Y.L. Li, Y.P. Zhang, X. Bai, L. Zhou

Pages: 27-34

Abstract
In particle swarm optimization (PSO) problems, inertial weight and learning factor are important parameters to determine the searching ability of the algorithm. This paper aims to overcome the defects (e.g. premature convergence and proneness to local optimum) of PSO in shortest path planning. For this purpose, nonlinear dynamic inertia weight (NDIW) and variable learning factor (VLF) algorithm were introduced to optimize the PSO for shortest path planning. The proposed algorithm was applied to solve the shortest path and compared with several other PSO algorithms. The simulation results show that the algorithm has strong searching ability and fast convergence, and outperforms the contrastive algorithms in road transport path planning.
Keywords: Particle Swarm Optimization (PSO); evolutionary computation; Adaptive Weight (AW); Dynamic Learning Factor (DLF); shortest path