Structural reliability-based design optimization with non-probabilistic credibility level

Abstract

With the increasing diversity of performance requirements for engineering structures, credibility has become an essential prerequisite for ensuring structural safety. This article introduces a reliability-based structural optimization with non-probabilistic credibility (NCRSO) method, to characterize the credibility of the optimization process. First the concept of uncertainty quantification for credible set is introduced. By utilizing a limited number of sample points, structural parameters with a high level of credibility are obtained and maintained throughout the optimization procedure. Additionally, a reliability index is established based on area ratio and credible feature distance, and a credible sequential optimization strategy is developed. Implementation of the proposed method significantly enhances the overall computational efficiency of structural optimization. The reliability index can directly utilized as the translation distance for deterministic constraints in the sequential strategy, minimizing unnecessary calculations and leading to improved convergence. Three engineering examples are eventually presented to effectively illustrate and emphasize the necessity of considering credibility when utilizing the structural optimization method under complex uncertainty from different perspectives.