摘要
标准粒子群算法随着迭代次数的增加,整个粒子种群的多样性呈下降趋势,种群很快在当前最优位置的吸引下容易陷入局部最优而无法逃脱。因此,如何增加种群多样性,使粒子逃脱局部最优,成为增强算法全局寻优能力的关键。为了克服粒子群算法早熟收敛的缺点和增加其粒子多样性,通过引入"吸收"、"再生变异"算子,设计了一种新的粒子群优化算法,通过对常用基准函数的数值试验,证明了新算法不仅能有效地避免早熟收敛,而且具有更好的收敛效果。
A disadvantage of the standard particle swarm algorithm is that with the increase of iteration times, the diversity of the whole particle population has a descending trend, which makes population trap easily into the local optimum and can not escape owing to the current optimal position attraction. Therefore, how to increase the diversity of population in order to make particle escape local optimum becomes the key for the enhanced algorithm to acquire global optimization ability. In order to overcome the shortcomings of the standard particle swarm algorithm premature convergence and increase the diversity, this paper designed a new particle swarm optimization algorithm by introdu- cing three operators called ＂absorption＂, ＂regeneration and variation＂. The numerical test results using some com- mon reference functions prove that the new algorithm not only can avoid effectively premature convergence but has better convergence effect.
出处
《计算机仿真》
CSCD
北大核心
2013年第6期308-311,365,共5页
Computer Simulation
基金
四川省教育厅青年基金项目(11ZB058)
关键词
粒子群算法
早熟收敛
多样性
吸收
再生变异
Particle swarm
Premature convergence
Diversity
Absorption
Mutation
