An Improved PSO Algorithm with an Area-Oriented Learning Strategy

Tengjun Liu, Jinzhuo Chen, Yuelin Rong, Yuting Zheng, Lijing Tan

Abstract: The classical particle swarm optimization (PSO) trains the particles to move toward the global best particle in every iteration. So, it has a great possibility of being trapped into local optima. To deal with this issue, this paper improves the learning strategy of PSO. Therefore, an area-oriented particle swarm optimization (AOPSO) is proposed, which contributes to leading the particles to move toward an area surrounded by some suboptimal particles besides the best one. 10 test functions are employed to compare the performance of AOPSO with the classical PSO and 3 other improved PSOs. AOPSO performs the best in 5 test functions and relatively better than some of the other algorithms in the rest, which sufficiently demonstrates the effectiveness of AOPSO.