Taguchi integrated Least Square Support Vector Machine an alternative to Artificial Neural Network analysis of electrochemical machining process

Abstract

The important performance parameter such as material removal rate(MRR) and surface roughness(SR) are influenced by various machining parameters namely voltage, feed rate of electrode and electrolyte flow rate in electrochemical machining process(ECM). In machining, the process of modelling and optimization are challenging tasks to qualify the requirements in order to produce high quality of products. There are a lot of modelling techniques that have been discovered by researchers. In this paper, the optimize settings of performance parameters; surface roughness and MRR are done by the Taguchi technique and the experimental result of MRR and SR was predicted by the Multi-layer Feed forward Neural network (MFNN) and Least square support vector machine (LSSVM). For Taguchi analysis three process parameters and two responses, MRR and SR were considered by L18 orthogonal array design and ANOVA result were performed. EN19 material used as the work piece for the experiment. After evaluating MFNN and LSSVM models, the best network found to be Least square support vector machine with RBF kernel. The mean square errors (MSE) between actual and predicted response obtained in both LSSVM model and MNFF model for the training and for testing datasets were concluded that LSSVM as more powerful machine learning tool and predict the MRR and SR successfully compared to other models. The performance of LSSVM is depend different kernel function that can separate the data from hyper plane for better prediction however we use Linear and RBF kernel. RBF kernel gives better prediction of MRR and SR with minimum MSE. Keywords - ANOVA, EN-19 tool steel, LS-SVM, MFNN, Taguchi technique