混合不确定性下的多学科协同可靠性分析方法

Reliability-based Multidisciplinary Design Optimization(RBMDO)has been considered as an effective method to solve the problem of complex product design optimization under the uncertainties. However, due to its excessive computational cost, there have some difficulties in the application of RBMDO. As an important part of RBMDO, multidisciplinary reliability analysis plays a direct leading role in the computational efficiency of RBMDO. The current multidisciplinary reliability analysis under mixed uncertainties has still been a nested or sequential execution mode, which is easy to cause the problems of poor discipline autonomy and expensively computational cost in the reliability analysis of complex products. Aiming at the multidisciplinary reliability analysis problem under mixed uncertainties, a multidisciplinary collaborative reliability analysis method was proposed to solve the computational problems. The method transformed the multidisciplinary limit state function into the standard normal space, and then the KKT conditions of interval reliability analysis were used to conduct model transformation for constructing the integral model of reliability analysis. The integral model was decomposed into several parallel single-subject optimization problems, and the condition of consistency constraints was used to coordinate all disciplines, which could realize the parallel solution of multidisciplinary reliability analysis. A numerical example and a composite cylinder design example were used to verify the feasibility and effectiveness of the proposed method.