Comparisons of neural network models on material removal rate in electrical discharge machining

Predictions for work removal based on physical models have been reported in electrical discharge machining (EDM) in recent years. However, when the change of polarity has been considered, few models have succeeded in giving consistent predictions. In this study, comparison of modeling the material removal rate of the work for various materials considering the change of polarity among six different neural networks together with a neuro-fuzzy network have been illustrated. The six neural networks are namely, the logistic sigmoid multi-layered perceptron (LOGMLP), the hyperbolic tangent sigmoid multi-layered perceptron (TANMLP), the fast error back-propagation hyperbolic tangent multi-layered perceptron (error TANMLP), the radial basis function networks (RBFNs), the adaptive TANMLP, and the adaptive RBFN. Also, the neuro-fuzzy network is the adaptive-network-based fuzzy inference system (ANFIS). Trained by the same experimental data selected with the method of design of experiment (DOE), the parameters of the above models have been optimized for further analysis. Based on the conclusions from the comparisons at checking the error among the network models, the best is the ANFIS with Bell-shape membership functions. Also, it can be concluded that the further experimental results have shown the accurate predictions based on the ANFIS model.