Investigation of Optimum Process Parameters Using Genetic Algorithm Based Neural Networks during EDM of 64WC-9Co

Abstract

Conductive ceramics like cobalt bonded tungsten carbide which is categorized as with high mechanical and physical properties are usually known to create major challenges during conventional and non-conventional machining. Electrical discharge machining (EDM) which is very prominent amongst the non-conventional machining methods is expected to be used quite extensively in machining it due to the favorable features and advantages that it can offer. This project was undertaken to study the machining performance of EDM with tungsten carbide using copper and graphite as electrodes. The effect of varying the machining parameters on the machining responses such as material removal rate (MRR), electrode wear ratio (EWR), was investigated. The experimental plan for both processes was conducted first on exploratory based and focused experiment conducted according to the design of experiments (DOE) and the results were statistically evaluated using analysis of variance (ANOVA). Taguchi methodology was employed in evaluating the machining performance of the SEDM process and mathematical models for MRR, EWR were developed. For verification of model results, conformation runs have been conducted. Results show that peak current was the most significant parameter that influenced the machining responses on EDM. Artificial neural network and Genetic algorithm neural network based multi objective optimization implemented for maximization of MRR and minimization of EW has been done by using the developed empirical models. Optimization results have been used for identifying the effect of control parameters on responses that developed data sets which give the performance of parameters of SEDM on conductive ceramics.