A Hybrid Particle Swarm Optimization Improved by Mutative Scale Chaos Algorithm

When using the standard particle swarm optimization to optimize the complex problems with high dimension, low convergence efficiency and falling into local optimization usually occur because of its inherent disadvantages. To avoid these disadvantages, a novel hybrid particle swarm optimization improved by mutative scale chaos method is proposed in this paper. This hybrid algorithm combines global high-speed convergence ability of particle swarm optimization with chaos method´s advantage, i.e., breaking away from local optimal points easily. The variance of the population´s fitness is used to judge premature state of the whole population. The searching space of chaos method can be reduced dynamically by mutative scale scheme, and then searching efficiency of the proposed algorithm is improved further. The test results for benchmark functions show that this novel hybrid algorithm not only surpasses the standard particle swarm optimization obviously in many respects, such as optimization precision, efficiency, success ratio and so on, but also has good stability and low sensitivity to different dimensions of functions.