Investigation and Analysis of MRR in Spark Erosion Machining Through Artificial Neural Network

Abstract

EDM is an advanced machining process for machining, hard material parts which are difficult to machined by conventional machining process. There are various types of products which can be produced by using Die-sinking EDM, such as dies, mould, parts of aerospace, automobile industry and surgical components can be finished machined by EDM. In present scenario numbers of researchers have explored a number of ways to improve EDM efficiency. The optimum selection of manufacturing condition is very important in manufacturing processes as they determine surface quality, dimensional accuracy of the obtained parts. EDM process is based on thermoelectric energy between the work piece and an electrode. A pulse discharge occurs in a small gap between the work piece and the electrode and removes the unwanted material from the parent metal through melting and vaporizing. ANN can be trained with GA and BP algorithms, so that the local least solution can be avoided and the training speed enhanced. The experiment has proved that the utilization of mirror processing conditions generated from the above method will consequently lead to both good small-area mirror processing results and desired processing precision and efficiency. Experimental data was gathered from Die sinking EDM process for copper-electrode and steel-workpiece (D2 tool steel). It is aimed to develop a behavioral model using input-output pattern of raw data from EDM process experiment. The behavioral model is used to predict MRR and then the predicted MRR is compared to actual MRR value. The results show good agreement of predicting MRR and TWR between them. A feed forward neural network based on back-propagation is a multilayered architecture made up of one or more hidden layers (layer 1 - 6 neurons & layer 2