摘要
采用多层前馈网络结构进行动态建模，并用Ｄａｖｉｄｏｎ最小二乘法作为在线学习算法，将辨识后得到的模型进行线性化．基于线性化模型设计广义预测控制器。将其与非线性前馈相结合，建立了一种适合于非线性系统的前馈补偿广义预测自校正控制器．
Using a multilayer forward neural networks to model the dynamic systems and Davidon least square algorithm as the networks'learning method, after identification, linearizes the obtained model. Then based upon the linear model, the generalized predictive controller is build. combining it nonlinear forward gain compensation, we constructs a forward compensation generalized predictive self tuning controller which is suitafle to nonlinear plant. The simulations verified the effectiveness of the presented controller.
出处
《南开大学学报（自然科学版）》
CAS
CSCD
北大核心
1999年第2期51-55,共5页
Acta Scientiarum Naturalium Universitatis Nankaiensis
基金
国家863应用基础研究项目
天津市自然科学基金
关键词
非线性控制
神经网络
广义预测控制
自校正控制
nonlinear control
neural networks
generalized predictive control
forward gain compensation
self tuning control
Davidon least square
