A Topographical Approach For MultimodalStructural Optimal Designs

This paper presents a two-stage multimodal optimization algorithm and its application to structural optimizations. The first stage of the two-stage algorithm serves to identify local optimum candidates which are defined as the best sampled design in each valley containing a local optimum from a number of sampled designs by using topographical identification techniques. The second stage of the optimization algorithm involves a number of local searches using the local optimum candidates as initial designs. A unique crystal growth approach is proposed to perform systematic identification of the local optimum candidates. This topographical multimodal optimization algorithm is applied to one illustrative structural optimization problem and comparison are made to results obtained by an existing topographical algorithm.