Composite First-Order Reliability Method for Efficient Reliability-Based Optimization of Electromagnetic Design Problems

This paper proposes a composite first-order reliability analysis method to effectively perform the reliability-based optimization of electromagnetic (EM) design problems. The proposed method utilizes both of two different ways, reliability index approach (RIA) and performance measure approach (PMA), for reliability analysis of probabilistic constraints. However, each performs a distinct function: the first merely checks the feasibility status of probabilistic constrains, and the second evaluates the failure probability of only active constraints selected from the feasibility identification. Such a unique scheme can substantially enhance computational efficiency during optimization process. The proposed method is applied to two EM design problems, and its numerical efficiency and accuracy are examined by comparison with existing methods.