Electrical Discharge Drilling (EDD) is widely used for machining difficult-to-cut titanium alloys. However, thermal damage in the form of recast layer thickness (RLT) significantly affects the surface integrity and reliability of drilled components. This study develops an artificial intelligence-based predictive framework using Adaptive Neuro-Fuzzy Inference System (ANFIS) integrated with Fuzzy C-Means (FCM) clustering and hybrid optimization techniques to model RLT during EDD. Experiments were conducted using a full factorial design considering discharge current, pulse-on time, pulse-off time, and dielectric pressure. A total of 81 experiments were performed, where 60 datasets were used for training and 21 for testing. Among all models, ANFIS achieved the best prediction accuracy (Mean average error = 0.0497, Root mean square error = 0.041, Variance accounted for = 82.33%, and coefficient of determination R 2 = 0.816), outperforming hybrid FCM models ( R 2 ≈ 0.56–0.68). Parametric analysis revealed that higher discharge current and pulse-on time increase RLT, whereas higher pulse-off time and dielectric pressure reduce RLT due to improved cooling and flushing. The proposed intelligent framework enables reliable prediction of recast layer formation and supports optimization of EDM drilling parameters for improved surface integrity of titanium alloy components.
Arif et al. (Wed,) studied this question.