Alireza Nikravan - Department of Mechanical Engineering, Technical and vocational University, Mashhad, Iran;
Farhad Kolahan - Associate Professor, Ferdowsi University of Mashhad, Mashhad, Iran;

In the present study artificial neural network (ANN)along with particle swarm optimization (PSO) andsimulated annealing (SA) algorithms have beenemployed for modeling and optimization of electricaldischarge machining (EDM) process of AISI۲۳۱۲ hotworked steel parts. The process input parametersconsidered here include voltage (V), peak current (I),pulse off time (Toff), pulse on time (Ton) and duty factor(η). The process quality measures are surface roughness(SR), tool wear rate (TWR) and material removal rate(MRR). The objective is to determine a combination ofprocess parameters to minimize TWR and SR andmaximize MRR independently (as single objective) andalso simultaneously as multi-criteria optimization. Theexperimental data are gathered based on Taguchi L۳۶orthogonal array design of experiments. The threeperformance characteristics (MRR, TWR and SR)obtained from experimental tests. Then, the outputs areused to develop the artificial neural network (ANN)model. Next, in order to determine the best set ofprocess parameters values for a desired set of processquality measures the developed ANN model isembedded into heuristic algorithms (SA and PSO) andtheir derived results have been compared. Validation ofthe results has been carried out through a series ofexperimental test run under the optimal machiningconditions. It is evident that the proposed optimizationprocedure is quite efficient in modeling andoptimization of EDM process parameters.

Electrical discharge machining (EDM),Taguchi technique, Design of experiments (DOE),artificial neural network (ANN), simulated annealing (SA)algorithm, particle swarm optimization (PSO) algorithm.