Ground Pentrating Radar Target Recogniation Methods And Its Applications In The Tunnel Lining Detection

With the continuous improvement and development of China's trunk highway network,the scale and quantity of the highway tunnel are growing as well.In order to avoid the accidents affect the safety of people's lives and property during the tunnel excavation process and the operational stage after the construction of tunnels,it is of great significance to adopt proper geophysical tools to carry out quality detection of tunnel linings.As a common tool with advantages of quick,high efficiency,high resolution and nondestructive,the ground penetrating radar is favored by the tunnel workers.Three typical diseases occured in the lining detection--the separation of liner,the non-dense and the waterlogged cavity--can be classified in holes of different shapes and sizes which filled with different materials.In this paper,through the test data obtained by using the ground penetrating radar to detect physical models,research and establish classification and recognition method of cavities with different filling materials,and apply it to the classification and identification of the disease in tunnel lining detection.The main research contents are as follow:(1)Through physical model tests,the corresponding initial radar data which were filled with dry sand,gravel,mud,water and other materials respectively were obtained,and use the data pretreatment process summarized in this paper of ground penetrating radar to deal with it.Then further summarize and analyze three typical diseases radar images:the separation of liner,the non-dense and the cavity filled with water in the detection of tunnel linings.(2)In this paper,the extraction technique of amplitude attribute,frequency spectrum and coherence in ground penetrating radar is introduced.Combining with the physical meaning of radar attributes,five kinds of radar attributes--the root mean square amplitude,the average peak amplitude,the time domain average energy,the similar coefficient and the average phase value of the-3dB bandwidth--are selected as the characteristic parameter of the ground penetrating radar target classification and recognition.(3)By establishing the Gauss Process for Binary Classification(GPC)model of different cavity fillers.The five radar attributes are taken as the vector input with the root mean square amplitude,average peak amplitude,time domain average energy,similarity coefficient and-3dB bandwidth average phase,which successfully identify the air,debris,mud and water filled in each cavity of the physical model,and put out the corresponding prediction probability.(4)Combined with the practical lining detection engineering of Cenxi tunnel,the GPC model verified by physical model is succed in identifying the classification of three typical deseases of lining:the separation of liner,the non-dense entity and the cavity filled with water,which proves the feasibility of the Gauss prediction model of tunnel lining detection of the ground penetrating radar,and has a splended application prospect.