A support vector machine-based prediction model for electrochemical machining process

Abstract

Manufacturing of quality products is one of the core measures to address competitiveness in industries. Hence, it is always necessary to accomplish quality prediction at early stages of a manufacturing process to attain high quality products at the minimum possible cost. To achieve this goal, the past researchers have developed and investigated the applications of different intelligent techniques for their effective deployment at various stages of manufacturing processes. In this paper, support vector machine (SVM), a supervised learning system based on a novel artificial intelligence paradigm, is employed for prediction of three responses, like material removal rate, surface roughness and radial overcut during an electrochemical machining (ECM) operation. Gaussian radial basis kernel function is adopted in this algorithm to provide higher prediction accuracy. Regression analyses are also carried out to validate the effectiveness of these prediction models. The SVM-based results show good agreement between the experimental and predicted response values as compared to linear and quadratic models. Among the four ECM process parameters, i.e. applied voltage, tool feed rate, electrolyte concentration and percentage of reinforcement of B4C particles in the metal matrix, tool feed rate is identified having the maximum influence on the considered responses.