人工神经网络的一种改进的B-P学习算法及其在雷达目标识别中的应用  

A Modified Back-Propagation Learning Algorithm for Artificial Neural Network and Its Applications to Radar Target Recognition

作　　者：梁民 孙仲康[1] 

机构地区：[1]国防科技大学

出　　处：《现代雷达》1992年第2期49-55,共7页Modern Radar

摘　　要：本文首先简要地介绍了人工神经网络(以下简称神经网络)的B-P学习算法。讨论了一些加速B-P学习算法收敛的措施,进而提出了一种改进的B-P学习算法(MB-P)。将这两种算法分别应用于舰船雷达目标识别系统。仿真结果表明,MB-P学习算法的收敛速度比B-P算法快得多,分别用这两种学习算法训练的神经网络对舰船雷达目标的识别率大致相同。The back-propagation (B-P)learning algorithm for artificial neural network is reviewed at first. Some modifications to B-P algorithm are discussed. The corresponding modified back-propagation (MB-P)algorithm is proposed and applied to marine radar target identification system. Simulation results show that characters of MB-P learning algorithm are better than B-P algorithm.

关 键 词：神经网络 B-P算法 目标识别 雷达 

分 类 号：TN959.7[电子电信—信号与信息处理]