Improved particle swarm optimization algorithm with density factor
一种引入密度因子的改进粒子群优化算法*

Based on the conventional linear decreased weight particle swarm optimization algorithm, proposed a novel improved PSO with a so-called density factors involved. Defined the density of one generation in the form of radial basis function with the average fitness value and the best one of the whole swarm, which was used as a metric of the assembling degree around the best fitness value. In the process of evolution, introduced a disturbing term in the LDW factor formula when the density factor was larger than a particular constant in order to scatter the particle swarm and leap the local minimum. The simulation tests show that the new PSO algorithm avoids the premature phenomenon in a sense especially for high dimensional and multiple extremum scenarios.