基于Memetic框架混合群智能算法的NARMAX模型参数辨识  

Parameter Identification of NARMAX Model Using a Hybrid Swarm Intelligence Algorithm Based on Memetic Framework

作　　者：田谦益[1] 王小北[2] 

机构地区：[1]漳州师范学院计算机科学与工程系,福建漳州363000 [2]福建省热带作物科学研究所科技信息管理科,福建漳州363000

出　　处：《山东科技大学学报（自然科学版）》2013年第1期88-94,共7页Journal of Shandong University of Science and Technology(Natural Science)

基　　金：福建省教育厅科技项目(JA09162)

摘　　要：针对NARMAX模型的参数辨识问题,提出一种新颖的混合群智能算法。该算法在Memetic算法框架的基础上,采用粒子群算法与人工蜂群算法融合作为全局搜索策略,采用单纯形优化法作为局部搜索策略。针对三个复杂程度不同的NARMAX模型进行的参数辨识仿真实验,结果显示,与标准人工蜂群算法和随机惯性权重粒子群算法相比,新算法无论在鲁棒性还是求解精度上都具有一定优势。Aiming at the parameter identification of NARMAX model, a novel chaos-artificial bee colony algorithm was proposed. Based on the Memetie algorithm framework,the new algorithm uses the fusion of the particle swarm optimization and the artificial bee colony algorithm as the global search strategy and uses the simplex optimization as the local search strategy. Through the simulation experiment on the parameter identification of three different NARMAX models in complexity, the results show that the new algorithm, compared with the artificial bee colony algorithm and the random inertia weight particle swarm optimization,has some advantages in the robustness or solutions＇ quality.

关 键 词：优化 粒子群算法 人工蜂群算法 非线性系统 参数辨识 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]