The received signals in ground penetrating radar (GPR) techniques contain very valuable information on the size and orientation of the reflectors as well as the properties of the propagation path of the electromagneti...
详细信息
The received signals in ground penetrating radar (GPR) techniques contain very valuable information on the size and orientation of the reflectors as well as the properties of the propagation path of the electromagnetic wave. Sometimes, these signals contain many overlapping reflected echoes due to the complexity of the studied geometry. Accurate estimation of these echoes is essential for the non-destructive evaluation of reinforced concrete structures. In this study, the matching pursuit (MP) algorithm is used to decompose the GPR signal into multiple echoes. The selection of the appropriate atoms is done by teachinglearningbasedoptimization (TLBO) algorithms which, at the same time, change five parameters (bandwidth, arrival time, center frequency, amplitude and phase), until a good representation of the original GPR signal is achieved. The effectiveness of the proposed method is tested using both synthetic signals generated from a gprMax simulation program, and a real data collected on a concrete slab specimen at different offset.
暂无评论