چكيده به لاتين
Abstract
Optimization using meta-heuristic algorithms is usually inefficient when the number of design variables or size of the problem is excessive, since searching in a large design space is highly time consuming for optimization algorithms. In this thesis, a hybrid algorithm that is called MDVC-UVPS is introduced to perform design optimization of large-scale skeletal structures. This method is based on vibrating particles system (VPS) algorithm, upper bound strategy (UBS), and multi design variable configurations (MDVC) cascade optimization. The VPS algorithm is a population-based meta-heuristic algorithm which is originated from damped free vibration of a single degree of freedom system and it is proven to be a suitable method for optimal design of structures. By employing the UBS concepts in the hybrid method, some of the unnecessary structural analyses are avoided in the course of optimization. Therefore, the computational time is reduced. Additionally, the problem is solved in consecutive stages, where different design variable configurations are utilized at each cascade optimization stage to reduce the complexity of the problem. Optimal design of dome shaped trusses with frequency constraints, optimal design of antennas with strength and displacement constraints and optimum seismic design of steel space frames are considered as benchmark examples to evaluate the performance of the proposed method. Significant reduction in computational time and better performance of the MDVC-UVPS method for all the investigated design examples show the effectiveness of the proposed method for structural optimization problems.
Keywords: Optimum seismic design, skeletal structures, meta-heuristic algorithms, vibrating particles system, cascade optimization.
