Indian Journal of Science and Technology

Abstract

Background/Objectives: Modelling and optimization of machining process is recognized to be an extremely challenging research area in current scenario. This study illustrates work suggestion, an intellectual approach in solving multi-response optimization problem involving Electrical Discharge Machining (EDM) of LM25 Al composite using Response Surface Methodology (RSM) combined with Radial Basics Function Neural Network (RBFNN) and Artificial Neural Network (ANN) techniques. Methods/Statistical Analysis: An experimental analysis was carried out in establishing the most significant machining parameters that throw in to MRR and SR. The experimental plan for these investigations was conducted according to the RSM. RBFNN and ANN is a computational intelligence model that consists of nodes that are interlinked. The optimization of EDM is performed by preferring input process parameters like discharge voltage, current, pulse-on time, pulse-off time, oil pressure and spark gap, and also output responses as Material Removal Rate (MRR) and Surface Roughness (SR) using Box - Behnken method. Findings: Each node performs a simple operation in computing its output from its input that is transmitted through links connected to other links. This is comparatively simple computational model because of the analogous structure that of neural system in human brain-nodes equivalent neurons and links corresponding to synapses that transmit signals between neurons. Back Propagation Neural Network (BPNN) is utilized to train the network for optimizing the EDM parameters. By simulation the result was authenticated with the target output awaiting the network error has congregated to threshold minimum. Applications/Improvements: Multi-response MRR and SR modelling were performed in the EDM process and via the investigation on the experiment the results are confirmed. Different process parameters consequences have also been premeditated.

Keywords: Electrical Discharge Machining, Metal Removal Rate, Neural Network, Response Surface Methodology, Radial Basics Function Neural Networks, Surface Roughness