Artificial neural network model for material removal rate of TI-15-3 in electrical discharge machining

Artificial Neural Network (ANN) plays an important role in predicting the linear as well as non-linear problems in distinctive fields of engineering. A lot of attempts have been carried out to develop models from experimental data using Artificial Neural Network. This paper is to develop a model for correlating the variables and electrical discharge machining (EDM) characteristics employing neural network technique. Machining is performed on new high-strength titanium alloy Ti-15V-3Cr-3A1-3Sn (Ti-15-3) while copper is used as :EDM tool. Training and testing are done with data that are found succeeding the experiment as design of experiments. To find out the suitable architecture of the network for this problem different architectures are studied altering the number of hidden layer and neuron. The model with 4-10-41 architecture is found the most suitable. The model is also verified following confirmation test with some other data. The confirmation test yields that the error is within the agreeable limits. Accordingly, ANN is demonstrated to be a practical and effective way for the evaluation of EDM material removal rate (MRR).