Optimization of Trusses Using Simulated Annealing for Discrete Variables

摘要

The simulated annealing (SA) algorithm is applied to the discrete structural optimization of trusses. Some approaches have been proposed to improved the basic procedures of the SA algorithm, such as the determination of initial temprature, the generation and acceptance of solutions, a recorder added, and the converage criterrion based on a newly defined relative precision. By using a multi-objective layout, the discrete optimization is combined with the continuous optizimation. The important algorithm control parameters, such as the number of random seeds and Markovian chain, are studied carefully by numerous tests. These improvements and studies have enhnced the robustness, efficiency and accuracy of the SA algorithm in the structural design optimazation problems. A certain number of traditional and a relatively complicated 200-bar problems are solved with the improved SA algorithm and compared with related examples in literatures. The numerical results have demonstrated that the improved SA algorithm of this paper has high-solution precision, and its solution efficiency has noticeably increased. It is hoped the SA algorithm can be applied in structural design optimization to make use of its advantages.