摘要
利用MT理论对一般动态系统的分解提出了混合神经网络建模的方法 ,混合神经网络模型采用动态神经网络和静态神经网络级联的方式 ,结合了二者的优点 ,克服了各自的不足之处 .探讨了采用混合神经网络对一般动态系统建模的方法 .研究结果表明 ,此方法对于一般动态系统 ,特别是复杂的非线性动态系统的建模有其独特的优点 .
Based on mathematical general systems theory, a general dynamical system can be divided into two parts. The composite neural networks was used to model general dynamical systems. The composite neural networks were composed of recurrent neural networks and multiplayer feedforward neural one. It has the advantages of the two networks. The methods on how to use this kind of neural networks to model a general dynamical system were discussed. The simulation results showed that this method had its own advantage in modeling general dynamical system, especially in modeling complex non-linear dynamical system.
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2003年第5期31-33,共3页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目 (79970 0 2 5 )
关键词
一般动态系统
MT理论
混合神经网络
general dynamical system
MT theory
composite neural networks
