Research On Topology Optimization Methods Based On Improved Binary Particle Swarm Optimization

Topology optimization aims at finding the optimum distribution of materials within thedesign domain under given loads, boundaries and constraints to improve the materialutilization and the performance of the structures for lightweight design. This paper deeplyanalyzed the research status both in home and abroad. To deal with the shortcomings ofthe existing research, such as low solution quality, small-scale problems and localoptimum, this paper studied different aspects of topology optimization problems forcontinuum structures, including model building methods, solving algorithms andengineering applications.Firstly, considering the local optimums solution of the conventional bi-directionalevolutionary structural optimization (BESO) methods, the solution quality of BESO canstill be improved. Besides, the conventional binary particle swarm optimization (BPSO)methods for topology optimization are not efficient enough, because they can only dealwith small-scale optimization problems. This paper presents the first study to combineBPSO with BESO for topology optimization. A dual coding method is employed torepresent material distribution and individuals. Three improved BPSO operators aredeveloped for the improved BPSO. The effects of different sets of parameters of theproposed method are discussed. A penalization mechanism is introduced to deal with theconnectivity problems of the structures. Several2D and3D examples of minimum meancompliance problems are performed to demonstrate the efficiency of this method.Secondly, the multi-loading cases of topology optimization have been studied. In thesecases, the objective function is treated as the linear weighted sum of the sub-cases. Themodel of these cases is built based on the IBPSO-BESO method. The results of thesimulation runs of several2D and3D examples indicate the efficiency of the proposedmethod.Finally, the model repetition constraint for topology optimization has been studied.This paper utilizes the unit cells method to build the model of model repetition constraintbased on the IBPSO-BESO method. And the proposed method has been improved forsolving these problems. From several examples one can see the effectiveness andversatility of the proposed method.