In this paper, a multi-level optimization algorithm is presented. The algorithm constructs a system level objective function that drives all disciplines toward each discipline's optimal point while satisfying all necessary discipline constraints and the system level constraints. Mathematical convergence of the system level optimization is demonstrated. Communication among the system and the discipline level optimizations is facilitated by the new algorithm. An analytical example, which has a non-convex Pareto front with coupled disciplines, is solved for highlighting the main capabilities of the new algorithm, and a ship design optimization analysis is conducted. The latter demonstrates how the new algorithm can be used for analyzing a complex engineering system.
展开▼