ANFIS based predictive model for wire edm responses involving material removal rate and surface roughness of Nitinol alloy

Abstract

Nickel titanium (Nitinol) alloys are functional materials which find extensive applications in aerospace, automotive and medical applications. Nitinol alloy is very hard to machine using conventional process due to its ductility, temperature sensitivity, severe work hardening and minimum thermal conductivity, resulting in massive tool wear and very poor surface finish. Wire electro discharge machining (WEDM) is the most versatile among all unconventional machining processes for machining difficult to cut materials. This work reports the relationship between WEDM machining parameters: pulse on time (Ton), pulse off time (Toff), peak current (Ipeak) gap voltage (V), and output parameters: metal removal rate (MRR) and surface roughness (SR) analyzed using artificial neuro fuzzy inference system (ANFIS) predicted model.Results attained from ANFIS model were distinguished from experimental values and it was found that anticipated values of the output attributes are in good agreement with actual values. Correlation Coefficient (R) of both MRR and surface roughness was almost 0.9945 which is nearly equal to one, and the coefficient of determination (R2) and avg error(%) erected for surface roughness were 0.9891 and 2.04 respectively whereas for material removal rate, coefficient of determination (R2) and avg error (%) were found to be 0.9738 and 1.70% respectively. Experimental data was employed for prediction modeling of MRR and SR with an accuracy of 98.3% and 98.75% respectively.