ANFIS-based forming limit prediction of stainless steel 316 sheet metals

Abstract

Effect of microstructure on the formability of the stainless sheet metals is a major concern for engineers in sheet industries. In the case of austenitic steels, existence of strain-induced martensite ($${\alpha }^{^{\prime}}$$ α ′ -martensite) in their micro structure causes considerable hardening and formability reduction. In the present study, we aim to evaluate the formability of AISI 316 steels with different intensities of martensite via experimental and artificial intelligence methods. In the first step, AISI 316 grade steels with 2 mm initial thicknesses are annealed and cold rolled to various thicknesses. Subsequently, the relative area of strain-induced martensite are measured using metallography tests. Formability of the rolled sheets are determined using hemisphere punch test to obtain forming limit diagrams (FLDs). The data obtained from experiments were further utilized to train and validate an artificial neural fuzzy interfere system (ANFIS). After training the ANFIS, predicted major strains by the neural network are compared to a new set experimental results. The results indicate that cold rolling has unfavorable effects on the formability of this type of stainless steels while significantly strengthens the sheets. Moreover, the ANFIS exhibits satisfactory results in comparison to the experimental measurements.