Numerical Control and Optimization of Springback in L-bending of Magnesium Alloy Through FE Analysis and Artificial Intelligence

Abstract

Springback is an undesirable phenomenon that extensively occurs during sheet metal forming processes. There are many parameters which have great influence on springback. Hence, selection of appropriate controllable parameters may lead to spingback reduction. In the present work an attempt has been made to find optimal combination of L-bending parameters (i.e. die temperature, step distance, lower punch radius, die clearance and step height) regarding minimum springback. Here, combination of finite element model (which was validated through trial experiments) and Taguchi experimental design were used to form design matrix. Then adaptive neuro-fuzzy inference system (ANFIS) was then applied to correlate intelligent relationships between process inputs and springback. The accuracy of developed ANFIS model was compared with FE model and experimental testing data. Finally, the teaching learning based optimization algorithm was combined with the developed ANFIS model to minimize the springback. The obtained optimal results were then compared with those derived from FE model and experiments and showed that the proposed approach can predict the optimal drawing process accurately. Furthermore, the optimum results were discussed carefully according to mechanical behavior of L-bending process and through implicit finite element model.