文献类型：期刊文献

中文题名：基于精英克隆选择的粒子群优化算法研究与应用

英文题名：Study and application of modified particle swarm optimization algorithm based on elitist clone choice

作者：年笑宇[1];王昕[2];王振雷[1];钱锋[1]

机构：[1]化工过程先进控制和优化技术教育部重点实验室(华东理工大学),上海200237;[2]上海交通大学电工与电子技术中心,上海200240

年份：2012

卷号：29

期号：1

起止页码：5

中文期刊名：计算机与应用化学

外文期刊名：Computers and Applied Chemistry

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD2011_2012】；

基金：国家高技术研究发展计划(863)(2008AA042902);上海科技攻关项目(10dz1121900);上海市重点学科建设项目资助(B504)

语种：中文

中文关键词：精英克隆选择;粒子群(PSO);变异

外文关键词：elitist clone choice; PSO; variation

摘要：粒子群算法(PSO)是一种基于迭代的智能算法,具有较好的全局搜索能力,但局部搜索能力较弱。针对粒子群算法容易陷入局部最优不足这一问题,本文提出了一种精英克隆选择的方法。该算法在基本粒子群算法的基础上保留了种群中的若干个精英粒子,然后将这些精英粒子进行克隆复制,并将复制之后的粒子进行变异操作,再将变异之后的粒子与变异前的粒子进行亲和度的比较,保留下来亲和度提高的粒子并替换之前适应值较差的粒子,通过这种方式增强了种群的多样性,从而避免了粒子陷入局部最优的问题。此外,本文引入了新的改进惯性权重的机制,根据粒子位置和速度的情况自适应地改变惯性权重,这样避免了粒子盲目运动,更有针对性的寻找最优解。对4个高维复杂函数寻优测试,分别从平均精度和标准差这两方面进行分析,结果表明改进之后的算法在寻优精度和稳定性方面都超过了基本PSO,从仿真图像中可以看出改进之后的算法在迭代末期跳出了局部最优更接近全局最优值。最后将这种改进的算法应用到优化乙烯、丙烯的收率模型中,应用结果表明当裂解原料属性发生改变时,本文提出的算法可以很快完成对操作变量的寻优,显著提高了"双烯"收率。
Particle Swarm Optimization （PSO） is an intelligent algorithm based on iteration. It has a good ability of searching the optimum in the whole, but the capacity of searching in the local is poor. Therefore, to solve this problem of converging to local optimum for PSO, a modified method based on elitist clone choice is proposed. This optimization retains several elitist particles of the tracLitional PSO and clones them, then makes the variation of them. Furthermore, the particles after variation will be compared with the ones before variation and the particles which fitness is improved will be kept to replace the particles which fitness is bad in the old swarm. By means of this method, the diversity of swam is increased, which prevents the particle from converging to the local optimum. In addition, a novel modified inertia weight method is proposed in this paper, which changes the inertia weight according to the position and velocity of the particle adaptively. It can prevent the particle from motion blindly and has a good ability to find global optimization. Then four complex functions with high dimension are employed for two aspects of the mean value and the standard deviation, the results show that the improved algorithm is much better than the traditional PSO in precision and stability. It can also be seen from the simulation result that the modified algorithm jumps out of the local optimum in the late iteration and is close to the global optimum. Finally the improved algorithm is applied to optimize the model of the ethane and propylene yield. Application results show that when the feature of feedstock changes it can help to find the optimal operating condition as soon as possible. Meanwhile, the yield of ethane and propylene is promoted observably.