非线性输出误差模型的两阶段递推辨识算法  

The two-stage Recursive Identification Algorithm for the Model of Nonlinear Output Error

作　　者：贾杰[1] 范奎伟 

机构地区：[1]中国科学院大学,北京100049 [2]南昌航空大学信息工程学院,南昌330063

出　　处：《南昌航空大学学报（自然科学版）》2014年第2期24-31,共8页Journal of Nanchang Hangkong University(Natural Sciences)

基　　金：国家自然科学基金(61263012);中国博士后科学基金(2012M510593)

摘　　要：输入非线性系统的输出误差模型在实际工业生产中是一类常见模型,针对含有色噪声的输出误差模型提出基于辅助模型的两阶段递推增广最小二乘算法。根据辅助模型思想和分解技术,将复杂的非线性辨识系统分解为系统模型和噪声模型子系统,再根据最小二乘思想分别辨识,其中噪声信息向量中存在的不可测噪声项用其估计值代替。最后与递推增广最小二乘算法在参数估计精度和收敛速度的比较,验证算法在此类模型应用的有效性,仿真结果表明该算法精度高,收敛速度快,计算量小。The output error model of nonlinear system is a kind of common models in actual industrial production. A class of two-stage recursive extended least squares algorithm based on auxiliary model is proposed for the nonlinear output error model with colored noise in this paper. According to the auxiliary model ideas and decomposition techniques, the identification of nonlinear complex system is decomposed into sub-system models and noise models. Then the sub-models parameters are separately identified through the ideas of the least squares, where the unpredictable noise term exists in the noise information vector are replaced by its estimated values. Finally, the algorithm is compared with the recursive augmented least squares algorithm in respects of the parameter estimation accuracy and convergence rate. The simulation results show that the algorithm has high precision, fast convergence rate and a small amount of computation.

关 键 词：非线性输出误差模型 有色噪声 两阶段最小二乘算法 辅助模型 递推辨识 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]