Modelling and multi-response optimisation of sinking micro-electrical discharge machining of AISI 4140 steel

Abstract

Improving productivity and surface quality in sinking micro-electrical discharge machining (S-MEDM) process is still a challenging task for manufacturing engineers. In order to fulfil these requirements, modelling and optimisation of the S-MEDM process is needed. In this paper, a prediction model for S-MEDM process has been developed using the coupled methodology of finite element method (FEM) and artificial neural network (ANN). First, a FEM-based model is developed incorporating realistic aspects such as Gaussian heat flux distribution, time dependent spark radius, temperature dependent workpiece material properties and phase change phenomenon. Further, an ANN model is developed utilising the data generated by FEM-based model for training and testing of the network. The trained ANN model has been used for the prediction of the material removal rate and surface roughness (Ra). Weighted principal component (WPC) method is effectively employed for multi-response optimisation of S-MEDM process using prediction data of ANN model. The optimum setting of process parameters gives an improvement of 22.68% in material removal rate and a decrease of 17.30% in surface roughness.