Globally-optimal prediction-based adaptive mutation particle swarm optimization

• DOI: 10.1016/j.ins.2017.07.038
• Publication Date: 2017-08-01T17:30:58Z
• Authors: Quanlong Cui; Qiuying Li; Gaoyang Li; Zhengguang Li; Xiaosong Han; Heow Pueh Lee; Yanchun Liang; Binghong Wang; Jingqing Jiang; Chunguo Wu
• Keywords: Adaptive mutation; Benchmark (surveying); Premature convergence; Local optimum

Particle swarm optimizations (PSOs) are drawing extensive attention from both research and engineering fields due to their simplicity powerful global search ability. However, there two issues needing be improved: one is that the classical PSO converges slowly; other tends result in premature convergence, especially for multi-modal problems. This paper attempts address these issues. Firstly, improve convergent efficiency, this proposes an asymptotic predicting model of globally-optimal solution, which used predict optimum based on extracting features reflecting evolutionary trend. The predicted then taken as third exemplar, a way similar individual historical best solution guiding process particles. To reduce probability population trapped into local phenomenon, adaptive mutation strategy, help particles escape away by using extended non-uniform operator. Finally, we combine entities develop prediction-based particle optimization (GPAM-PSO). In numerical experimental parts, compare proposed GPAM-PSO with 11 existing variants 22 benchmark problems 30-dimensions 100-dimensions, respectively. Numerical experiments demonstrate could accuracy efficiency remarkably, means combination strategy accelerate convergence phenomenon effectively. Generally speaking, performs most efficiently robustly. Moreover, performance problem demonstrates practical application algorithm.