Application of GRNN and multivariate hybrid approach to predict and optimize WEDM responses for Ni-Ti shape memory alloy

The current investigation aims at highlighting the application potential of a smart approximation tool, which is general regression neural network (GRNN) during machining of nickel-titanium (Ni-Ti) shape memory alloy using WEDM. A series of experiments were accomplished based on Taguchi's orthogonal layout design. Pulse on time, discharge current, wire tension, wire speed and flushing pressure were considered as five distinct WEDM parameters, whereas arithmetic mean roughness, maximum peak to valley height, root mean square roughness, and micro-hardness were selected as the major responses to be investigated. The aforementioned WEDM responses were predicted with the help of the projected GRNN model and compared with the experimental results. The investigation was further extended to ascertain the optimum combination of input parameters using a hybrid approach. This was done by combining VIKOR method with the Fuzzy logic system. The prediction error of the GRNN model was noted as ±5% within the studied range of machining parameters. Finally, the adequacy of the multivariate VIKOR-Fuzzy approach was verified by performing confirmation test which exhibited improvement in WEDM responses.