In this paper, the multidisciplinary design optimization (MDO) problems are considered as a type of complex adaptive system. A novel MDO method, called adaptive subspace optimization, was proposed based on the paradigm of complex adaptive system. Each subspace (discipline or subsystem) design optimization can be considered as an agent, which senses the environment and finds an optimum solution using a specific optimizer. The environment for each subspace is a set of values, which consists of the values of the current design variables and their associated state (behavior) variables of all other subspace. System behavior evolves over time. As a consequence, convergence often results from the interactions among the subspaces. Two design optimization examples have been used to test and demonstrate the proposed MDO method. Initial results with the method for two problems are quite encouraging. Satisfactory convergence and good agreements with the benchmark results have been obtained.
[1]
S. Forrest,et al.
Modeling Complex Adaptive Systems with Echo
,
1994
.
[2]
Jaroslaw Sobieszczanski-Sobieski,et al.
Multidisciplinary aerospace design optimization - Survey of recent developments
,
1996
.
[3]
Marc A. Stelmack,et al.
Framework for Multidisciplinary Design Based on Response-Surface Approximations
,
1999
.
[4]
J. Renaud,et al.
Approximation in non-hierarchic system optimization
,
1992
.
[5]
John E. Renaud,et al.
Response surface based, concurrent subspace optimization for multidisciplinary system design
,
1996
.
[6]
Ilan Kroo,et al.
Aircraft design using collaborative optimization
,
1996
.
[7]
Jaroslaw Sobieszczanski-Sobieski,et al.
Bilevel Integrated System Synthesis with Response Surfaces
,
2000
.