Artificial intelligence for experimental investigation and optimal process parameter selection in PM-EDM of nimonic alloy 901

Electrical discharge machining (EDM), a widely used non-contact machining method, employs electric discharge to remove conductive material from workpieces. This study focused on experimentally investigating and optimizing input process parameters for the PM-EDM process of a Nimonic alloy 901 (NA-901) workpiece with a silver electrode. The silicon carbide (SiC) powder particles were explored for their exceptional properties, including high temperature resistance, hardness, thermal conductivity, and resistance to corrosion and oxidation. The study evaluated the impact of input process parameters such as servo voltage ($V_s$), powder concentration ($C_p$), pulse-on-time ($T_{\mathrm{on}}$), and peak current ($I_p$) on surface roughness rate (SRR),tool wear rate (TWR), and material removal rate (MRR). The Taguchi design approach with an L18 orthogonal array was used to identify the optimal parameter combination based on signal-to-noise (S/N) ratio analysis. To improve optimization, a feed-forward backpropagation neural network (FF-BPNN)was utilized to approximate solutions. The results of the experimental MRR confirmation test (E-MRR) were compared to the MRR values predicted using the FF-BPNN model (P-MRR). Similarly, the SRR (E-SRR) and the TWR (E-TWR) were compared to the predicted SRR and TWR obtained from the proposed FF-BPNN model. In summary, this study presents an experimental examination and optimization of input process parameters in the PM-EDM process of an NA-901 workpiece with a silver electrode.The use of SiC powder particles, the impact of input process parameters, and optimization were explored using Taguchi design and FF-BPNN techniques. This study's results demonstrate these approaches' effectiveness in achieving optimal PM-EDM process parameters.Finally, results revealed E-MRR as 6.894, P-MRR as 6.8913, E-SRR as 0.891, P-SRR as 0.897, E-TWR as 0.116, and P-TWR as 0.015.