<<
[an error occurred while processing this directive] | [an error occurred while processing this directive] >>
Machine Recognition Method of Tunnel Lining Voids Based on SVM Algorithm
（1 School of Civil Engineering，Dalian University of Technology，Dalian 116024; 2 School of Hydraulic Engineering，Dalian University of Technology，Dalian 116024; 3 Key Laboratory of Geotechnical and Underground Engineering，Tongji University，Shanghai 200092;4 Department of Geotechnical Engineering，Tongji University，Shanghai 200092）
Abstract Ground penetrating radar (GPR) is one of the most effective detection methods for tunnel lining voids. However, the difficulties in data explanation are always the key to restrict its wide application. Based on support vector machine (SVM) algorithm, a set of machine recognition method of GPR image for tunnel lining voids is established. This method includes pre-processing of GPR data, feature extraction and SVM recognition. Firstly, the GPR image needs to be preprocessed by time-zero correction, filtering, migration and gain and so on to improve the signal-noise ratio (SNR). Secondly, each time-domain trace of GPR image is segmented and three statistics, namely variance, mean absolute deviation and fourth-order moment are extracted from the segmented signal as image features. Finally, the SVM model is trained by using the known data, and the data from a numerical simulation and a model experiment are used to test the trained SVM model. The results show that the proposed method can not only accurately recognize all voids in the tunnel lining and surrounding rock, but also accurately estimate the cover depths and lateral ranges of the voids.
Abstract：
Ground penetrating radar (GPR) is one of the most effective detection methods for tunnel lining voids. However, the difficulties in data explanation are always the key to restrict its wide application. Based on support vector machine (SVM) algorithm, a set of machine recognition method of GPR image for tunnel lining voids is established. This method includes pre-processing of GPR data, feature extraction and SVM recognition. Firstly, the GPR image needs to be preprocessed by time-zero correction, filtering, migration and gain and so on to improve the signal-noise ratio (SNR). Secondly, each time-domain trace of GPR image is segmented and three statistics, namely variance, mean absolute deviation and fourth-order moment are extracted from the segmented signal as image features. Finally, the SVM model is trained by using the known data, and the data from a numerical simulation and a model experiment are used to test the trained SVM model. The results show that the proposed method can not only accurately recognize all voids in the tunnel lining and surrounding rock, but also accurately estimate the cover depths and lateral ranges of the voids.
QIN Hui1 TANG Yu2 XIE Xiongyao3,
4 WANG Zhengzheng1
.Machine Recognition Method of Tunnel Lining Voids Based on SVM Algorithm[J] MODERN TUNNELLING TECHNOLOGY, 2020,V57(2): 13-19
2020, Vol. 57 