Abstract: To overcome the disadvantage of Particle Swarm Optimization(PSO) algorithm such as premature, bad convergence precision, based on feedback of swarm diversity, a PSO algorithm with Dynamic Learning Objects(PSO-DLO) is presented. In the algorithm swarm diversity is used to control the learning objects, the strategy relieves the lost of swarm diversity, which is helpful for the global search. Experiments of three typical multi-modal functions indicate that the algorithm can effectively avoid premature and achieve better global search ability.

Key words: Particle Swarm Optimization(PSO), premature, feedback, swarm diversity, multi-modal function

摘要： 针对粒子群优化算法容易早熟、收敛精度低等问题，基于群体多样性反馈的思想，提出一种动态学习对象的粒子群优化算法。该算法采用群体多样性动态控制粒子的学习对象，减缓群体多样性的丧失速度，有利于群体的全局寻优。对3种典型多峰函数的仿真结果表明，该算法可以有效避免早熟问题，具有较好的全局寻优能力。

关键词: 粒子群优化, 早熟, 反馈, 群体多样性, 多峰函数

CLC Number: 

• TP 18