基于粒子群算法的ANFIS模型参数优化  

Parameters Optimization of ANFIS Based on Particle Swarm Optimization

作　　者：王介生[1] 

机构地区：[1]辽宁科技大学电子信息与工程学院,辽宁鞍山114044

出　　处：《石油化工高等学校学报》2007年第3期41-44,共4页Journal of Petrochemical Universities

基　　金：国家自然科学基金项目(60474058)

摘　　要：粒子群优化算法是一类全局随机进化算法,算法通过粒子间的相互作用发现复杂搜索空间中的最优区域。根据粒子群算法对整个参数空间进行高效并行搜索的特点,提出了最小二乘法和粒子群优化算法相结合的混合学习算法对自适应神经-模糊推理系统网络结构参数进行优化设计。混合学习算法提高了网络参数辨识的收敛速度,仿真结果表明本算法的有效性。Particle swarm optimization （PSO） is a global stochastic evolutionary algorithm. It tries to find optimal regions of complex searching space through the interaction of particles in the population. Based on the PSO algorithm characteristics of searching the parameter space concurrently and efficiently, the structure parameters of adaptive network--based fuzzy inference system （ANFIS） were tuned by the hybrid algorithm which combined particle swarm optimization with least--square method. The new algorithm greatly raises the convergence speed of network parameters identification and computation. It is demonstrated by numerical simulations that the designed algorithm is effective.

关 键 词：自适应神经-模糊推理系统 粒子群优化算法 最小二乘法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]