Experimental study and machining parameter optimization on powder-mixed EDM of Nimonic 901 using feed-forward backpropagation neural networks

The earliest and most widely utilized procedure of unconventional machining is referred to as electric discharge machining (EDM), which is a process of non-contact machining for eliminating the material from the workpiece (i.e. electrically conductive) with the usage of electric discharge series. This article proposes an experimental investigation and optimal selection of input process parameters involved in powder-mixed EDM (PM-EDM) process of a workpiece made of Nimonic 901 alloy with an electrode of copper-tungsten (Cu-W). In addition, silicon carbide (SiC) is used as a powder particle due to its more resistance to mechanical steadiness at greater temperatures, high hardness, exceptional thermal conductivity, low coefficient of thermal expansion, corrosion and oxidation etc. The impact of input process variables like concentration of mixed powder (C-p), servo voltage (V-s), peak current (I-p) and pulse-on-time (T-on) on output response features assessed in terms of tool wear rate (TWR), surface roughness (SR) and material removal rate (MRR) was also examined. Furthermore, the Taguchi design approach with L-18 orthogonal array is employed to find the optimal combination of process parameters using the analysis of signal-to-noise (S/N) ratio. Furthermore, back-propagated neural network (BPNN) with feed-forward (FF) architecture is utilised to determine the approximate solutions and best fit for the optimization and search issues. Finally, the findings of the experimental MRR (E-MRR) confirmation test are compared to the MRR values derived using FF-BPNN model, i.e. P-MRR. Similarly, the experimental SR (E-SR) and experimental TWR (E-TWR) results were also compared to the SR and TWR computed utilizing FF-BPNN model, i.e. P-SR and P-TWR, respectively.