| U-D分解渐消记忆滤波神经网络控制器在船舶操纵中的应用 |
| |
| 引用本文： | 常依斌,刘以建,李海量.U-D分解渐消记忆滤波神经网络控制器在船舶操纵中的应用[J].上海海事大学学报,2002,23(4):6-8. |
| |
| 作者姓名： | 常依斌 刘以建 李海量 |
| |
| 作者单位： | 1. 上海海运学院,工学院,上海,200135
2. 上海海运学院,研究生部,上海,200135 |
| |
| 摘 要： | 针对前馈网络BP算法所存在的收敛速度慢且常遇局部极小值等缺陷 ,提出一种基于U D分解的渐消记忆推广Kalman滤波学习新方法。与EKF相比 ,该方法不仅大大加快了学习收敛速度、数值稳定性好 ,而且比BP算法需较少学习次数和隐节点数 ,学习效果也更好。将这种学习算法应用在船舶操纵的神经网络控制器中 ,仿真结果表明该方法是提高网络学习速度、改善学习效果的一种有效方法 ,可有效解决非线性系统的控制问题。
|
| 关 键 词： | 船舶操纵 前馈神经网络 BP算法 推广Kalman滤波 U-D分解滤波 |
| 文章编号： | 1000-5188(2002)04-0006-0004 |
| 修稿时间： | 2002年6月27日 |
A U-D Factorization-Based Extended Kalman Filter Learning Algorithm for Feedforward Neural Networks Controller & Its Application in Ship-Maneuvering |
| |
| CHANG Yi bin,LIU Yi jian,LI Hai liang.A U-D Factorization-Based Extended Kalman Filter Learning Algorithm for Feedforward Neural Networks Controller & Its Application in Ship-Maneuvering[J].Journal of Shanghai Maritime University,2002,23(4):6-8. |
| |
| Authors: | CHANG Yi bin LIU Yi jian LI Hai liang |
| |
| Institution: | CHANG Yi bin 1,LIU Yi jian 1,LI Hai liang 2 |
| |
| Abstract: | BP(Back Propagation) algorithm is a popular algorithm for training multilayered feedforward neural networks. Later EKF(Extended Kalman Filter) is suggested to train feedforward neural networks and improve the convergence of BP algorithm. But EKF algorithm increases computation cost and may suffer from numerical stability. Then a new fast EKF learning algorithm using U D factorization is proposed. It can greatly improve the numerical stability, computational efficiency, accuracy and convergent rate. U D factorization algorithm is applied to a neural networks controller to maneuver ships. The results of simulation show that the new proposed algorithm is a robust and efficient learning algorithm. It can effectively resolve the control problem of nonlinear dynamic system. |
| |
| Keywords: | ship maneuvering feedforward neural networks BP learning algorithm EKF algorithm U D factorization |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《上海海事大学学报》浏览原始摘要信息 |
| 点击此处可从《上海海事大学学报》下载免费的PDF全文 |
