Approach of non-probabilistic reliability topology optimization using evidence theory

Abstract

It is of great importance to consider uncertainty due to insufficient data or imprecise information in topology optimization design. In this paper, evidence theory is presented to handle the imprecise data situation. The plausibility measure based on evidence theory is introduced to overcome the difficulty of constructing the precise probabilistic constraint. Furthermore, the topology problem in this evidence-based optimization design is solved by combined strategy of the differential evolution and superior topology technique. In order to overcome the difficulty of intensive computational cost in calculating plausibility measure, an improved method of evolutionary optimization design integrating with imprecise extremum idea is proposed. Two truss examples are given to demonstrate the proposed approach. The research indicates that deterministic results may be the failure solutions under epistemic uncertainty. Although evidence-based optimum topology designs are more conservative than deterministic results in aspects of weight and optimal structural topology layout, it gains a more robust design under epistemic uncertainty.