摘要
为了克服标准量子粒子群优化(SQPSO)算法易陷入局部最优的缺点,引入变异机制,基于进化阶段的概念,提出了自适应阶段变异量子粒子群优化(APMQPSO)算法。以四种不同的变异概率减小方式阶段性地对QPSO算法中的全局最优位置进行柯西变异,形成了四个不同的APMQPSO算法。用五个典型的测试函数进行仿真实验,并将四个APMQPSO算法与SQPSO算法的实验结果进行了比较。实验结果表明,对于单峰函数优化问题,基于变异概率线性变化的APMQPSO算法较为有效;而对于多峰函数优化问题,基于变异概率非线性变化的APMQPSO算法则具有很强的优化能力。
The standard quantum particle swarm optimization（SQPSO） algorithm may sink into local optimum.To overcome this shortcoming,this paper introduced the mutation mechanism.Based on the concept of evolution period,it proposed adaptive period mutation-based QPSO algorithms（APMQPSOs）.It used four kinds of mutation probability decreasing methods to periodically mutate global best position with cauchy random numbers in QPSO algorithm,thus formed four different APMQPSO algorithms.It adopted five typical test functions to conduct simulation experiment,and compared experimental results of four APMQPSOs and SQPSO with each other.The experiment results show that APMQPSOs with linear variation mutation probability are effective for unimodal function optimization problems,while algorithms with nonlinear variation mutation probability have very strong optimization abilities for multimodal ones.
出处
《计算机应用研究》
CSCD
北大核心
2012年第6期2035-2039,2051,共6页
Application Research of Computers
基金
国家自然科学基金资助项目(90818025)
关键词
量子粒子群优化算法
进化阶段
变异算子
变异概率
函数优化
QPSO algorithm
evolution period
mutation operator
mutation probability
function optimization
