features in advanced alloys such as Ti-6Al-4V. The present study investigates the effects of gap voltage, peak current, pulse-on time, and flushing pressure on material removal rate (MRR), tool wear rate (TWR), taper, and overcut during micro-hole drilling. A total of 31 experiments were conducted using a central composite design, and the experimental data were used to develop predictive models based on multivariate regression analysis, fuzzy logic, artificial neural networks, and an adaptive neuro-fuzzy inference system (ANFIS). The models were evaluated using five statistical indices: mean absolute percentage error (MAPE), root mean squared log error, root mean squared percentage error, root relative squared error, and correlation coefficient (R). Results show that ANFIS provides the most accurate predictions, with MAPE values below 5% for MRR and TWR and R > 0.95 across all responses, outperforming other approaches. The novelty of this work lies in its comparative framework, which highlights the superiority of ANFIS in capturing nonlinear input-output relationships in micro-EDM. The study concludes that ANFIS can be effectively applied for optimizing micro-EDM parameters, and future research should extend this methodology to other difficult-to-machine materials and explore advanced ANFIS variants for improved machining performance.