Abstract

A common robust topology optimization is formulated as a weighted sum of expected and variance of the objective functions for the given uncertainties. This has recently been applied to topology optimization with uncertainties in loading, [1]. Figure 1(a) shows the Pareto front of solutions found using uniformly distributed weightings. This front suffers from crowding for weight values < 0.5 and is sparsely populated for weights > 0.625. In the general case, the two goals of multi-objective optimization are; to find the most diverse set of Pareto optimal solutions, and, to discover solutions as close as possible to the true Pareto front. This paper presents schemes to achieve both these goals.