No abstract

No abstract

No abstract

A borehole radar (BHR) prototype system was developed for the exploration of complicated oil and gas reservoirs. To verify the performance of the system, single-hole reflection imaging experiments were carried out in an abandoned limestone mine. In the physical experiments, the cliff wall and a metal plate were selected as the imaging targets to evaluate the detection performance of the prototype system. The average filter method was used to remove the background noise, then the frequency–wave number (F-K) imaging algorithm was adopted for radar imaging. The unknown fractures surrounding the borehole produced complex reflections that were not beneficial to effectively extract the target echo when the down-hole sensor was shifted along the borehole. However, by fixing the down-hole sensor and shifting the target, the detection range of the radar system extended up to about 10 m in the limestone formation. A 2-D finite-difference time-domain (FDTD) modeling method was also implemented to simulate the experimental procedure, and demonstrated that the prototype system can provide enough accuracy to predict the echo signal characteristics and reproduce the radar response in the formation. The combination of field experiment, theoretical analysis, and numerical simulation not only objectively validated the fundamental performance of the radar prototype, but also generated some new concepts for further improvement on the radar system design.

Electromagnetic (EM) inversion is a quantitative imaging technique that can describe the dielectric constant distribution of a target based on the EM signals scattered from it. In this paper, a novel deep neural network (DNN) based methodology for ground penetrating radar (GPR) data inversion, known as the Ü-net is introduced. The proposed Ü-net consists of three parts: a data compression unit, U-net, and an output unit. The novel inversion approach, based on supervised learning, uses a neural network to generate the dielectric constant distribution from GPR data. The GPR data can be compressed and reshaped the size using data compression unit. The U-net maps the object features to the dielectric constant distribution. The output unit meshes the dielectric constant distribution more finely. A novel feature of the proposed methodology is the application of instance normalization (IN) to the DNN EM inversion method and a comparison of its performance to batch normalization (BN). The validity of this technique is confirmed by numerical simulations. The Mean-Square Error of the test data sets is 0.087. These simulations prove that the instance normalization is suitable for GPR data inversion. The proposed approach is promising for achieving quality dielectric constant images in real-time.