摘要
针对标准粒子群算法容易陷入局部极值和精度低的问题,提出一种嵌入极值优化算法的粒子群优化算法。在线性下降的惯性权重粒子群算法运行过程中,间隔一定迭代次数与极值优化算法相结合,利用其波动性增加种群的多样性,并有效结合粒子群算法较强的全局探索能力和极值优化算法精细的局部搜索性能,以较高精度收敛到全局极值。仿真实验结果表明,该混合算法是一种求解高维多峰连续函数极值的有效方法。
To overcome the problems of premature convergence frequently and low accuracy computationin in Particle Swarm Optimization (PSO), a novel hybrid algorithm, called hybrid EPSO, is proposed. Extremal Optimization(EO) is introduced into PSO at a certain iteration intervals. The hybrid algorithm uses the volatility of EO to increase the diversity of population, on the other hand, it elegantly combines the exploration ability of PSO with the fine-grained ability of EO, it avoids prelnature convergence of PSO with high accuracy. Simulation experimental results show that the hybrid algorithm is an effective way to locate global optima of continuous multimodal functions of high dimensions.
出处
《计算机工程》
CAS
CSCD
北大核心
2011年第8期172-174,共3页
Computer Engineering
基金
国家自然科学基金资助项目"过程控制系统的一类设定点优化方法研究"(60874070)
中南大学研究生学位论文创新基金资助项目"过程控制系统设定点全局优化的粒子群算法研究"(2009ssxt190)
关键词
粒子群优化算法
极值优化
混合柯西-高斯变异
混合算法
Particle Swarm Optimization(PSO) algorithm
Extremal Optimization(EO)
hybrid Cauchy-Gaussian mutation
hybrid algorithm