Optimum Sizing of Truss Structures Using a Hybrid Flower Pollinations

Abstract

The truss sizing optimization problem, where the cross-sectional areas of truss members are optimized, is a nonlinear and complex problem. The problem becomes quite challenging especially when a large number of variables exist. Metaheuristics proved to be working efficiently for this problem type. Flower pollination algorithm (FPA) is a recently developed metaheuristic search algorithm, and it has been successfully applied to structural optimization problems in the literature. In this paper, the formulation of FPA is enhanced to improve its local and global searching capabilities for better performance. For this purpose, a hybrid algorithm is developed by imitating the algorithmic structure of FPA while the mutation and crossover operators of differential evolution (DE) with elitism strategy are replaced with Levy flights of FPA to explore search space more efficiently and exploitation ability of FPA is improved with the addition of mutation factor and crossover operator of DE working in the local search formula of FPA. The performance of the proposed algorithm is tested with benchmark truss sizing optimization problems and the results are compared with other metaheuristics. The results show that hybrid FPA improves the capabilities of the original one, produces more robust solutions than both FPA and DE, generates quite competitive results for the test problems compared to the previous studies with different algorithms. Therefore, the hybrid FPA proves to be a promising optimization algorithm alternative for engineering practice.