Abstract:
Aiming at the shortcoming of slower searching speed in the traditional particle swarm optimization (PSO), this paper introduces an average speed to describe the activeness of particle swarm, which is utilized to regulate the inertia weight and learning factors such that the searching speed is accelerated. Besides, this paper incorporates simulated annealing (SA) into PSO. By means of PSO’s fast parallel and SA’s probability sudden jump, the population diversity may be maintained and the local convergence is effectively avoided. Moreover, the improved algorithm is verified through 4 typical test functions. Finally, this proposed method is applied to the optimization on the model for wind speed probability distribution of wind farm. Compared with traditional statistic strategy, the present algorithm may attain higher precision.