INTELLIGENT HYBRID OPTIMIZATION OF WEDM FOR Ti-6Al-4V USING TAGUCHI DEEP LEARNING AND ANOVA FOR ENHANCED SURFACE INTEGRITY AND PREDICTION ACCURACY

Wire Electrical Discharge Machining (WEDM) is a high-precision, non-traditional manufacturing process crucial for machining advanced materials such as Ti-6Al-4V, extensively used in aerospace and biomedical applications. This study introduces a novel hybrid optimization framework that integrates Taguchi’s method, Deep Neural Networks (DNNs), and Analysis of Variance (ANOVA) to enhance machining efficiency and surface integrity. A Taguchi L27 orthogonal array was employed to systematically evaluate the influence of key process parameters – pulse-on time (Ton), pulse-off time (Toff), and wire feed rate (Wf) – on surface roughness (Ra). Experimental trials conducted on an ELECTRONICA Sprintcut 734 WEDM machine, utilizing deionized water and a 0.5 mm brass wire electrode, revealed that Ton was the most dominant factor, contributing 38.02 % to the Ra variation, followed by the Ton-Toff interaction (27.80 %), while Toff (7.28 %) and Wf (4.92 %) had comparatively low impacts. The optimal machining parameters (Ton = 3 µs, Toff = 4 µs, Wf = 2 m/min) achieved a minimum Ra of 1.50 µm, significantly improving the surface finish. A trained ANN model demonstrated a predictive accuracy exceeding 95 %, validating its effectiveness in process modeling. ANOVA results (p = 0.05) confirmed the statistical significance of process parameters, reinforcing the reliability of the proposed optimization approach. This research establishes a robust, data-driven methodology for precision machining control, outperforming conventional single-method optimizations. Future studies could explore real-time adaptive control, hybrid machining strategies, and AI-driven predictive analytics to further advance WEDM performance in high-precision manufacturing applications.