基于U模型的输入加权自适应逆控制  

AN input weighted adaptive inverse control method based on U model

作　　者：李联飞[1] 李文 史伟[1] 苏振华[1] 贾海亮[1] LI Lian-fei;LI Wen;SHI Wei;SU Zhen-hua;JIA Hai-liang(China National Heavy Machinery Research Institute Co.,Ltd.,Xi’an 710032,China;;Rizhao Steel Holding Group Co.,Ltd.,Rizhao 276806,China)

机构地区：[1]中国重型机械研究院股份公司,陕西西安710032 [2]日照钢铁控股集团公司,山东日照276806

出　　处：《重型机械》2018年第2期50-54,共5页Heavy Machinery

摘　　要：U模型面向控制,通过实时辨识网络的结构参数,进而通过多项式求根得到控制量,而不需要另外进行控制器的设计。本文将U模型同神经网络进行了结合,通过同系统并联运行的U模型网络及改进的神经网络的结构参数,各自求取控制量。由辨识误差的实时大小对两个控制量进行加权,完成控制给定。仿真表明结合后的网络在理想状态下以及噪声存在状态下都具有更好的控制效果。U model is a control oriented model which calculates its control input by solving the root of a polynomial equation.In this paper,a combination of the U model and NN is given.The proposed model and U model runs with the plant in parallel,they calculate their network parameters and control input separately.An algorithm is proposed to weight these two control inputs by a function of their identification error which forms a real control input.Simulation shows that this proposal performs better both in normal and noise existed condition than the un-weighted ones.

关 键 词：U模型 控制输入加权 逆模型 参数辨识 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]