Collaboration Pursuing Method for MDO Problems

Abstract

Multidisciplinary Design Optimization (MDO) problems are dominated by couplings among subsystems formulated from different disciplines. Effective and efficient collaboration between subsystems is always desirable when solving MDO problems. This work proposes a new sampling-based methodology, named the Collaboration Pursuing Method (CPM), for MDO problems. In the CPM, a new collaboration model, reflecting both physical and mathematical characteristics of couplings in MDO problems, is formulated to guide the search of feasible design solutions. The interdisciplinary consistency among coupled state parameters in MDO problems is reflected and maintained by the collaboration model. An adaptive sampling strategy is also developed to speed up the search of local optimal solutions. The new method is implemented using MATLAB ® 6.0 and successfully applied to four test problems including an engineering design application.