Title: An optimised back propagation neural network approach and simulated annealing algorithm towards optimisation of EDM process parameters

Authors: Masoud Azadi Moghaddam; Farhad Kolahan

Addresses: Department of Mechanical Engineering, Ferdowsi University of Mashhad, P.O. Box 91775-1111, Mashhad, Iran ' Department of Mechanical Engineering, Ferdowsi University of Mashhad, P.O. Box 91775-1111, Mashhad, Iran

Abstract: The present research addresses the multi-criteria modelling and optimisation of electrical discharge machining (EDM) process, via optimised back propagation neural networks (OBPNN) and simulated annealing (SA) algorithm. The process response characteristics considered are material removal rate, surface roughness, and tool wear rate. The process input parameters include voltage, peak current, pulse off time, and pulse on time and duty factor. The three performance characteristics are combined into a single objective using weighted normalised grades (WNG) obtained from experimental study based on Taguchi method, to develop the artificial neural network (ANN) model. In order to enhance the prediction capability of the proposed model, its architecture is tuned by SA algorithm. Next, the developed model is embedded into SA algorithm to determine the best set of process parameters values for an optimal set of outputs. Experimental results indicate that the proposed optimisation procedure is quite efficient in modelling and optimisation of EDM process parameters. [Received 25 January 2015; Revised 12 April 2015; Accepted 3 May 2015]

Keywords: electrical discharge machining; EDM parameters; Taguchi methods; multicriteria optimisation; artificial neural networks; ANNs; back propagation neural networks; simulated annealing; electro-discharge machining; modelling; material removal rate; MRR; surface roughness; surface quality; tool wear rate; voltage; peak current; pulse off time; pulse on time; duty factor.

DOI: 10.1504/IJMR.2015.071616

International Journal of Manufacturing Research, 2015 Vol.10 No.3, pp.215 - 236

Received: 14 Feb 2015
Accepted: 03 May 2015
Published online: 04 Sep 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article