On the Coordination of Multidisciplinary Design Optimization Using Expert Systems
Andrew Price, Andy J. Keane, Carren M. E. Holden · AIAA Journal · 2011
Multidisciplinary Design Optimization (MDO) of complex systems in the enterprise is typically broken down along domain specialist lines with associated expertise, tools, method and process. This paper investigates an approach to MDO that exploits heuristic control of the bounds of the common design variables across the domains of a decomposed problem. By bringing the common design variable vector to a single point in the design space, in the context of internally consistent and multiple discipline feasible state variables (and consistently resolved local design vectors), the MDO problem is solved. We present a system level management system in which an expert system is used to coordinate the activities of the domain level optimizers. Motivated by move limit and trust region algorithms a rule base has been developed to manage the bounds on the common design variable vector, control the exchange and relaxation of state coupling variables and control the specification of the domain level optimization problems. Through application of the rule base across a couple of representative MDO problems assembled from the literature the viability and performance of the method are discussed.