We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
A novel polynomial chaos expansion-based method for feedback-coupled multidisciplinary design optimization under metamodel uncertainty.
- Authors
Liu, Zhao; Song, Zhouzhou; Zhu, Ping
- Abstract
Uncertainty-based multidisciplinary design optimization (UMDO) is a practical methodology to cope with uncertainties in the design of sophisticated systems. To alleviate the computing burden, the time-consuming computer simulation models are often replaced by metamodels. Nonetheless, inconsistency between metamodel and simulation model, or metamodel uncertainty, could be introduced into the multidisciplinary design optimization process due to lack of data. The optimal solution may deviate from the true result or even become infeasible if the metamodel uncertainty is neglected. In this research, a new UMDO approach based on polynomial chaos expansion (PCE) for feedback-coupled systems is proposed aiming at improving the accuracy and efficiency of UMDO process under metamodel uncertainty. In this approach, PCE is utilized for the Kriging metamodel uncertainty propagation. The decoupled formulation is used to solve the UMDO. PCE is integrated into the decoupled UMDO framework naturally since the PCE coefficients could be regarded as design variables, which can assure the satisfaction of disciplinary consistency by matching the distribution mean and variance functions of coupling variables under stochastic uncertainty. Then, a modified particle swarm optimization algorithm is proposed to implement the UMDO efficiently. The effectiveness of the proposed approach is verified by a mathematical problem and a fire-detection satellite design problem.
- Publication
Structural & Multidisciplinary Optimization, 2022, Vol 65, Issue 4, p1
- ISSN
1615-147X
- Publication type
Article
- DOI
10.1007/s00158-022-03207-y