并行混沌粒子群优化研究及应用  

Research and Application of Parallel Chaos Particle Swarm Optimization

作　　者：王华秋[1] 曹长修[1] 

机构地区：[1]重庆大学自动化学院

出　　处：《计算机仿真》2005年第11期98-101,共4页Computer Simulation

基　　金：国家教育部博士点基金项目(98061117);重庆市教委基础研究项目(020612);重庆市科委攻关项目(20020828)

摘　　要：粒子群优化算法(PSO)是进化计算领域中的一个新的分支,其源于鸟群和鱼群群体运动行为的研究。针对粒子群优化容易陷入局部极值点的问题,该文提出并行混沌粒子群算法,该算法保持了群体多样性从而避免了早熟,经过计算机仿真实验这是一种高效的优化算法。针对转炉提钒过程是一个多元非线性反应过程而难以建立终点控制模型的问题,提出了并行混沌粒子群RBF算法的方法建模。克服了RBF中心个数选择的随机性。试验表明,用该算法预测冷却剂加入量误差较小,具有工程实用性。The particle swarm optimization algorithm is a new branch in evolution computing field. It is originated from the research of swarm intelligence moving actions of birds and fishes. To improve performance of original particle swarm optimization algorithm and avoid trapping to local excellent result, the paper presents a parallel chaos particle swarm optimization. The proposed algorithm can maintain the individual diversity and restrain the degeneration phenomenon. The simulation results are satisfactory. The converter vanadium recover is a very sophisticated reaction, and it is difficult to build up end - point control model. The paper presents a modeling method based on parallel chaos PSO RBF neural network, which can perfectly resolve the problem of random selection of RBF cluster center number. Finally actual project results indicate that the error is less than before and the algorithm has the engineering practicability.

关 键 词：并行算法 混沌 粒子群优化 计算机仿真 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP338.6[自动化与计算机技术—控制科学与工程]