摘要
为了解决饱和非线性维纳(Wiener)系统的参数辨识问题,提出了基于辅助模型的可变遗忘因子随机梯度算法。首先,由于饱和非线性的特殊结构,采用了切换函数变换非线性表达式,使所有未知参数包含在一个向量中,将系统模型转换为线性回归形式。然后,为了获得未知的中间变量,构造辅助模型,运用辅助模型的输出替换信息向量中的未知内部变量。最后,为了提高随机梯度算法的收敛速度,在算法中引入了可变遗忘因子。仿真结果表明,与传统的随机梯度算法相比,所提算法的参数估计更精确,且收敛速度更快,验证了所提算法的有效性。
In order to solve the problem of the parameter identification for Wiener systems with saturation nonlinearity,an auxiliary model-based variable forgetting factor stochastic gradient algorithm is proposed.Firstly,due to the special structure of the saturation nonlinearity,a switching function is employed to transform the expression of the nonlinearity,so that all the unknown parameters are included in one vector,and the system model is transformed as a linear regression form.Then,in order to obtain the unknown intermediate variables,the auxiliary model is constructed,and unknown inner variables in the information vector are replaced with the outputs of the auxiliary model.Finally,to improve the convergence speed of stochastic gradient algorithm,the variable forgetting factor is introduced into the algorithm.Simulation results show that the proposed algorithm has higher parameter estimation accuracy and faster converge speed than the conventional stochastic gradient algorithm,which verifies the effectiveness of the proposed algorithm.
作者
汪菲菲
马君霞
WANG Feifei;MA Junxia(School of Intelligent Equipment Engineering,Wuxi Taihu University,Wuxi 214063,China;School of Internet of Things Engineering,Jiangnan University,Wuxi 214122,China)
出处
《控制工程》
CSCD
北大核心
2024年第4期760-768,共9页
Control Engineering of China
基金
国家自然科学基金青年项目(61803183)
江苏省自然科学基金资助项目(BK20180591)。
关键词
参数估计
随机梯度
饱和非线性
可变遗忘因子
Parameter estimation
stochastic gradient
saturation nonlinearity
variable forgetting factor
