Research on GPR Identification of the Voids behind Tunnel Lining Based on SVM (宇宙・航行エレクトロニクス)

概要

論文の詳細を見る
At present, ground penetrating radar (GPR) as a non-destructive detection method is widely applied in tunnels' exploration. Nevertheless, it has been facing many difficulties on the interpretation of radargram images. In this paper, we focus on the identification of forward simulation images for the detection of voids behind tunnel lining by GPR based on Support Vector Machine (SVM). First of all, we obtain training data, testing data and predicting data for the SVM classifier model by applying Finite Difference Time Domain (FDTD) method. Then, we divide the A-SCAN data into different segments; calculate the variance, standard absolute deviation, fourth moment of each segment as statistical features, which can be collected as the SVM model input samples, train and test the SVM model with training samples and testing samples. Finally, the trained SVM model can be used to predict the estimating samples in order to identify the voids behind tunnel lining. From the results, the GPR image of the voids behind tunnel lining can be automatically identified by SVM. However, the recognition of the shape and distribution of voids may need certain improvement.
2011-11-17