Predictive analysis and multi objective optimization of wire-EDM process using ANN

Abstract

Metal Matrix Composites (MMCs) plays very important role in structural, aerospace and automotive industries, because of their good properties like light weight, high specific strength and good wear resistance. Among various nonconventional machines, wire EDM is widely used in cutting of hard materials with good mechanical properties and high accuracy. So, in this paper, composite Al7075 + 10%Al2O3 is selected as material and stir casting process is used in composite fabrication and machining is done using Wire EDM. Increasing the Material Removal Rate (MRR) value and decreasing the Surface Roughness (SR) are the main objectives of this paper. There are five input parameters such as pulse-on, voltage; pulse-off, bed speed and current are considered to study the output parameters and to achieve the objectives. Performance analysis of the experimental values is done using Artificial Neural Network (ANN) mathematical model and the model is evaluated using R-square value. Artificial Neural Network model gives better and accurate results compared with linear regression model. In practical situation, it is hard to find the best for all the objectives simultaneously. Pareto optimal solutions are non-dominated solutions when all the objectives are considered. So, in this study, Solution for multi objective optimization is determined based on Pareto optimal optimization.