A Combined Taguchi–ANOVA and Deep Learning Approach for Optimizing Die-Sinking EDM of Pentagonal Cavities over EN-24 Alloy Steel

Electrical Discharge Machining (EDM) is one of the most commonly adopted non-traditional manufacturing processes for producing precise and complex geometries in hard-to-machine materials. In this paper, an optimal system using Combined Taguchi–ANOVA and Deep Neural Network (DNN) techniques is suggested to enhance machining effectiveness and increase prediction accuracy. In this study, EDM was employed to machine pentagonal cavities on EN-24 alloy steel using a copper tool electrode. The machinability of the alloy was investigated using the Taguchi method, considering three critical process parameters: discharge current, pulse-on time, and pulse-off time. Material removal rate (MRR), tool wear rate (TWR) and surface roughness (SR) were evaluated as key performance measures to assess the overall machining efficiency and quality. Signal-to-noise (S/N) ratio analysis was done to explore the effect of input parameters on output responses. The study findings indicated that discharge current was the dominant factor for all performance parameters and greatly impacted both the material removal and tool wear behaviour. Moreover, The ANOVA method was applied to measure the impact of each process parameter which validated the findings based on S/N ratio and established a statistical significance between process factors. The best machining conditions found using S/N ratio and ANOVA revealed 9 A, 250 μs pulse-on and 35 μs pulse-off for maximal MRR, whereas 5 A, 50 μs pulse-on and 25 μs pulse-off were the best for minimizing both TWR and SR. The developed DNN model demonstrated the capability to predict MRR, TWR and SR within the investigated experimental domain. The proposed framework offers a data-based way for modelling and optimizing EDM process performance.