Improved Particle Swarm Optimization Algorithm

Particle Swarm Optimization is a kind of swarm intelligence (SI) technology, whose Prinple is simple,it is easy yo implement. PSO has strong commonality and global optimization characteristic.It is a effective method of solving nonlinear problem. However, some studies reported that the PSO algorithm might sometimes get stuck in a near optimal solution, which made us difficult to improve solution accuracy by fine tuning. By analyzing the influence of average absolute value of all particles' velocity on search ability of the PSO, this paper proposed an improved particle swarm optimization with nonlinear dynamic adaptive velocity variation. In the improved algorithm, the average absolute value of velocity was used as an index to represent the briskness of all the particles. The average absolute value of velocity changes along with a given nonlinear variation of ideal velocity by feedback control for tuning inertia weight to improve the search ability in the multidimensional space. Experimental results showed that the proposed algorithm remarkably improved the ability of PSO to jump out of the local optima and significantly enhanced the convergence precision.