Support vector machine regression for predicting dimensional features of die-sinking electrical discharge machined components

Abstract

Die-sinking electrical discharge machining produces components with low repeatability as the process is inherently stochastic. Effects of its inputs and process parameters on the components’ dimensions are difficult to predict. This paper investigates the influence of input parameters like gap voltage, current, and pulse characteristics like percentage of “open”, “normal”, “arc” and “short” pulses on the dimensional features of the machined components. It discusses the methodology for extraction and estimation of amount of area machined, undercut and dimension by image processing. Support vector machine regression is applied to predict the dimension features based on the input and condition parameters.