Optimization and Prediction of Responses Using Artificial Neural Network and Adaptive Neuro-Fuzzy Interference System during Taper Profiling on Pyromet-680 Using Wire Electric Discharge Machining

In the present study, taper cutting is performed with the aid of a uniquely designed fixture. This is attempted to avoid the difficulties in tapering using wire electric discharge machining like wire break, dimensional error, guide wear, non-uniform flushing and low surface quality. An investigation of output parameters was made during taper machining using a fixture. The cutting rate (CR) and surface roughness (SR) were considered for response surface optimization (RSM) as they were important response parameters that indicate the quality of a machined component. It is observed that servo gap voltage and pulse act contrastingly on the output parameters. For achieving a trade-off of input parameters with output responses, RSM optimization is selected during taper profiling. There were 3-5% variations for both CR and SR when compared to experimental and RSM optimal values. The tapered profile slots of different angles like 0°, 15° and 30° were machined on Pyromet-680 using optimal machining parameters. The effect of different profiling parameters like wire distance between guides (WD), dwell time (DT), profile offset (PO) and cutting speed override (CO) on output responses like CR and SR was analyzed. Adaptive neuro-fuzzy interference system (ANFIS) and artificial neural network (ANN) models have been established for the prediction of the output responses. The validation is performed by experimentation, and the prediction errors ranged from 0 to 5% for both the responses CR and SR in ANFIS models. So ANFIS models proved to be the most efficient as there is an improvement of 45-50% in prediction compared to ANN models.