摘要
对系统辨识原理、基于神经网络(ANN)的动态系统辨识进行了分析,针对动态系统辨识模型描述的复杂性,为简化ANN辨识建模的输入/输出关系的表达,提高算法的简洁性,采用了状态方程辨识模型,并给出了基于ANN的动态系统状态方程辨识模型。为比较分析不同网络结构的辨识建模效果及网络模型泛化能力,针对三种不同网络结构方案进行了辨识建模仿真研究。仿真结果显示,基于ANN的动态系统状态方程模型的辨识建模是有效的,并且简单合理的网络结构方案,可提高网络辨识模型的泛化能力。
The identification modeling principle of system and identification modeling for dynamic system based on artificial neural network (ANN) are analyzed. In order to avoid complex description and improve the simplicity of algorithm for model of dynamic system, a model of state equation has been adopted. And the identification model of state equation for dynamic system based on ANN is given. For comparing the identification effects and generalization ability of different network architectures of ANN model, three kinds of different network architecture schemes of state equation models for dynamic system are simulated. Simulation results show that the identification model of state equation for dynamic system based on ANN is effective. And the simple and reasonable architecture of ANN can raise the generalization ability of ANN model.
出处
《计算机仿真》
CSCD
2006年第10期144-146,共3页
Computer Simulation
关键词
神经网络
动态系统
状态方程
辨识建模
仿真
ANN
Dynamic system
State equation
Identification modeling
Simulation