摘要
提出采用多层局部回归神经网络建立多变量非线性系统多步预测模型的方法,神经 网络模型可提供多步预测控制所需要的系统输出预测值及输出向量对控制向量的雅可比矩 阵.仿真试验表明这种动态神经网络的预测模型具有较高的精度.
The multistep predictive modeling of MIMO nonlinear system based on multilayer local recurrent neural networks is presented. The predictive outputs and Jocobian matrixs of output vecter versus input vecter are proffered by the neural networks predictive model for the multistep predictive control systems. The results of simulation show that predictive model of the dynamic neural networks can reach higher degree of accuracy.
出处
《北京科技大学学报》
EI
CAS
CSCD
北大核心
2000年第2期190-192,共3页
Journal of University of Science and Technology Beijing
基金
国家"八五"攻关项目! 85-311-02-11-04
关键词
非线性系统
局部回归网络
神经网络
预测模型
multivariable nonlinear system
multilayer local recurrent neural networks
multistep predictive modeling
Jocobian matrixs