| 基于递归模糊神经网络的鲁棒BP算法研究 |
| |
| 引用本文： | 沈晓蓉,范耀祖,李敬.基于递归模糊神经网络的鲁棒BP算法研究[J].航天控制,2004,22(3):26-30. |
| |
| 作者姓名： | 沈晓蓉 范耀祖 李敬 |
| |
| 作者单位： | 1. 北京航空航天大学,北京,100083
2. 北京航天指挥控制中心,北京,100094 |
| |
| 摘 要： | 讨论了一种基于递归模糊神经网络的鲁棒BP算法的实现。该算法采用极大似然法修正传统BP算法的指标函数 ,利用M估计器和自举算法在线估计样本的误差分布 ,适合于动态非线性系统的辨识 ,逼近精度高 ,收敛速度快 ,在小噪声扰动时处于稳定 ,对过失误差有很强的鲁棒性。仿真结果证明了该算法的有效性
|
| 关 键 词： | 递归模糊神经网络 鲁棒BP算法 极大似然法 M估计器 自举法 |
| 文章编号： | 1006-3242(2004)03-26-05 |
| 修稿时间： | 2003年11月10 |
A Study on Robust BP Algorithm Based on Recursive Fuzzy Neural Network |
| |
| Shen Xiaorong,Fan Yaozu,Li Jing.A Study on Robust BP Algorithm Based on Recursive Fuzzy Neural Network[J].Aerospace Control,2004,22(3):26-30. |
| |
| Authors: | Shen Xiaorong Fan Yaozu Li Jing |
| |
| Institution: | Shen Xiaorong2 Fan Yaozu2 Li Jing1 1.Beijing Aerospace Control Center,Beijing 100094 2.Beijing University of Aeronautics and Astronautic,Beijing 100083 |
| |
| Abstract: | A robust BP algorithm bas ed on recursive fuzzy neural network that is resistant to the noise effects and is capable of rejecting gross errors during the approximation process is derived. The index function of robust BP algorithm is amended with maximu m likelihood method, while the M-estimator and bootstrap method are employed to estimate the error probability distribution. Compared the BP alg orithm, the robust BP algorithm not only have higher approximati on accuracy, but also have faster convergence rate. The simulation results in th e last section illustrate that the robust BP algorithm is effect ive.< |
| |
| Keywords: | Recursive fuzzy neural network Robust BP algorithm Maximum lik elihood method M-estimator Bootstrap method |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
