Particle Swarm Group Search Optimizer And Its Application To The Truss Structural Design

Modern intelligent optimization algorithm is mainly studied in this thesis, especially these two swarm intelligent optimization algorithms, namely particle swarm optimization (PSO) and group search optimization (GSO). Based on these two intelligent optimization algorithms and combined with the "harmony search" algorithm, a new hybrid intelligent optimization algorithms is proposed in this thesis, namely particle swarm group search optimization (PSGSO).The concept of structure optimization problem and the development history of optimal design algorithm are introduced in this thesis.The basic idea, the basic principle and realization process of PSO algorithm and GSO algorithm are introduced. The main content of this thesis is introduced in details, such as the generation, principle and the realization of PSGSO algorithm as well as the combination of "harmony search" algorithm. There are three types of models of algorithms, the first one is applicable to continuous variable problems, the second one is applied to discrete variable problems and the third one which contains continuous variables and discrete variables in design variables is applicable to the mixed variable conditions.Based on MATLAB platform, the main program of PSGSO algorithm is compiled. Combined with the finite element method, the algorithm is applied to the examples of classical truss structure cross-section optimization designs. Optimization calculation contains two design types including continuous and discrete variables.The time efficiency of the PSGSO algorithm is also studied in this thesis. The results of several examples showed that the proposed algorithm has amazing adaptability to less particle size cases, which greatly improved the calculation efficiency. Therefore, it is a kind of efficient swarm intelligent optimization algorithm.The PSGSO algorithm is applied to shape optimization design with mixed variables, which includes discrete variables and continuous variables. The result shows that the algorithm similarly has good convergence rate and accuracy to the shape optimization problems.