Study and optimisation of WEDM parameters of AISI P20+Ni using RSM and hybrid deep neural network

Abstract

ABSTRACTThe main task in Wire Electrical Discharge Machining (WEDM) is to select the proper tool material for machining the workpiece to get better surface finishing and high cutting speed. To overcome these challenges, nowadays, coated electrodes are used. The main objective of this research work is to study the impact of certain input parameters such as pulse off-time, pulse on-time, servo voltage, and peak current in WEDM on Material Removal Rate (MRR), Recast Layer Thickness (RLT), and Surface Roughness (SR) output parameters by using zinc-coated brass wire as an electrode and AISI P20+Ni (DIN 1.2738) as the workpiece. One of the hardest materials used to make die for plastic moulds. The study is carried out experimentally and then compared with the design done by response surface methodology (RSM) using ANOVA and hybrid deep neural network (DNN) predicted values, which are optimised based on Manta ray foraging optimisation (MRFO) to obtain more precise outputs. The developed RSM and hybrid DNN+MRFO were designed in Design Expert 12 and MATLAB R2020a platforms and achieved 90% average prediction accuracy in the DNN+MRFO method. The experimentally obtained maximum MRR and minimum RLT and SR are 2.657 mm3/min, 7.9 µm, and 0.351 µm, respectively. The regression results showed that DNN+MRFO outperforms the RSM model. The achieved desirability at confirmatory analysis for RSM and DNN+MRFO are 0.844 and 0.921, respectively.