非线性动态系统辨识的神经网络结构和可行性研究  

The Structures and Feasibility of Using Neural Networks for the Identification of Nonlinear Dynamic Systems

作　　者：王科俊[1] 李国斌 李殿璞[1] 

机构地区：[1]哈尔滨工程大学自动控制系

出　　处：《哈尔滨工程大学学报》1997年第2期52-57,共6页Journal of Harbin Engineering University

摘　　要：在讨论各种非线性动态系统辨识模型的基础上，给出相应模型的神经网络实现方案，并首次提出实现非线性动态系统的回归状态模型的新型神经网络结构—神经网络状态空间辨识模型．从理论上证明了使用神经网络实现这些模型的可行性．In this paper,based on discussing a variety of models for the identification of nonlinear dynamic systems,the neural network methods for these models are given.A novel neural network structure-the neural network state space identification model for the recurrent state model of nonlinear dynamic systems,is proposed.The feasibility of using neural networks for the identification of nonlinear dynamic systems is proved with the approximative theories of neural networks.

关 键 词：神经网络 非线性动态系统 系统辨认 可行性 

分 类 号：O231[理学—运筹学与控制论] TP13[理学—数学]