摘要
针对静态模糊神经网络对动态系统辨识精度低的特点,在T-S模糊神经网络标准结构基础上,通过在输入层与状态层间加入可以记忆暂态信息的递归层,一种新的T-S递归型模糊神经网络(TSRFNN)被提出,来提高对动态系统的辨识能力.同时,给出了参数的动态BP学习算法.通过仿真实验,证明提出的TSRFNN对动态非线性系统的辨识比传统静态模糊神经网络(TFNN),具有更快的网络收敛速度,更高的辨识精度,更适合于动态系统的辨识.
In order to improve the identification ability of dynamic system, a new T-S recurrent fuzzy- neural network (TSRFNN) is proposed based on the traditional static T-S fuzzy neural networks (TSNN), A recurrent layer is added between the states layer and inputs layer in the new fuzzy-neural network to memory the temporal information. The dynamic back propagation (DBP) learning algorithm are deduced to demonstrate the system, It is shown that the proposed TSRFNN is better than TFNN used in dynamic system identification by simulation.
出处
《北京建筑工程学院学报》
2006年第3期44-47,共4页
Journal of Beijing Institute of Civil Engineering and Architecture
基金
北京建筑工程学院青年基金项目(资助号:1005023)
关键词
T—S递归型模糊神经网络
动态系统
自适应学习算法
辨识
T-S recurrent fuzzy-neural networks
dynamic system
adaptive learning algorithm
iden- tification