Particle Swarm Optimization with non-linear velocity

Particle Swarm Optimization (PSO), a population based optimization technique, has two intrinsic problems of slow convergence and tendency to converge prematurely. In order to overcome these problems, we propose an improvement to the velocity update equation of the standard PSO algorithm in which particles of a swarm tend to move towards the global best position more rapidly as compared to the local best position. Two different non-linear weight factors are multiplied with the two parts of the velocity update equation; one that tends to move the particle to the global best position, while the other tends to move the particle back to its local best position achieved so far. By introducing the separate weight factors, a significant improvement in the results is seen. We test the proposed algorithm on six benchmark functions and the simulation results are presented. The results indicate that the proposed algorithm does not converge prematurely and its convergence speed is faster than the standard PSO algorithm.