Improved Particle Swarm Optimizer Based on Nonlinear Inertia Weight Dynamic Changing
Feng Qian · 2008
This paper proposes a particle swarm optimizer (PSO) with nonlinear inertia weight dynamic changing (NDPSO) to improve the speed of convergence and fine tune the search in the multidimensional space. The decreasing exponent of inertia weight is selected by the region which particle stays in the swarm and each particle will be having different inertia weight according to the place which it stays. If the particle is far from the optimal point of the swarm the exponent of inertia weight will be smaller in order to fly to the optimal position quickly, on the other hand if particle flies near to the optimal position the exponent of inertia weight will be larger for particle searching for the optimal value carefully. The propose NDPSO is tested with a set of 4 benchmark problems and compares with linearly decreasing weight PSO (SPSO), nonlinearly decreasing weight PSO (CS-PSO). Experiment results indicate that the proposed PSO improves the search performance on the benchmark functions significantly.