Multi-objective optimization of partition temperature of steel sheet by NSGA-II using response surface methodology

Abstract

For the manufacturing of automotive B-pillar parts with tailored properties, a multi-objective optimization of partition temperature of steel sheet is conducted to obtain the magnetic conductor parameters corresponding to optimal mechanical properties by using Response Surface Methodology (RSM) and Non-dominated Sorting Genetic Algorithm (NSGA-II). Based on the proposed zone induction heating-stamping, the numerical simulations are implemented to optimize the design variables (width, thickness, spacing of the magnetic conductor) on the objective functions (the maximum Z1 and minimum temperature Z2). The regression model of quadratic polynomial is established, the significance of each item is analyzed by analysis of variance (ANOVA). The Pareto-optimal fronts are obtained by NSGA-II algorithm. The results reveal that the most significant factor for Z1 and Z2 is the linear term of width and thickness respectively. On the basis of Pareto optimization solution set, the final optimal solution is selected through decision-making. In other words, the objective functions Z1 = 950 °C and Z2 = 650 °C are obtained, the corresponding magnetic conductor parameters as design variables are Width = 9.05 mm, Thickness = 2.96 mm and Spacing = 2.19 mm. Through the simulation and experimental verification of rectangular steel sheet, the multi-objective optimization of partition temperature of steel sheet by NSGA-II using RSM is effective. The regression model provides theoretical and experimental basis for the customization of steel sheet temperature field under the zone induction heating-stamping.