Optimization of wire-cut EDM parameters using artificial neural network and genetic algorithm for enhancing surface finish and material removal rate of charging handlebar machining from mild steel AISI 1020

Wire-cut electrical discharge machining (wire-cut EDM) was the primary method for producing Bren gun charging handlebars due to its high precision demands. However, the frequent breakage and damage of the wire electrode presented significant obstacles, negatively impacting surface quality, production time, and manufacturing costs. To address these challenges, an L9 orthogonal array was employed to investigate surface quality, material removal rate, and wire surface quality. Following experimentation, fresh and broken electrodes underwent examination using a scanning electron microscope (SEM). Surface damage and roughness were analyzed through SEM observations and histogram mean values. Initially, Artificial Neural Network (ANN) models were developed. The optimal ANN model (4–9-2) was then integrated with a multi-objective Genetic Algorithm (GA) for optimization. This combined approach yielded the following optimal parameter combinations: peak current of 2.513 A, pulse on time of 25.642 µs, wire feed rate of 9.999 m/min, and pulse off time of 7.975 µs. Three confirmatory experiments were conducted using these optimized parameter values. The errors between the GA-predicted and confirmatory results for surface roughness (Ra) and material removal rate (MRR) were 1.19% and 2.29%, respectively. Compared to the company’s previous performance, the confirmatory results demonstrated a 39.37% improvement in Ra while maintaining wire integrity, albeit with a 10.63% decrease in MRR. This research highlights a successful approach to enhancing machining quality and optimizing the wire-cut EDM process for defense applications.