Neural Network-Based Optimization of the Electrical Discharge Drilling Process Parameters

Abstract

The current work aims to optimize process parameters, including the current (I), voltage (V), pulse on time (O), and pulse off time (F) of the electrical discharge drilling of the small hole (EDDH) to reduce the dilation of the drilled hole (DH) as well as the tapper of the drilled hole (HT) and enhance the material removal rate (MRR). The radial basis function network (RBFN) was used to develop EDDH responses, while the modified quantum-behaved particle swarm optimization algorithm (MQPSO) was applied to produce feasible solutions. The evaluation by an area-based method of ranking (EAMR) approach was used to select the best optimality. The obtained results indicated that the optimal I, V, O, and F are 5 A, 60 V, 40 μs, and 45 μs, respectively. The DH and TP are reduced by 32.8% and 28.0%, while the MRR is improved by 66.3%. The RBFN models could be effectively applied to present non-linear data. The DH and TP models were significantly affected by the O and F, while the I and V had effective influences on the MRR. The outcomes could be used to improve the drilled quality indicator and productivity in industrial EDDH applications.

Keywords

• dilation
• electrical drilling
• process parameters
• removal rate
• tapper