A New Approach to Multidisciplinary Design Optimization via Internal Decomposition

摘要: Traditional approaches to MDO problem decomposition have shown poor performance when solving problems with strong interactions between disciplines. We present a new strategy for that aims overcome this diculty. The method, called Block Approximation Krylov Renement, or BAKR, decomposes the solution of linear system at each iteration an interior point algorithm. By decomposing inside optimization algorithm, rather than original design problem, we maintain global and local convergence properties algorithm while reducing overall computational cost solution. Preliminary test results show reductions in both number function evaluations, demonstrating potential future application large problems.