Research On Azimuth Recognition And 3D Imaging Algorithm Of Directional Borehole Radar

Ground Penetrating Radar(GPR)is a widely used geophysical method.GPR can detect the target nondestructive,but the depth of the detection is limited.The borehole radar can be close to the target,and in the depth of detection,the borehole radar has a great increase.But the traditional borehole radar cannot recognize the azimuth angle of the target in a single borehole,and using several boreholes will greatly increase the cost.Directional borehole radar can be used to accurately measure the orientation of the target in a single hole.Directional borehole radar consists of two forms including directional transmission and directional reception.This paper studies directional reception.The directional borehole radar in this paper is composed of a transmitting antenna and four omni-directional receiving antennas which form a uniform circular array.Four receiving antennas are designated as the east,south,west,north antennas,with clockwise rotation.When the north antenna and geographical north are in the same direction,other antennas point corresponding direction.Azimuth recognition is performed on the basis of the difference in the phase of the received signal from the four receiving antennas.Traditional azimuth recognition algorithms include the Inverse Tangent Method and the Residual Method.The Inverse Tangent Method is to synthesize the magnetic field data from the received electric field data,and use the tangential relation between the east-west direction component of the magnetic field data and the north-south direction component to recognize the direction.The Residual Method assumes that the signal is incident at different azimuth angles,calculates the received signal of the receiving antenna under the corresponding azimuth.The residual is the absolute value obtained by the difference between the calculated value and the measured value.The azimuthal angle corresponds to the minimum value of the residual.The spatial spectrum method is a method of estimating the direction of arrival by calculating the energy of various azimuths in the space.Compared to the Inverse Tangent Method and the Residual Method,the spatial spectrum method has a stronger ability to resist random noise.The Multiple Signal Classification Method(MUSIC Method)is a traditional spatial spectrum method.The MUSIC algorithm is not only suitable for uniform linear array,but also suitable for uniform circular array.The MUSIC algorithm needs to obtain the covariance matrix of matrix composited by the received signal,and then seek the eigenvalues and eigenvectors of the covariance matrix.The eigenvectors corresponding to the large eigenvalues constitute the subspace of the signal.Noise subspace consists of eigenvectors corresponding to small eigenvalues.Using the orthogonal relation between the signal subspace and the noise subspace,we can construct the spatial spectrum of the MUSIC algorithm.In this paper,a Multi Signal Classification Algorithm based on Beam Space(BS-MUSIC Method)is proposed and implemented on the basis of MUSIC algorithm.The BS-MUSIC Method combines the received signals of four receiving antennas to two beams,and then the two beams are used as channels to recognize the azimuth angle with the MUSIC Method.Therefore,the amount of computation is reduced from the processing of the original four dimensional arrays to the processing of a two-dimensional array,thus reducing the amount of computation and saving the operation time.Linear Prediction Algorithm is also a spatial spectrum estimation algorithm.In this paper,the linear prediction algorithm is improved,and a linear prediction algorithm based on directional drilling radar system is proposed and realized.The linear prediction algorithm is to obtain the weight coefficient by predicting the future or previous data.We use the weight coefficient to construct the linear prediction space spectrum,and then get the azimuth angle.The method of predicting the weight coefficient by predicting the future data is called the forward prediction algorithm.The method of predicting the weight coefficient by predicting the past data is called the backward prediction algorithm.The method of predicting the weight coefficient at the same time for the future and before the data is called the bidirectional prediction algorithm.Linear prediction algorithm is only suitable for uniform linear array.In this paper,four receiving antennas of a directional borehole radar system are divided into a uniform linear array of East-West and a uniform linear array of North-South.In order to improve the accuracy of operation,the virtual antenna is added to the two uniform linear arrays,so that the East-West uniform linear array and the North-South uniform linear array have four receiving antennas respectively.We apply the linear prediction algorithm to the two linear arrays and integrate the azimuth recognition results of the two linear arrays.We finally get the azimuth.The interaction between the borehole,the probe,and the receiving antenna causes serious interference to the azimuth recognition.In this paper,the FDTD algorithm with subgrades is used to simulate the above situation.When there is only one transmitting antenna and one receiving antenna in the well hole,the greater the dielectric constant of the fluid is,the stronger the direct wave signal is,the more the reflected signal is,and the more serious the waveform distortion is.When four receiving antennas are located in the four directions in the East,South,West and North,the larger the array radius of the receiving antenna is,the better the division of the received signal is.The larger the relative dielectric constant of the fluid is,the stronger the resonance of the received signal is.By numerical simulation,resonance is formed by the oscillations of reflected waves back and forth between the receiving antennas.The greater the diameter of the probe is,the weaker the direct wave signal is.The greater the diameter of the probe is,the little difference in the intensity of the reflected signal is,when the borehole and the pipe are considered simultaneously.The greater the relative permittivity of the fluid in the well is,the intensity of both direct and reflected signals increases.The greater the relative dielectric constant of the probe is,the more obvious the resonance phenomenon in the received signal is.Due to the existence of borehole and probe tube,the electromagnetic wave will be reflected and refracted,which will affect the phase information of the received signal,and cause serious interference to the azimuth recognition.The azimuth angle obtained by the azimuth identification algorithm under the uniform medium will be called as the apparent azimuth angle.In this paper,we get the relationship between the apparent azimuth angle and the true azimuth by finding the propagation path of the electromagnetic waves that are transmitted to each receiving antenna,and call this relation the correction relation of the non-uniformity around the borehole.We verify the correction relationship by GprMax.The results show that the correction relationship can effectively eliminate the influence of borehole and probe.Based the correction relation of the non-uniformity around the borehole,we improve the azimuth recognition algorithm,and propose and implement the algorithm of synthesizing vertical and horizontal slices based on azimuth.The migration algorithm is used to offset the vertical slices of all directions,and the 3D images of the offset data are processed.In this paper,the FDTD algorithm with subgrades is used to synthesize data,and the main situation in production practice is simulated.In this paper,three models are simulated,which includes isolated targets,well penetrating targets,and the combination of isolated and well penetrating targets.After calculation,the azimuth recognition algorithm and the imaging algorithm get the effective verification of the above model,so the algorithm proposed in this paper has universal applicability.Finally,the experiment is designed to test the azimuth recognition algorithm.Experiments of design include variable distance experiment and azimuth experiment.The experimental data are calculated by the azimuth recognition algorithm,and all the results are in line with the target model.Therefore,the azimuth recognition algorithm proposed in this paper has some practical application value.