Non-sorting multi-objective optimization of flexible roll forming using artificial neural networks

The main defects due to flexible roll forming (FRF) processes include longitudinal bow and wrinkling. In this study, experimental and numerical analyses were performed using three different blank shapes to characterize the effects of the process parameters on defects in parts fabricated by FRF with and without leveling roll. Owing to the complexity of the FRF process, two algorithms were combined for its optimization. Artificial neural network-based Non-dominated Sorting Genetic Algorithm II (NSGA-II) was used to optimize the effective parameters of the FRF process, such as the sheet thickness, yield strength, and blank shape, with respect to the target bend angle to minimize the longitudinal bow and wrinkling of the product. The back-propagation neural network (BPNN) was used to identify two objective functions, while non-sorting multi-objective algorithm simulation was used to optimize the input parameters to minimize the objective functions. The results showed that the sheet thickness had the greatest effect on the minimization of the two objective functions, followed by the yield strength and blank shape, respectively.