A data-driven approach for robust wing jig shape design

This paper presents a data-driven robust jig shape design process, providing an initial application case to understand the opportunities of accounting for the implementation of uncertainty optimization in a real industrial use case. The study utilizes an industrially relevant dataset defined for academic usage, which includes both variations of aerodynamic performance (lift/drag improvement) and structural loads (shear, moment, and torque) with the variations of wing twist and structural stiffness parameters. The objective is optimizing the wing jig shape in the presence of uncertainties in structural stiffness, aiming to balance the confidence of achieving aerodynamic performance targets and the risks of exceeding the design loads margins. First, Gaussian process emulators are constructed based on given dataset to represent the relationship between aerodynamic performance as well as loads and input uncertain parameters. Then, aero performance and loads analysis are conducted separately to evaluate the impact of target improvements and design load margins on uncertainty optimization solutions. Finally, Non-dominated Sorting Genetic Algorithm II(NSGA-II) is employed to determine the optimal jig twist that satisfies various optimization objectives, effectively managing the trade-offs between aero performance and structural safety under stiffness uncertainties.