基于改进BP网络的地震动信号目标识别
详细信息 本馆镜像全文    |  推荐本文 | | 获取馆网全文
摘要
应用人工神经网络进行目标识别是当前模式识别的重要方法之一。前向多层神经网络及其BP算法是发展较为成熟的一种。该文对BP算法加以改进 ,使得其性能有所提高 ,收敛速度加快。针对战场监视传感器系统中处于一级警戒的地震动传感器 ,对在良好土质地面实测的人员脚步、汽车、坦克的地震动信号进行分析 ,利用小波变换和小波包分解提取能量特征 ,采用两级级连网络进行目标识别 ,识别率在 94.5 %以上
It is one of the important methods of pattern recognition to apply neural networks to target classification. Forward propagation multi layers neural networks and its BP algorithm are used widely. In this article, some measures are taken to improve BP algorithm, and to make its performance better and its convergence speed quicker.The seismic sensor is an essential sensor in battlefield watching system.By testing,a great number of seismic signals are obtained on footsteps, wheeled vehicle and tank. These signals are processed using wavelets transform and wavelets package. The energy spectrum features of these signals are extracted, and two series connected BP neural networks identify them. The results of 94.5% proper identification are attained.
引文
1 徐佩霞 ,孙功宪 小波分析与应用实例 .合肥 :中国科学技术出版社 .1 996
    2 黄得双 .神经网络模式识别系统理论 .北京 :电子工业出版社 .1 996
    3 郁文贤 .军事电子信息处理中的人工神经网络技术 .国防科技大学学报 ,1 998,2 0 (3) :92~ 98
    4 赵玉霞 .多传感器监视系统的目标识别 :[硕士学位论文 ] .南京 :南京理工大学 ,1 999

版权所有:© 2023 中国地质图书馆 中国地质调查局地学文献中心