地球物理学进展
地毬物理學進展
지구물이학진전
PROGRESS IN GEOPHYSICS
2009年
6期
2302-2307
,共6页
支持向量机%路基病害%探地雷达%信号识别
支持嚮量機%路基病害%探地雷達%信號識彆
지지향량궤%로기병해%탐지뢰체%신호식별
support vector machine%roadbed disease%ground penetrating radar%signal recognition
本文通过对支持向量机原理以及路基病害特征的讨论和分析,提出了基于支持向量机的探地雷达回波信号识别算法;利用所提出的算法对探地雷达实测数据进行处理,试验结果表明该算法的识别效果优于神经网络算法,并且该算法克服了神经网络的过学习和局部极小值的缺点,是一种适合路基病害识别的高效算法.
本文通過對支持嚮量機原理以及路基病害特徵的討論和分析,提齣瞭基于支持嚮量機的探地雷達迴波信號識彆算法;利用所提齣的算法對探地雷達實測數據進行處理,試驗結果錶明該算法的識彆效果優于神經網絡算法,併且該算法剋服瞭神經網絡的過學習和跼部極小值的缺點,是一種適閤路基病害識彆的高效算法.
본문통과대지지향량궤원리이급로기병해특정적토론화분석,제출료기우지지향량궤적탐지뢰체회파신호식별산법;이용소제출적산법대탐지뢰체실측수거진행처리,시험결과표명해산법적식별효과우우신경망락산법,병차해산법극복료신경망락적과학습화국부겁소치적결점,시일충괄합로기병해식별적고효산법.
Based on analysis of the principle of support vector machines proposed characteristics of roadbed diseases,a new GPR echo signal recognition algorithm is proposed for identifying the ground-penetrating radar measured data. The result shows that this algorithm is better than the neural network recognition algorithm and overcomes the shortcomings of the local minimum value and over-learning.The algorithm is suitable for roadbed disease recognition and is an efficient algorithm.
