Prediction of Cutting Temperature using an Artificial Neural Network and Response Surface Methodology in turning Al based Metal matrix composite with coated carbide insert

Abstract

The objective of this paper is to develop two predictive models for tool-workpiece interface temperature in turning 5052 Al based metal matrix composite with coated carbide insert during both dry and wet conditions using artificial neural network and response surface methodology. In ANN model, a feed forward multilayer neural network with twenty neurons was found as the optimum network for wet model. In RSM model, the analysis of variance showed that feed rate is the most influential parameter which affects the cutting temperature followed by cutting speed. Two quadratic equations were formulated which can be used to predict the temperature for both dry and wet models. The values of regression coefficient (R²) prove the ability of two predictive models to predict accurately. ANN model exhibits less mean absolute percentage error (MAPE) than RSM model. So it can be concluded that ANN models can predict better than RSM models.