Abstract: Quantum-behaved particle swarm optimization algorithm is analyzed, particles searching action and local attract point are studied. To the different searching environment in searching progress, the searching actions are divided into four models. The proposed algorithm can self-learn the optimization problem, and utilize a suitable learning model, then the whole optimization performance is increased. The comparison and analysis of results with the proposed method and other improved QPSO based on CEC2005 benchmark function are given, the simulation results show the modified algorithm can greatly improve the QPSO performance.

Key words: Particle Swarm Optimization（PSO） algorithm, self-learning, local attract point, searching model

摘要： 分析了量子行为粒子群优化算法，着重研究了算法中群体粒子的搜索行为，对算法中局部吸引点进行了分析，提出针对粒子在搜索过程中所处的不同搜索环境，将粒子的搜索行为分为四种类型，并能够自适应地学习优化问题环境，采用合适的学习模式，提高算法整体优化性能；将改进后的自学习量子粒子群算法与其他一些改进方法通过CEC2005 benchmark测试函数进行了比较，最后对结果进行了分析，仿真结果显示自学习方法能够显著改善量子粒子群优化算法的性能。

关键词: 粒子群算法, 自适应学习, 局部吸引点, 搜索模式