论文部分内容阅读
A new hybrid reliability analysis method is developed based on probability and ellipsoidal convex models.Random distributions are used to deal with the uncertain parameters with sufficient information,while the ellipsoidal convex model is employed to deal with the uncertain-but-bounded variables which are correlated with each other.Apparently,a complex nesting optimization problem will be involved in this hybrid reliability analysis and the traditional optimization method will lead to extremely low efficiency.To alleviate the computational burden,an efficient decoupling strategy is developed to solve the nesting optimization problem.By using Karush-Kuhn-Tucker necessary condition,the complex nesting optimization problem can be transformed to an equivalent single- layer optimization model with only random variables.Then,the sequential quadratic programming (SQP) method is adopted to directly solve this equivalent model.The present method is applied to the two numerical examples and the results are compared with the traditional nesting optimization method.The results indicate that the present method has a fine accuracy,and also an acceptable efficiency if not many uncertain parameters are involved.