Design of Particle Swarm Optimization with Random Flying Time

In order to improve the searching performance of Particle Swarm Optimization (PSO) algorithm, a new PSO algorithm with random flying time was designed in this paper. In traditional PSO, the flying time of particles remains the same and is fixed to one when the positions are updated which lead to the "oscillation" phenomena. Randomness was introduced to the location updating formula of the particle, and the new definition of location update formula was given. We have, in the form of theorem, also proved that before PSO finds the optimal value, the position of the particle keeps changing continuously, the flying distance of the particle in the new algorithm is smaller than that in the original algorithm, which can avoid the particle "fly over" the optimal location and reduce the possibility of "oscillation" phenomena. Experimental results show that the algorithm is of advantage of improving the convergence speed and calculation accuracy. The design of PSO with random flying time both increased the randomness of particles and improved the searching performance.