Research On The Evolutionary Structural Optimization Against Buckling

Structural stability is an important aspect in structural design, especially for large span structures. So the problems that exist in stability optimization are studied in this paper, and the optimization methods basing on the evolutionary optimization are proposed.Evolutionary structural optimization (ESO) method can be used in sizing optimization via proper improvement. In sizing optimization, buckling sensitivity number is calculated in the same way as classic ESO, but new iteration criterion are proposed here to overcome the deficiencies exist in the classical method, and the method efficiency is proved through the example.For bimodal and multimodal problems, because the eigenvector of an eigenvalue is no longer unique, so the sensitivity number calculated before is not correct anymore, thus two different methods are selected to solve these problems. Furthermore, comparisons of these two methods are given, and the results show that the latter one is more powerful in computation efficiency.Considering the constraints such as stress, displacement and stiffness, an improved ESO method is proposed. By limiting the elements set in which the element's section area can be adjusted, the structural is always optimized in the feasible domain. A ten-bar model is optimized by this method, furthermore, comprehensive sensitivity method is also used in this model, and the results show that the proposed method is more suitable for the structures with only stress constraints.Because ESO method is basing on the gradient information, so the results may be a local optimum solution. Genetic algorithm is more powerful then gradient information based method when searching for a global optimum solution. Hence to ensure the reliability of the result, both genetic algorithm and ESO method are used for more accuracy results. Through this combination, the optimization process is more reliable and efficient comparedto only use one method. Penalty function is introduced when using GA to solve the problems with constraints. Finally a lattice dome is optimized to illustrate the optimization progress and the optimal result shows the feasibility of this method.