A Multi-Objective Memetic Algorithm for RBDO and Robust Design
Abstract: 2010 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must first be obtained from the IEEE. Product design optimization under model or input variable uncertainty is commonly required, in which robustness and reliability are two important attributes of the design. In structure design, it is critical to maintain the design feasibility (or reliability); while at the same time, to counter manufacturing variations, robust design is employed in order to obtain high product quality. It is necessary, therefore, to establish a multi-objective optimization problem that combines both robustness and reliability considerations, where the product’s performance variation and the performance function are simultaneously optimized, subject to probabilistic constraints for design feasibility. Based on the idea of the reliability index and the Most Probable Point (MPP), probabilistic constraints are converted into deterministic constraints. Then an efficient Multi-Objective Memetic Algorithm (MOMA) is presented to minimize robustness and performance value, subject to deterministic constraints. A classical I-beam example with probabilistic constraints illustrating the MOMA concept and applicability is proposed in the paper, the result is a Pareto frontier of robustness and performance value. Product design optimization under uncertainty has been widely discussed in recent years. To design and manufacture high quality products, techniques are employed to minimize variations that exist in materials properties and loading in manufacture. Robust design, proposed by Taguchi, is a method which focuses on minimizing performance variation without eliminating the sources of variation. In product structure design, though robustness objective can be achieved given the product performance, it is necessary to guarantee the robustness design to maintain design feasibility at an expected probabilistic level (or called reliability). Then the paradigm of Reliability-based design optimization (RBDO) is proposed for design under uncertainty. Therefore, it is necessary to establish a multi-objective optimization problem to integrate robustness and reliability, where the performance variation and performance measure are simultaneously minimized, subject to probabilistic constraints.
Keywords: RAMS 2010 Proceedings - Product Reliability - Reliability Model