摘要
针对一类非线性系统的未知时变参数,提出了一种模型算法学习辨识方法。该方法主要是利用系统模型估计出了辨识参数的偏差,利用这一参数偏差的估计来修正辨识参数,不断进行迭代。并严格证明了系统经过迭代学习后,辨识器的输出能够完全跟踪参数真值,同时得到了该算法收敛的范数形式充分条件。该方法不仅可以实现非线性系统未知时变参数在有限时间区间上的完全辨识,而且还克服了传统迭代学习辨识器中凭借经验选取学习增益的盲目性,加快了参数辨识器的收敛速度。仿真结果验证了该方法的有效性。
A learning identification method of model algorithm is proposed aiming at a class of nonlinear system with unknown time-varying parameters.This method mainly uses system model estimation to identify parameter deviation,which is used to modify identification parameters,and iterates continuously.It is strictly proved that output of identifier can track true values of parameters after iteration learning,and meanwhile sufficient condition of norm form that the method converges to can be drawn.This method can not only realize complete identification of nonlinear system unknown time-varying parameters in finite time region,but also overcome the blindness that traditional iteration learning identifier chooses learning gain by experiences,and can fasten the convergent speed of parameter identifier.Simulation results verify the validity of this method.
出处
《武汉理工大学学报》
CAS
CSCD
北大核心
2011年第12期115-119,共5页
Journal of Wuhan University of Technology
基金
国家自然科学基金(61100103)
黑龙江省教育厅科学技术研究支持项目(12511600)
关键词
非线性系统
时变参数
模型算法学习
辨识器
nonlinear systems
time-varying parameters
model algorithm learning
identifier
