基于GPR信号频带介电谱特征的公路早期病害检测*

现有的探地雷达(GPR)技术都是在公路病害产生后进行病害定位,而不是在病害发生的早期对病害进行预测。为了实现公路早期病害检测,本文构建了公路早期病害识别系统,该系统充分利用GPR信号的频带介电谱特征,分别抽取训练样本和测试样本的频带介电谱特征向量,以识别公路早期病害情况。实验结果表明：频带介电谱特征向量能较好的表征GPR信号对应的地电信息,适用于GPR信号的分类；该系统能对公路基层的密实度特征和含水量特征进行有效识别,为公路早期病害检测提供了技术保障；此外,该系统具有很强的实时性,对各个测试样本进行特征向量提取和识别的平均耗时不大于0.8ms。

Dielectric Spectrum Feature Vector of Ground Penetrating Radar Signal in Frequency Bands for Detecting Expressway Subgrade Diseases

All of the existing ground penetrating radar (GPR) technologies was only used to locate expressway subgrade diseases, but not carry out disease forecast of embankments. In order to predict the diseases, a diseases recognition system was formulated in the paper, which made the fullest use of the dielectric spectrum of GPR signal in different frequency bands. The dielectric spectrum feature vectors of the training and testing samples are separately extracted to identify the road early diseases. The experimental results show that the dielectric spectrum feature vector can be applied on target classification of GPR signal. The proposed system can identify the compaction degree and water content of embankments and provide guarantees for disease forecast of embankments. Besides, it has characteristics of excellent real-time ability. In the process of the extraction of feature vector and pattern recognition, the average elapsed time is less than 0.8 milliseconds.