A borehole radar directional detection method and device based on a circular multi-receiving antenna array

Abstract

The application discloses a borehole radar directional detection method and device based on a ring-shaped multi-receiving antenna array. The method comprises the following steps: acquiring the azimuth angles of each receiving antenna in the ring-shaped multi-receiving antenna array and a plurality of radar image maps corresponding to the azimuth angles one by one; determining a preset medium wave velocity and inversely calculating an optimal wave velocity based on a joint target function, that is, after offset imaging of each candidate wave velocity and adaptive binary processing based on local energy entropy, calculating an image focusing index and an abnormal point azimuth standard deviation, and taking the minimum weighted sum of the two as the optimal wave velocity; after re-imaging based on the optimal wave velocity, calculating the target azimuth angle by weighting the azimuth angles of each receiving antenna based on a phase consistency factor. Through the cooperation of adaptive partition binary, joint target function velocity selection and phase consistency weighting, the application overcomes the problems that a fixed threshold value is difficult to consider both strong background and weak anomaly, a single focusing index velocity selection has multiple solutions, and azimuth weighting is easily affected by multipath interference, and the precision of three-dimensional directional detection of an abnormal body is improved.

Description

Technical Field

[0001] This invention belongs to the field of borehole radar detection technology, specifically relating to a borehole radar directional detection method and device based on a circumferential multi-receiver antenna array. Background Technology

[0002] Borehole radar is a geophysical exploration method that places a transceiver antenna inside a borehole to transmit high-frequency electromagnetic waves into the surrounding strata and receive reflected waves, thereby detecting geological anomalies within a certain range around the borehole. Compared with ground-penetrating radar, borehole radar antennas can penetrate deep into the strata and get close to the target, overcoming the shortcomings of ground-penetrating radar such as shallow detection depth and susceptibility to surface weathering. It has advantages such as a large detection range, low cost, and high efficiency in the exploration of adverse geological bodies such as karst, faults, fracture zones, and water-bearing structures, and has broad application prospects.

[0003] In practical engineering, the detection of geological anomalies around boreholes often requires not only determining their burial depth and distance from the borehole, but also their azimuth relative to the borehole. Traditional borehole radar typically uses omnidirectional antennas, which can only determine the depth of the anomaly and its distance from the borehole, but cannot determine its azimuth, thus failing to meet the requirements for three-dimensional directional anomaly detection. To obtain azimuth information, a detection method has been developed that uses a single-transmitter, multi-receiver directional antenna and performs offset imaging on the data from each receiving antenna before azimuth identification, making it possible to determine the azimuth of anomalies. However, in the process of using the above detection method to achieve directional detection, existing technologies still have the following problems in the key stages of data processing.

[0004] Firstly, when binarizing offset images to extract anomalous morphology, existing techniques typically employ a fixed threshold method. This involves using a threshold uniformly applied to the entire image (e.g., a global threshold obtained by multiplying the average energy of the entire offset image by a preset coefficient) to differentiate point-by-point. Since the energy distribution between the background and anomalous regions in an offset image differs significantly, and the energy of strongly reflective and weakly reflective parts of the same anomalous object also varies considerably, using a fixed threshold across the entire image is insufficient. Figure 1 When using a fixed threshold, a high threshold will filter out weaker real anomalies, resulting in their loss. Conversely, a low threshold will misclassify background clutter as anomalies, creating strong background interference. A fixed threshold struggles to simultaneously address the conflicting needs of suppressing strong backgrounds and preserving weak anomalies, leading to inaccurate anomaly morphology in the binarized data and impacting the reliability of subsequent processing.

[0005] Secondly, offset imaging relies on a pre-set electromagnetic wave propagation speed. The accuracy of the preset wave velocity directly determines the imaging quality and the estimation accuracy of the spatial location of anomalies. Current technologies typically use only a single image focusing index as the criterion for retrieving the optimal wave velocity. This involves trying different speeds within a certain range and selecting the wave velocity that provides the best image focusing. However, in complex underground media, such as those containing water, the relationship between the electromagnetic parameters of the medium and the wave velocity is no longer singular. Multiple different wave velocities may correspond to similar focusing effects, causing the curve of the focusing index changing with wave velocity to flatten near the optimal value, or even exhibiting multiple similar extreme values. This results in multiple solutions for wave velocity selection, making it difficult to uniquely and accurately determine the true wave velocity. Furthermore, deviations in wave velocity selection can further lead to image defocusing and inaccurate anomaly localization.

[0006] Third, when calculating the azimuth angle of an anomaly using weighted data from multiple receiving antennas, existing technologies typically employ a simple arithmetic mean of the azimuth angles of each antenna, or a weighted average with the reflected wave amplitude (energy) as the weight. The weights in these methods depend solely on the signal strength, without considering the signal's authenticity or reliability. In actual detection, interference signals from non-target directions, such as side reflections and multiple reflections, often have amplitudes that are not weak or even strong. Weighting by amplitude or energy would assign them higher weights, significantly deviating from the calculated azimuth angle and reducing the accuracy of directional detection.

[0007] In summary, existing technologies for directional detection of geological anomalies based on borehole radar suffer from several limitations. These include the difficulty in achieving both strong background suppression and weak anomaly preservation through binary fixed thresholding, the ambiguity of single focusing index velocity selection in complex aquifers, and the azimuth weighting relying solely on signal strength, making it susceptible to multipath interference. These issues limit the accuracy of anomaly morphology extraction, wave velocity inversion, and azimuth determination, hindering the achievement of high-precision three-dimensional directional detection. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the aforementioned background technology and provide a borehole radar directional detection method and device based on a circumferential multi-receiver antenna array, so as to achieve high-precision directional detection of the depth, distance and azimuth of anomalies.

[0009] The technical solution adopted in this invention is: a borehole radar directional detection method based on a circumferential multi-receiver antenna array, comprising the following steps: The azimuth angles of each receiving antenna in the circumferential multi-receiver antenna array are obtained, as well as multiple radar image maps generated from the radar electromagnetic wave data of each receiving antenna, each corresponding to a specific receiving antenna. A preset medium wave velocity is determined, and the optimal wave velocity is inverted based on a joint objective function: Multiple candidate wave velocities are acquired within a velocity scanning interval determined by the preset medium wave velocity; for each candidate wave velocity, offset imaging processing is performed on multiple radar images to obtain multiple offset images corresponding to each receiving antenna; adaptive binarization processing based on local energy entropy is performed on each offset image to obtain multiple candidate binarized images; the image focus index and the standard deviation of the azimuth angle of outliers are calculated based on the multiple candidate binarized images corresponding to each candidate wave velocity; an objective function is constructed including the product of a first weighting coefficient and the image focus index, and the product of a second weighting coefficient and the standard deviation of the azimuth angle of outliers; the candidate wave velocity that minimizes the objective function is selected as the optimal wave velocity, and the offset imaging processing is performed on multiple radar images using the optimal wave velocity to obtain multiple final offset images corresponding to each receiving antenna; The adaptive binarization process based on local energy entropy is as follows: The local energy entropy of each pixel within a preset local window and the global average entropy of the entire offset image are calculated respectively; when the local energy entropy is less than or equal to the global average entropy, the corresponding pixel is determined to be located in the background area, and the local grayscale mean within the local window is used as the binarization threshold for that pixel; when the local energy entropy is greater than the global average entropy, the corresponding pixel is determined to be located in a suspected abnormal area, and the product of the global grayscale mean of the entire offset image and a magnification factor not less than one is used as the binarization threshold for that pixel; the offset image is binarized point by point according to the determined binarization threshold to obtain the corresponding binarized image; The image focusing index is an index characterizing the degree of dispersion of each suspected anomaly point relative to its spatial centroid. The smaller the image focusing index, the more focused the anomalous energy is in space. The standard deviation of the azimuth angle of the anomaly point is as follows: the pixels located in the suspected anomaly area and whose binarization result is one in the candidate binarized image corresponding to each receiving antenna are identified as suspected anomaly points. The azimuth angle of the receiving antenna to which each suspected anomaly point belongs is taken as the azimuth angle of the suspected anomaly point. The standard deviation of the azimuth angle of all suspected anomaly points relative to the mean of the azimuth angles of all suspected anomaly points is the standard deviation of the azimuth angle of the anomaly point. The target azimuth angle is calculated based on a phase consistency factor weighted average: Adaptive binarization is performed on the final offset images corresponding to each receiving antenna. The suspected anomalous areas obtained from the adaptive binarization are identified as target anomalous bodies, and the depth corresponding to these anomalous bodies is determined as their depth position. At the depth position of the final offset images corresponding to each receiving antenna, the maximum reflected wave amplitude and corresponding phase are extracted, and the average phase value of all corresponding phases of the receiving antennas is calculated. The product of the maximum reflected wave amplitude and the phase consistency factor of each receiving antenna is used as the weighting coefficient for each receiving antenna. The azimuth angle of each receiving antenna is then weighted and averaged using the weighting coefficient to obtain the target azimuth angle of the target anomalous body. The phase consistency factor is an exponential function with a base of the natural constant and an exponent of the negative of the phase deviation, where the phase deviation is the absolute value of the difference between the corresponding phase and the average phase value.

[0010] In the above technical solution, the calculation of the local energy entropy of each pixel within a preset local window includes: calculating the pixel proportion corresponding to each gray value within the local window, accumulating the product of each pixel proportion with its corresponding base-2 logarithm, and taking the negative of the accumulation result as the local energy entropy.

[0011] In the above technical solution, the magnification factor of not less than 1 ranges from 1.0 to 1.5; the size of the preset local window is 5 by 5 pixels.

[0012] In the above technical solution, the image focus index is calculated as follows: the spatial centroid of the lateral distance coordinates and longitudinal depth coordinates of all suspected anomalies is determined, and the sum of the squares of the differences between the lateral distance coordinates and longitudinal depth coordinates of each suspected anomaly point and the corresponding coordinates of the spatial centroid is averaged over all suspected anomalies. The average value obtained is the image focus index. The standard deviation of the azimuth angle of the anomaly point is the arithmetic square root of the square mean of the difference between the azimuth angle of the receiving antenna corresponding to all suspected anomalies and the mean of the azimuth angles of all suspected anomalies.

[0013] In the above technical solution, the sum of the first weighting coefficient and the second weighting coefficient is equal to 1; within the velocity scanning interval, multiple discrete velocity values ​​are extracted as each candidate wave velocity according to a preset velocity scanning interval.

[0014] In the above technical solution, the velocity scanning range is a velocity range determined by extending a preset range upwards and downwards to both sides based on the preset medium wave velocity.

[0015] In the above technical solution, determining the preset medium wave velocity includes: selecting a region with hyperbolic features in the radar image, extracting the time and spatial coordinates of the hyperbola vertex and at least two reflection points in the region, establishing a wave velocity solution equation based on the geometric path relationship, deriving the preliminary electromagnetic wave propagation speed and using it as the preset medium wave velocity.

[0016] In the above technical solution, the offset imaging processing is an offset imaging processing based on the frequency-wavenumber method, including: performing a two-dimensional Fourier transform on the radar image in the lateral distance and time directions to obtain wavenumber-frequency domain data; performing wavefield extension on the wavenumber-frequency domain data along the longitudinal depth direction; performing a two-dimensional inverse Fourier transform on the extended data, and taking the imaging time as zero through the mapping relationship between frequency and wavenumber based on the medium wave velocity used in the offset imaging processing, to obtain the offset image.

[0017] In the above technical solution, the azimuth angle of each receiving antenna and the multiple radar images corresponding to each receiving antenna are obtained in the following way: the initial azimuth angle and radar electromagnetic wave data of each receiving antenna in the one-transmit-multiple-receive antenna array are obtained, the initial azimuth angle corresponding to the electromagnetic wave data of each receiving antenna is corrected by combining the probe moving distance and rotation angle, and multiple radar images corresponding to each receiving antenna are generated based on the corrected azimuth angle and the radar electromagnetic wave data.

[0018] This invention provides a borehole radar directional detection device based on a circumferential multi-receiver antenna array, including a borehole radar probe. The probe has a transmitting antenna and a circumferential multi-receiver antenna array on the same cross section. The circumferential multi-receiver antenna array is composed of at least three receiving antenna elements arranged at equal angles along the circumference of the probe. The main lobe direction of each receiving antenna element is approximately perpendicular to the probe axis and points radially outward. The probe also has an independent receiving channel corresponding to each receiving antenna element, used for synchronous or time-division sampling of radar wave fields excited by the same transmitting source. The system includes a processor and a memory, wherein the memory stores a computer program that, when executed by the processor, implements the steps of the method described in the above technical solution.

[0019] The beneficial effects of this invention are: In terms of the overall method of this invention, based on acquiring the azimuth angles of each receiving antenna in a multi-transmitter system and their corresponding radar image maps, the depth, range, and azimuth of an anomalous body are determined sequentially through optimal wave velocity inversion based on a joint objective function, adaptive binarization based on local energy entropy, and azimuth angle calculation based on phase consistency factor weighting. Compared to existing technologies that can only determine the depth and range of anomalies, or those that can determine direction but have limited accuracy in wave velocity inversion, morphology extraction, and azimuth calculation, this invention organically combines the improvements of the above three stages into a single detection process. This enables robust and accurate three-dimensional directional detection of anomalies under complex geological conditions, significantly improving the overall reliability and accuracy of directional detection.

[0020] In adaptive binarization based on local energy entropy, this invention dynamically divides an image into background and suspected anomalous regions based on the relationship between the local energy entropy of a pixel's neighborhood and the global average entropy of the entire image. It then applies opposite threshold rules to the two types of regions: the background region uses the local gray-level mean that fluctuates with the neighborhood as the threshold, while the suspected anomalous region uses the global gray-level mean multiplied by an amplification factor as the threshold. This invention overcomes the dilemma of existing fixed-threshold schemes where suppressing strong backgrounds results in the loss of weak anomalous regions, while retaining weak anomalous regions results in the inclusion of strong backgrounds. It achieves adaptive differentiation between background and anomalous regions. Pixels with local energy entropy not exceeding the global average entropy are classified as background and excluded from subsequent speed selection and orientation criteria for suspected anomalous points, thus suppressing the interference of cluttered background reflections on subsequent processing. Pixels with local energy entropy higher than the global average entropy are classified as suspected anomalous regions, with a threshold higher than the global benchmark highlighting strong energy reflections while retaining relatively weak true anomalous signals. This improves the recognition rate of weak anomalous regions and reduces the false positive rate, providing a more reliable and concise anomalous body morphology for subsequent wave velocity inversion and orientation calculation.

[0021] In the optimal wave velocity inversion based on a joint objective function, this invention, in addition to the image focusing index, further introduces the standard deviation of the azimuth angle of outliers as a second criterion to characterize azimuth consistency, and uses the minimum weighted sum of the two as the criterion for selecting the optimal wave velocity. Since spatial focusing and azimuth consistency are two independent dimensions stemming from whether the wave velocity is correct, when the focusing index tends to flatten near the optimal value due to the complexity of the medium and water content, and is insufficient to distinguish between true and false signals, the azimuth consistency criterion can still provide effective discrimination. The combined minimum values ​​of the two are mutually locked. Therefore, this invention overcomes the problem of wave velocity ambiguity in complex water-bearing media where existing technologies rely solely on a single focusing index for velocity selection, making the determination of the optimal wave velocity more unique and robust, significantly improving the physical rationality of wave velocity estimation and the accuracy of outlier location, and effectively suppressing false and missed anomalies caused by incorrect wave velocity settings.

[0022] In azimuth calculation based on phase coherence factor weighting, this invention uses the product of the maximum reflected wave amplitude of each receiving antenna and the phase coherence factor as the weighting coefficient. This ensures that the weight depends simultaneously on the signal strength and the consistency of the signal phase with the mainstream phase of each antenna. Therefore, for interference signals from non-target directions such as side reflections and multiple reflections, which have strong amplitudes but significantly deviated phases, this invention can automatically assign them lower weights and suppress them. Compared to existing schemes that rely solely on arithmetic averaging or amplitude weighting based on signal strength and are easily biased by strong interference, this invention effectively suppresses the influence of multipath interference on azimuth calculation, contributing to the accuracy of azimuth calculation for anomalous objects and the precision of directional detection.

[0023] Based on the above scheme, this invention limits the local energy entropy to the negative of the sum of the products of each gray level percentage and its base-2 logarithm, providing a clear and feasible calculation method for local energy entropy; it sets the magnification factor to a value not less than one and not greater than 1.5, and the local window size to 5 x 5 pixels, which can effectively raise the judgment threshold in suspected abnormal areas to highlight strong reflections without suppressing weak anomalies due to excessive magnification, while achieving a balance between statistical stability and spatial resolution in local entropy estimation; it limits the image focusing index to the mean of the sum of squares of the differences between the coordinates of each abnormal point and its spatial centroid, enabling it to sensitively characterize the spatial convergence of abnormal energy, and it corresponds mathematically with the azimuth standard deviation to form a complete system. A unified measurement of the spatial and azimuth distribution of anomaly points is implemented; the sum of two weighting coefficients is limited to one, and candidate wave velocities are extracted at preset intervals; the velocity scanning interval is limited to the interval extended to both sides based on the preset wave velocity, enabling efficient joint optimization within a reasonable search range; the initial wave velocity is inverted based on geometric relationships using the spatiotemporal coordinates of the hyperbola vertex and reflection point, providing reasonable initial values ​​that fit the measured medium for subsequent migration and accurate velocity measurement; the migration imaging is limited to the frequency-wavenumber method, efficiently completing wavefield extension and imaging in the frequency-wavenumber domain, ensuring accurate convergence of diffraction energy; the azimuth angle is corrected using the initial azimuth angle, moving distance, and rotation angle, eliminating azimuth deviation caused by probe rotation and ensuring the accuracy and reliability of the azimuth angle used for orientation calculation. The above preferred implementation further improves the feasibility, robustness, and accuracy of the scheme based on the aforementioned approach.

[0024] As for the detection device of the present invention, the device can realize each step of the above-mentioned directional detection method by executing the computer program stored in the memory through the processor. Therefore, it also has all the beneficial effects corresponding to the aforementioned method, and has a simple structure and is easy to integrate and deploy. The detection device of this invention arranges at least three receiving antenna units at equal angles along the circumferential direction on the same cross-section of the borehole radar probe, forming a 360° azimuth-covering circumferential spatial sampling receiving array. This allows the same reflection / scattering source to have distinguishable incident angles and path lengths in different receiving channels. A single-transmitter, multi-receiver configuration with a single transmitting antenna and multiple circumferential receiving antennas in a common aperture configuration is used to achieve multi-azimuth synchronous or time-division sampling without increasing the borehole diameter or probe outer diameter. Each receiving antenna is connected to an independent RF front-end and sampling channel, ensuring amplitude and phase consistency between channels. This allows the same reflection / scattering source to be received by multiple receiving antennas simultaneously, providing a consistent hardware foundation for the aforementioned azimuth weighting based on the amplitude and phase consistency factors of the reflected waves from each receiving antenna. Furthermore, the transmitting antenna, the circumferential receiving antenna array, and related RF modules are integrated into the same rigid or semi-rigid probe, meeting the entry and pushing requirements of engineering boreholes. Therefore, this invention introduces an azimuth dimension on top of the traditional two dimensions of borehole depth and radial distance, achieving circumferential azimuth resolution.

[0025] It is particularly important to note that the adaptive binarization, joint objective function velocity selection, and phase consistency weighting of this invention are not simply isolated superpositions, but rather functionally interdependent and mutually conditional: the reliable anomaly morphology extracted by adaptive binarization forms the basis for the joint objective function to calculate the focusing index and azimuth standard deviation; the accurate wave velocity obtained by the joint objective function inversion is a prerequisite for phase consistency weighting to extract the true amplitude and phase from a fully focused image. These three elements progressively enhance each other in image domain purification, velocity domain calibration, and angular domain orientation, jointly serving the high-precision three-dimensional orientation detection of anomalies in complex water-bearing, multi-path media. Therefore, the overall effect achieved exceeds the simple sum of the effects that each element could achieve individually, forming a synergistic organic whole. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the borehole radar directional detection method based on a circumferential multi-receiver antenna array in an embodiment of the present invention. Figure 2 This is a schematic diagram illustrating the principle of determining the initial medium wave velocity based on hyperbolic features in an embodiment of the present invention. Figure 3 This is a schematic diagram of the borehole radar directional detection device based on a circumferential multi-receiver antenna array in an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the principle of inverting the optimal wave velocity based on the joint objective function in an embodiment of the present invention; Figure 5 This is a schematic diagram of the cross-sectional structure of the circumferential receiving antenna array in an embodiment of the present invention.

[0027] Among them, 1-probe tube, 2-transmitting antenna, 3-1, 3-2, 3-3, 3-4-receiving antenna units, 4-main lobe direction, 5-hole wall, 6-shielding partition. Detailed Implementation

[0028] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments to facilitate a clear understanding of the present invention, but these descriptions do not constitute a limitation on the present invention.

[0029] Example 1 This embodiment provides a borehole radar directional detection method based on a circumferential multi-receiver antenna array. For ease of understanding, the directional radar data acquisition device upon which this method is based will be described first, followed by a detailed explanation of the entire data processing flow, i.e., steps S1 to S6. The overall flow of this method can be found in [link to relevant documentation]. Figure 1 .

[0030] The directional radar data acquisition device used in this embodiment can be a single-transmitter, multi-receiver directional drilling radar acquisition device, which mainly includes: a main control unit, a radar probe, a probe cable, a pulley counter, a pulley counter cable, and a tripod. During horizontal drilling tests, a push rod is used to guide the radar probe into the hole. The tripod is mounted at the hole opening, the pulley counter is connected to the main control unit via the pulley counter cable, and the radar probe is suspended inside the borehole via the probe cable and can move along the borehole axis.

[0031] The main controller can record the data from the pulley counter to determine the position of the radar probe in the vertical or horizontal direction, i.e., to determine the drilling distance; and control the start and stop of the transmitting antenna, and store the data from the electronic compass and each receiving antenna.

[0032] The radar probe is equipped with a single-transmitter, multiple-receiver antenna array, meaning that a transmitting antenna and multiple receiving antennas are arranged on the same cross-section of the probe, forming a circumferential multi-receiver antenna array. "Single-transmitter" refers to an omnidirectional transmitting antenna that emits electromagnetic waves in a 360° direction; "multiple-receiver" refers to multiple receiving antennas, typically three or four, arranged at equal angles along the circumference of the probe. Their main lobes are approximately perpendicular to the probe's axis and point radially outward, and each antenna synchronously or time-divisionally samples the electromagnetic wave signal excited by the same transmitting source through a corresponding independent receiving channel. Since each receiving antenna is a directional antenna, with its main lobe pointing to its respective circumferential azimuth, the same reflection / scattering source exhibits azimuth-related amplitude differences in different receiving antennas and their receiving channels, and can be received simultaneously by multiple receiving antennas. Therefore, the azimuth angle of each receiving antenna can be weighted based on the reflected wave amplitude and phase consistency factor to interpret the azimuth angle of the geological anomaly. The aforementioned single-transmitter, multiple-receiver (circumferential multi-receiver antenna array) structure can be implemented using corresponding antenna and receiving channel structures known in the art.

[0033] Furthermore, the radar probe also includes: a cap for waterproofing and dustproofing and protecting the probe's connectors; a fixed bearing for securing the probe's components; a metal casing for protecting the internal components; a non-metallic casing placed outside the transmitting and receiving antennas to avoid shielding electromagnetic signals; an electronic compass for recording the probe's azimuth; a main control board for controlling the transmitting antenna's operation and storing data from the electronic compass and receiving antennas; and a battery for powering the probe's components. In practical use, a circumferential three-, four-, or more-received receiving antenna array can be selected according to azimuth resolution requirements.

[0034] During the borehole radar test, as the radar probe moves, a pulley counter records the probe's movement information, an electronic compass records the probe's azimuth information, an omnidirectional transmitting antenna emits electromagnetic waves in a 360° direction, and multiple directional receiving antennas receive the electromagnetic wave signals in real time. Finally, the main control computer saves the probe's movement distance, azimuth angle, and received electromagnetic wave information for subsequent data processing. It should be noted that each receiving antenna corresponds to an azimuth angle, and each receiving antenna obtains a radar image during the test; therefore, the azimuth angle, receiving antenna, and radar image are all crucial data. Figure 3 Each item corresponds one-to-one and the quantity is exactly the same.

[0035] Based on the data acquired by the above acquisition device, this method sequentially executes steps S1 to S6, namely: step S1 distance and azimuth correction, step S2 preliminary calculation of medium wave velocity, step S3 radar data offset imaging, step S4 obtaining adaptive binarized image of offset image, step S5 accurate calculation of medium wave velocity, and step S6 obtaining abnormal azimuth angle.

[0036] Step S1: Perform range and azimuth correction on the radar image.

[0037] The input for this step is the raw data stored by the acquisition device, including pulley counter data, electronic compass azimuth data, electromagnetic wave signals received by each receiving antenna, and probe movement distance information.

[0038] The processing procedure is as follows: Based on the pulley counter data, the distance information of the radar image is corrected; based on the electronic compass data, the azimuth angles of multiple receiving antennas during the test are corrected. Since the borehole radar may rotate during the test, the azimuth angle corresponding to the acquired radar data may differ from the initial azimuth angle. Therefore, it is necessary to correct the actual azimuth angle corresponding to the radar data at each test position (i.e., at any borehole distance). Specifically, let's assume that at the start of data acquisition... The initial azimuth angles of the receiving antennas are The radar probe moves inside the hole. After a distance, a rotation occurs, and the angle of rotation is denoted as . Then the receiving antenna is in The corrected azimuth at the distance is This yields target radar data that perfectly corresponds to the initial azimuth angle at any borehole distance, with the initial azimuth angle as the reference, and the target radar data is then converted into radar imagery.

[0039] The output of this step is multiple radar image images corrected for range and azimuth, each corresponding to a receiving antenna and its azimuth. This step provides the basic data for all subsequent steps; the multiple azimuth angles output and the radar image image corresponding to each azimuth angle are the common inputs for steps S2 to S6.

[0040] In a preferred embodiment, the azimuth angles of each receiving antenna and the corresponding multiple radar images are obtained as follows: The initial azimuth angles and radar electromagnetic wave data of each receiving antenna in the multi-receiver array are acquired; the initial azimuth angles corresponding to the electromagnetic wave data of each receiving antenna are corrected by combining the probe's movement distance and rotation angle; and multiple radar images corresponding to each receiving antenna are generated based on the corrected azimuth angles and the radar electromagnetic wave data. Thus, by dynamically correcting the azimuth angles through movement distance and rotation angle, azimuth deviations caused by probe rotation can be eliminated, ensuring the accuracy and reliability of the azimuth angles used for subsequent orientation calculations.

[0041] Step S2: Preliminary calculation of medium wave velocity.

[0042] The input for this step is the radar image output from step S1. The processing procedure is as follows: Figure 2As shown, a region with obvious hyperbolic features is selected in the radar image, and the hyperbola vertices and two reflection points are manually added. The vertex coordinates are marked as follows: The coordinates of the two reflection points are denoted as follows: and Let the burial depth of the anomaly be... Considering that system delay and vacuum delay will cause an error between the pickup point time and the actual time, denoted as The speed of electromagnetic wave propagation The derivation process is as follows.

[0043] Abnormal body burial depth Equal to speed When traveling one way The product of, i.e.: (1) , With signal arrival point The movement paths of the ... (2) Substituting equation (1) into equation (2) and rearranging the terms, we get: (3) From the reflection point Similarly, we can conclude that: (4) The unknown can be eliminated by subtracting equation (4) from equation (3). After sorting, the speed of electromagnetic wave propagation is obtained: (5) In the formula, The speed of electromagnetic wave propagation. , , They are the first position, the second position, and the third position, respectively. , , These are the first time point, the second time point, and the third time point, respectively; where the first position refers to the position where the borehole radar first detects the vertex of the hyperbola feature region.

[0044] The output of this step is the preliminary electromagnetic wave propagation speed in the space where the anomaly is located. ,Should As the preset medium wave velocity for offset imaging in step S3; when precise localization of anomalies is required, step S5 will precisely optimize the wave velocity based on the velocity scan to achieve the optimal wave velocity. Alternative .

[0045] As a preferred embodiment, determining the preset medium wave velocity includes: selecting a region with hyperbolic characteristics in the radar image, extracting the time and spatial coordinates of the hyperbola vertices and at least two reflection points within the region, establishing a wave velocity equation based on geometric path relationships, deriving the preliminary electromagnetic wave propagation velocity, and using it as the preset medium wave velocity. This method directly utilizes the hyperbolic diffraction pattern of the radar image itself to invert the wave velocity. Compared to consulting empirical parameter tables, it can provide an initial wave velocity that more closely matches the measured medium, providing a reasonable search starting point for subsequent migration imaging and accurate velocity measurement.

[0046] Step S3: Radar data offset imaging.

[0047] The input for this step is the radar image output from step S1, and the preset electromagnetic wave propagation speed. The initial wave velocity in step S2 is used during the first imaging. In step S5, candidate wave velocities are selected during the velocity scan, and the optimal wave velocity determined in step S5 is selected during final imaging. .

[0048] The processing procedure is as follows: Taking horizontal drilling as an example (vertical drilling is similar, only the data processing needs to be adjusted accordingly). direction and (Direction interchange) Based on the frequency wavenumber method, borehole radar data migration imaging is carried out, and the main steps are as follows.

[0049] Borehole radar data exist direction and Perform a two-dimensional Fourier transform on the direction to obtain wavenumber-frequency domain data. : (6) right along By extending the wave field in the direction, we obtain : (7) right Performing a two-dimensional inverse Fourier transform, we obtain : (8) Will and Substitute into equation (8) and take the imaging time. The offset imaging results were obtained. : (9) In equations (6) to (9), For the offset imaging results, for The two-dimensional Fourier transform result, The horizontal distance coordinate is... For vertical depth coordinates, Sampling time, Angular frequency, The speed of electromagnetic wave propagation. and They are respectively direction and Wave number in direction.

[0050] The output of this step is an offset image. The obvious hyperbolic diffraction features in the radar image are focused and converged after offset imaging, and the offset image can intuitively show the location and approximate range of the anomaly. This offset image serves as the input for adaptive binarization in step S4; in the velocity scan of step S5, this step will be repeated for each candidate wave velocity.

[0051] In a preferred embodiment, the migration imaging processing is a frequency-wavenumber-based migration imaging processing method, comprising: performing a two-dimensional Fourier transform on the radar image in the lateral range and time directions to obtain wavenumber-frequency domain data; performing wavefield extension on the wavenumber-frequency domain data along the longitudinal depth direction; performing a two-dimensional inverse Fourier transform on the extended data, and setting the imaging time to zero based on the mapping relationship between frequency and wavenumber using the medium wave velocity adopted in migration imaging processing, to obtain the migration image. The frequency-wavenumber method completes wavefield extension and imaging in the frequency-wavenumber domain, has high computational efficiency, and can accurately converge the diffraction hyperbola to the diffraction point, laying the foundation for subsequent anomaly identification and localization.

[0052] Step S4: Obtain the adaptive binarized image of the offset image.

[0053] The input for this step is the offset image output from step S3. The processing involves performing adaptive binarization based on local energy entropy on the offset image.

[0054] The offset image obtained in step S3 is usually quite cluttered. Traditional methods use a fixed threshold for binarization, which is equivalent to binarizing the entire image. Figure 1 Knife-cutting: A high threshold suppresses strong background areas but also drowns out weaker real anomalies, resulting in their loss; a low threshold preserves weak anomalies but misinterprets cluttered reflections from strong background areas as anomalies, creating strong background interference. A fixed threshold makes it difficult to simultaneously satisfy the contradictory needs of "suppressing strong background" and "preserving weak anomalies."

[0055] Conventional offset image binarization schemes first calculate the average energy of the entire offset image, then multiply this global average energy by a preset threshold coefficient to obtain a global threshold that applies uniformly to the entire image, and then classify pixels point by point based on this threshold. The fundamental limitation of this type of global single-threshold scheme is that the threshold does not change with the local content of the image, forcing background areas and abnormal areas to share the same discrimination threshold, and failing to escape the aforementioned dilemma of raising the threshold to discard weak abnormalities and lowering the threshold to retain strong backgrounds. The essential difference between this step and the previous one is: first, it introduces the local energy entropy of the image, a quantity that characterizes the local texture complexity, rather than just using grayscale energy itself; second, it dynamically divides pixels into two categories, background areas and suspected abnormal areas, based on the relationship between the local energy entropy and the global average entropy, rather than classifying the entire image as a single threshold. Figure 1 Third, apply different and opposite threshold rules to the two types of regions: the background region uses a local grayscale mean that varies with the neighborhood, and the suspected abnormal region uses a global grayscale mean multiplied by an amplification factor, thus forming a dynamic dual threshold for the region, rather than a single global threshold.

[0056] This invention introduces local energy entropy of the image as a partitioning criterion. Local energy entropy characterizes the complexity of gray-level distribution within a pixel's neighborhood: background areas have simple textures and concentrated gray levels, resulting in lower local entropy; areas containing anomalies have rich textures and dispersed gray levels, resulting in higher local entropy. Based on this, partitioning is performed using the global average entropy of the entire image as a boundary, and thresholds are applied to each partition separately, thereby achieving adaptive threshold changes with image content. It should be noted that using entropy as an information metric to distinguish between simple textured backgrounds and textured anomalies, and designing contrasting threshold strategies for the two types of regions accordingly (adhering to local conditions to suppress fluctuations in the background area, and raising the threshold on a global baseline to highlight strong reflections and retain weak anomalies in the anomaly area), cannot be directly obtained from a global thresholding scheme through obvious parameter adjustments: it requires first recognizing that the inevitable failure of a fixed threshold stems from the fact that the background and anomalies are bound by the same criterion, and then reconstructing the discrimination criterion from the perspective of local information content. Specifically: Obtain the offset image In the preset local window (like Local energy entropy within a pixel : (10) In the formula, The total number of gray levels (e.g.) ); For window The inner grayscale value is The percentage of pixels.

[0057] Calculate the global average entropy for the entire image. : (11) In the formula, and These represent the number of pixels in the image horizontally and vertically, respectively. For each pixel... Perform the following judgment: If its local entropy is less than or equal to the global average entropy ( If a point is identified as a background area, its local grayscale mean is calculated. As a binarization threshold; if its local entropy is greater than the global average entropy ( If a region is identified as a suspected abnormal area, the global grayscale average of the entire image is calculated. With a magnification factor of not less than one The product of these values ​​is used as the binarization threshold. Adaptive thresholding. for: (12) In the formula, For Center, window The local grayscale mean within; This is the global grayscale mean of the entire offset image. The offset image is checked point by point: if the amplitude at that point is greater than or equal to... Then set to Otherwise set to Obtain the binarized image : (13) In the background region, the pixel is binarized using the neighborhood mean that fits the local background as the threshold. Since the local energy entropy of the background pixel does not exceed the global average entropy, it is not classified as a suspected anomalous region and is therefore not included in the subsequent speed selection and orientation of suspected anomalous points, thus suppressing the interference of cluttered background reflections on subsequent processing. In the suspected anomalous region, the threshold is set by multiplying the global mean by a magnification factor greater than one. This threshold is raised relative to the global benchmark, which can highlight strong energy reflections. At the same time, since it is based on the global benchmark rather than the local high benchmark of the anomalous region, it is not raised excessively, thus preserving the relatively weak effective signals in the suspected anomalous region.

[0058] This step adaptively distinguishes pixels into background and suspected anomaly regions using local energy entropy. Pixels in the background region whose local energy entropy does not exceed the global average entropy are excluded from the subsequent suspected anomaly point calculation for speed selection and orientation. This suppresses the interference of cluttered background reflections on subsequent processing. At the same time, strong energy reflections are highlighted in the suspected anomaly region with a threshold higher than the global benchmark, while relatively weak real anomaly signals are preserved. Compared to a fixed threshold, this step improves the recognition rate of weak anomalies and reduces the false positive rate, providing more reliable anomaly identification results for subsequent speed selection and orientation.

[0059] In a preferred embodiment, calculating the local energy entropy includes: calculating the pixel proportion corresponding to each grayscale value within the local window, summing the products of each pixel proportion and its corresponding base-2 logarithm, and taking the negative of the summation result as the local energy entropy. More preferably, the magnification factor of not less than 1 ranges from 1.0 to 1.5, and the size of the preset local window is... The value range is 1 pixel; this range can effectively raise the judgment threshold and highlight strong reflections in suspected abnormal areas, without suppressing weak anomalies due to excessive magnification. The window can strike a balance between the statistical stability of entropy estimation and spatial resolution.

[0060] Step S5: Accurately calculate the medium wave velocity.

[0061] The input for this step is the radar image from step S1 (taking a segment of the radar image as the research object) and the pre-set velocity scanning range. With velocity scan interval The azimuth angles of the receiving antennas corresponding to each anomaly point; the principle for retrieving the optimal wave velocity is described in [reference needed]. Figure 4 .

[0062] Migratory imaging requires pre-setting the electromagnetic wave propagation speed. Improperly set wave velocity results in poor imaging and large errors in estimating the location of anomalies. Traditional methods use the image focusing index as the sole criterion for velocity selection; however, in complex underground media containing water, multiple different preset wave velocities may correspond to similar focusing effects, leading to multiple solutions (ambiguities) in velocity selection and making it difficult to uniquely and accurately determine the optimal wave velocity.

[0063] Existing offset velocity estimation schemes generally rely solely on a single image focusing criterion for velocity selection. This involves imaging each candidate wave velocity within the velocity scanning range, calculating the focusing criterion for each candidate wave velocity, and selecting the candidate wave velocity with the optimal (minimum) focusing criterion as the optimal wave velocity. The fundamental limitation of this single focusing criterion scheme is that focusing only characterizes the convergence of diffraction energy in spatial location. However, in complex underground media, such as those containing water, the relationship between dielectric constant and wave velocity is no longer singular. Multiple different candidate wave velocities may correspond to similar spatial focusing effects, resulting in a smooth, extremely small range or even multiple similar minimum values ​​for the focusing criterion on the velocity axis. This leads to multiple solutions (ambiguities) in velocity selection, making it difficult to uniquely determine the true wave velocity. The essential difference in this approach is that, in addition to the spatial focusing criterion, a completely different criterion is introduced: "the azimuth consistency of the receiving antenna corresponding to the anomaly point," and the two are combined into a joint objective function for optimization. It is worth emphasizing that in existing technologies, azimuth information is only used for final target orientation (i.e., determining velocity first, then using azimuth imaging or determining azimuth). However, this step moves azimuth consistency forward and uses it in reverse as a constraint condition in the velocity inversion stage, allowing azimuth information to participate in velocity determination. This approach of using angle domain consistency feedback to constrain velocity domain inversion is not inspired by existing technologies that only use focusing indicators to select velocity and where azimuth is only used for terminal orientation.

[0064] This invention introduces an azimuth consistency constraint as a second criterion, in addition to the focusing index, to form a joint objective function. The physical basis for this is that when the preset wave velocity is correct, the diffraction energy of the same real anomalous object in the corresponding images of each receiving antenna is not only fully focused in space (small spatial dispersion of the anomalous point), but also the azimuth of the receiving antennas corresponding to each anomalous point exhibits high consistency (small azimuth standard deviation). However, when the preset wave velocity is mismatched, the imaging defocus not only disperses the anomalous points in space (increased spatial dispersion), but also causes stray points that should not be judged as anomalous to be misjudged as anomalous in images of receiving antennas with different orientations, thus widening the azimuth distribution of the antennas to which the anomalous points belong (increased azimuth standard deviation). Therefore, although both spatial focusing and azimuth consistency originate from the correctness of the wave velocity, they impose constraints on the wave velocity from two independent dimensions: spatial position convergence and azimuth distribution convergence. The key is that when the focusing criterion becomes flat and insufficient to distinguish true from false within a certain velocity neighborhood due to the complexity of the medium, the independent dimension of azimuth consistency can still provide effective discrimination. When the two are combined, their minimum values ​​mutually lock, thus converging the flat, multi-solution minimum interval under a single focusing criterion into a unique, sharp optimal solution. This construction, using an azimuth criterion independent of focusing but also pointing to the correct wave velocity to eliminate the multi-solution nature of the focusing criterion, is precisely what distinguishes this step from existing single-criterion velocity selection, and is not obvious. The specific steps are as follows: Set up velocity scanning ranges and intervals, and take a certain segment of radar imagery as the research object to analyze each candidate wave velocity. Repeat step S3 to perform the migration imaging process; for each candidate wave velocity, perform adaptive binarization processing in step S4 to obtain the corresponding binarized image. .

[0065] Calculate the wave velocity for each candidate Focusing index of the binarized image To characterize the degree of spatial focusing (convergence) of anomalous energy: (14) In the formula, Candidate wave velocity All abnormal points located within the suspected abnormal area and whose binarization result is 1 in the binarized image (i.e., A set of points; The centroid of the horizontal distance coordinates and vertical depth coordinates of all anomalies within this set; This represents the total number of outliers. Focus Index The focusing index characterizes the degree of dispersion of each anomalous point relative to its spatial centroid: when the wave velocity is correct, the diffraction energy converges, the anomalous points are compactly distributed, and the dispersion is small; when the wave velocity is mismatched, the image is out of focus and trails, the anomalous points are scattered, and the dispersion is large. Therefore, the smaller the focusing index, the more focused the anomalous energy is in space. It should be noted that the focusing index used in this step is the second moment dispersion of the anomalous point relative to its spatial centroid, which describes the spatial dispersion range of the anomalous points; it is different from the measurement method that only uses the proportion of the number of anomalous points (such as the ratio of the total number of anomalous points to the total number of pixels) to characterize the degree of focus - the latter only counts how many points are judged as anomalous, without reflecting whether these points are compactly clustered or divergent in space. Using the second moment dispersion can more directly and sensitively characterize the spatial convergence state of diffraction energy, and it and the azimuth standard deviation mentioned below are both discrete measures obtained for the same set of anomalous points. The two are mathematically consistent and together constitute a unified characterization of the discreteness of the spatial distribution and azimuth distribution of the anomalous point set.

[0066] For each candidate wave velocity Statistically count all suspected outliers in the binarized image (i.e. Standard deviation of azimuth of the point : (15) In the formula, For the first The receiving antenna azimuth angles corresponding to the suspected anomalies; This is the average azimuth angle of all suspected anomalies under this candidate wave velocity; This represents the total number of suspected anomalies in the binarized image under the candidate wave velocity. The optimal wave velocity is then determined. : (16) In the formula, , These are the weighting coefficients, and their sum is... This can be adjusted based on data quality: if the offset imaging effect is good, it can be adjusted... Set slightly larger (e.g.) , Conversely, Set slightly larger (e.g.) , It should be noted that the focus index... It has the dimension of the square of spatial distance, and the standard deviation of the azimuth angle. Both have angular dimensions, but their dimensions and numerical scales are different; weighting coefficients , While reflecting the relative importance of the two criteria, it also serves the purpose of dimensional and scale normalization, meaning that in actual calculations, it can first be... and Each criterion is normalized to a dimensionless comparable interval (e.g., divided by its maximum value within the scan interval) and then weighted and summed to make the two criteria numerically comparable. Thus, the focusing index... With azimuth standard deviation Both are discrete measures obtained from the same set of outliers, and both reach the minimum value at the optimal wave velocity. The candidate wave velocity corresponding to the minimum value after normalization and weighted summation is the optimal wave velocity, thus forming a dual convergence criterion of "spatial dispersion + azimuth dispersion".

[0067] By introducing azimuth consistency constraints, this step solves the problem of velocity ambiguity in complex water-bearing media by relying solely on focusing exponent velocity selection. This significantly improves the physical rationality and positioning accuracy of medium wave velocity estimation, and effectively suppresses false or missed anomalies caused by incorrect velocity settings. The output of this step is the optimal electromagnetic wave propagation velocity. and with The final offset image and the final binarized image obtained after re-executing steps S3 and S4 are used for step S6 to calculate the abnormal azimuth angle.

[0068] As a preferred embodiment, the image focus index is calculated as follows: Pixels located within the suspected anomaly region and whose binarization result is one in the candidate binarized image are considered suspected anomaly points. The spatial centroid of the lateral distance coordinates and longitudinal depth coordinates of all suspected anomaly points is determined. The sum of the squares of the differences between the lateral distance coordinates and longitudinal depth coordinates of each suspected anomaly point and the corresponding coordinates of the spatial centroid is averaged across all suspected anomaly points. The resulting average value is the image focus index. The standard deviation of the anomaly point azimuth angle is the arithmetic square root of the square mean of the differences between the azimuth angle of the receiving antenna corresponding to all suspected anomaly points and the mean azimuth angle of all suspected anomaly points. More preferably, the sum of the first weighting coefficient and the second weighting coefficient is equal to one, and multiple discrete velocity values ​​are extracted as candidate wave velocities within the velocity scanning interval according to a preset velocity scanning interval. Even more preferably, the velocity scanning interval is a velocity interval determined by extending a preset range upwards and downwards based on the preset medium wave velocity, thereby efficiently locking the optimal wave velocity within a reasonable neighborhood using the initial wave velocity obtained in step S2 as the search center.

[0069] Step S6: Calculate the abnormal azimuth angle.

[0070] The input for this step is based on the optimal wave speed. The obtained multiple final offset images (corresponding to multiple receiving antennas and their azimuth angles), and the maximum reflected wave amplitude extracted from each final offset image at the depth of the anomalous body. With phase and the azimuth angles of each receiving antenna. .

[0071] Traditional methods typically use simple arithmetic averages or amplitude (energy) square weighted sums to calculate abnormal azimuth angles, ignoring the differences in signal quality between different receiving antennas. In actual detection, interference signals from non-target directions (such as side reflections or multiple reflections), even with large amplitude values, can significantly deflect the calculated azimuth angle, reducing directionality accuracy.

[0072] Existing multi-antenna azimuth calculation schemes either use a simple arithmetic average of the azimuth angles of each antenna or a weighted average with the square of the reflection amplitude (energy) as the weight. A common limitation of these two schemes is that their weights depend solely on the signal strength (amplitude or energy), completely disregarding the signal's authenticity / reliability. However, in actual detection, interference from non-target directions such as side reflections and multiple reflections often has a strong or even considerable reflection amplitude. Weighting by amplitude or amplitude squares can result in higher weights, significantly skewing the azimuth result. The fundamental difference in this approach is that, in addition to the amplitude, a phase consistency factor representing signal reliability is introduced into the weighting. This ensures that the weights depend simultaneously on signal strength and whether the phase is consistent with the mainstream phase of each antenna, thus automatically reducing the weight of interfering antennas that, while strong, have a phase deviation.

[0073] This invention proposes an abnormal azimuth angle calculation method based on signal quality weighting, which suppresses the negative impact of interference signals by introducing a phase consistency factor. The physical basis is that reflected signals from the same real target should have a consistent phase across all receiving antennas; while interference signals from non-target directions such as the side or multiple reflections often deviate significantly from the mainstream phase. Accordingly, the final offset images corresponding to each receiving antenna are first subjected to adaptive binarization processing in step S4, and the resulting suspected abnormal areas are identified as target anomalies, thereby determining their depth position.

[0074] It should be noted that after performing the adaptive binarization process on the final offset image, the resulting binarized image may contain only one suspected anomalous area, or it may contain multiple suspected anomalous areas distributed at different depths. When there is only one suspected anomalous area in the binarized image, that suspected anomalous area is identified as the target anomalous body, and the depth corresponding to its center in the longitudinal depth direction is determined as the depth position of the target anomalous body. When there are multiple suspected anomalous areas in the binarized image, the target anomalous body and its depth position can be determined in either of the following ways: First, count the number of pixels or the sum of reflected wave energy in each suspected anomalous area, select the suspected anomalous area with the largest number of pixels or the largest sum of reflected wave energy as the target anomalous body, and determine the depth corresponding to the center of that suspected anomalous area in the longitudinal depth direction as the depth position; Second, treat each suspected anomalous area as a target anomalous body, and calculate the target azimuth angle of each target anomalous body separately according to the phase consistency factor weighting method in the subsequent step, thereby achieving sequential orientation of multiple anomalous bodies. Those skilled in the art can choose one of the above methods according to actual detection needs; when only a single main target needs to be oriented, the former method is preferred, that is, the suspected anomalous area with the most pixels or the largest sum of reflected wave energy is taken as the target anomalous body. After determining the depth position of the target anomalous body, that is, at the depth position of the final offset image corresponding to each receiving antenna, the maximum reflected wave amplitude and corresponding phase of each receiving antenna are extracted respectively.

[0075] The formula for calculating abnormal azimuth angles is: (17) (18) In the formula, This is the final abnormal azimuth angle. This represents the total number of receiving antennas. For the first The azimuth angle of each receiving antenna. For the first Weighting coefficients for each receiving antenna, For the first The maximum reflected wave amplitude of each receiving antenna For the first The phase of each receiving antenna, The average value is the phase value. In equation (18), the natural constant is used as the base and the phase deviation is used as the average value. The exponent term of the term with the negative of the exponent is the phase coherence factor: the greater the phase deviation, the smaller this factor, and the lower the weight of the corresponding receiving antenna. Thus, the weighting coefficients... It also incorporates signal strength (from amplitude) (Reflected by) and signal reliability (reflected by phase consistency factor).

[0076] Phase consistency weighting can automatically reduce the weight of receiving antennas whose phase deviates far from the mainstream, thereby effectively suppressing interference signals from non-target directions (side, multiple reflections), which helps to improve the calculation accuracy of abnormal azimuth angles and the accuracy of directional detection.

[0077] As a preferred implementation, considering that the azimuth and phase of each receiving antenna are periodic quantities (azimuth with a period of 360 degrees and phase with a period of 2π), when calculating the standard deviation of the azimuth of the outlier point according to formula (15), the outlier azimuth according to formula (17), and the phase consistency factor according to formula (18), the circular (vector) statistical method can be used to replace the corresponding linear arithmetic operation, so as to avoid the deviation caused by simple linear averaging when the azimuth or phase crosses the 0 degree (360 degrees) or ±π boundary, thereby further improving the robustness of the calculation of the standard deviation of the azimuth and the outlier azimuth.

[0078] Specifically, the azimuth standard deviation in equation (15) is replaced with the circumferential standard deviation. That is, the azimuth of the receiving antenna corresponding to each suspected anomaly point is first expressed as a unit complex exponential vector and the vector sum is calculated. The average resultant vector length is obtained from the normalized modulus of the resultant vector, and then the circumferential standard deviation of the azimuth is calculated according to the following formula: (19) Replace the arithmetic weighted average in equation (17) with a vector weighted average, that is, perform a complex exponential vector weighted summation on the azimuth angles of each receiving antenna according to their weighting coefficients, and take the argument of the composite vector as the final abnormal azimuth angle, as shown in the following equation: (20) In equation (20), arg(·) represents the argument of a complex number; correspondingly, the phase average value in equation (18) is taken as the circumferential average value of the phases of each receiving antenna, that is, obtained by taking the argument of the sum of the unit complex exponential vectors of each phase, and the phase deviation is taken as the circumferential phase difference between the corresponding phase and the circumferential phase average value (taking the principal value, the value range is 0-π). Thus, when the target azimuth or phase is close to 0 degrees (360 degrees) or ±π periodic boundaries, the above circumferential (vector) processing can still give the correct azimuth and azimuth dispersion measure, constituting a preferred improvement to the aforementioned linear calculation method; while when the target azimuth is obviously concentrated and does not cross the above boundaries, the calculation results of the circumferential method and the linear method tend to be consistent.

[0079] It should be noted that steps S4, S5, and S6 are not simply a superposition of each other that can be arbitrarily separated or replaced, but rather they support each other functionally, depend on each other, and form an organic whole. This organic nature can be explained by the fact that removing or replacing any one of the steps will lead to a significant decrease or even failure of the overall effect.

[0080] First, step S4 is a prerequisite for step S5 to be valid. Step S5 has two criteria: the focus index... With azimuth standard deviation --None of these calculations are performed directly on the grayscale shifted image; instead, they are based on the set of outliers extracted during the adaptive binarization in step S4. Above. If step S4 is missing, or replaced by conventional global single-threshold binarization, then in cases of beam velocity mismatch and image defocus, weak anomalies will be submerged and strong backgrounds will be misjudged. The extracted anomaly point set will be incomplete and contain a large number of false points. Consequently, the calculated focus index and azimuth standard deviation will be severely distorted, and the joint optimization in step S5 will lose its reliable basis. It is evident that the clean and reliable anomaly point morphology provided by step S4 for step S5 is a necessary condition for the latter to be effective.

[0081] Secondly, step S5 is a prerequisite for the accuracy of step S6. The maximum reflected wave amplitude value extracted in step S6... With phase All are taken from the optimal wave speed The final offset image obtained from the re-imaging. If step S5 is missing and the initial wave velocity from step S2 is used directly for imaging, the image will be underfocused in complex water-bearing media due to multiple wave velocity solutions and estimation errors. The depth position of the target anomaly will be inaccurate, and the amplitude and phase extracted at that position will be distorted. The phase consistency weighting in step S6 will then become a fine weighting on erroneous data, and its orientation accuracy cannot be guaranteed. It is evident that the "final image that is fully focused at the correct wave velocity" provided by step S5 for step S6 is a necessary condition for the latter to be effective.

[0082] Thirdly, these three methods address three aspects of the same medium problem in different processing domains: image clutter and weak anomalies easily being obscured (image domain) – solved by adaptive partitioning binarization in step S4; complex water-bearing media leading to multiple wave velocity solutions and failure of focusing criteria (velocity domain) – solved by introducing azimuth consistency constraints in step S5; and multipath interference such as lateral reflections and multiple reflections causing azimuth deviation (angle domain) – solved by phase consistency weighting in step S6. The three methods progress step-by-step from image domain purification to velocity domain calibration to angle domain orientation, with the output of each step forming a reliable input for the next, all serving the overall technical goal of achieving high-precision directional detection of anomaly depth, distance, and azimuth in complex water-bearing, multipath-bearing media. Because of the aforementioned interdependent and mutually conditional coupling relationships between each step, the directional detection accuracy achieved by this scheme exceeds the simple sum of the effects achievable by using the three processes in isolation. It constitutes a synergistic organic whole, rather than a feature combination that can be easily pieced together. Finally, the azimuth angle obtained in step S6 is... By combining the distance information obtained in step S1 with the depth information obtained in steps S3 and S5, the orientation and location of the anomaly in three-dimensional space can be realized, that is, the determination of depth, distance and orientation.

[0083] Example 2 This embodiment provides a borehole radar directional detection device based on a circumferential multi-receiver antenna array, the structure of which is described in [reference needed]. Figure 3 The device includes a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements the steps of the borehole radar directional detection method based on a circumferential multi-receiver antenna array described in Embodiment 1, namely, steps S1 to S6.

[0084] In a specific implementation, the processor can be a central processing unit, a digital signal processor, a field-programmable gate array, or other devices with data processing capabilities; the memory can be random access memory, read-only memory, flash memory, or other computer-readable storage media. The device can be integrated into the main control unit of the aforementioned directional radar data acquisition device, where the main control unit performs the aforementioned data processing while acquiring data; or it can function as a standalone computer device, reading the raw data stored in the main control unit and performing the aforementioned processing offline.

[0085] Based on the functional divisions, the computer program, when executed, can form the following processing modules: a data preprocessing module for implementing range and azimuth correction in step S1 and offset imaging processing in step S3; an adaptive binarization module for implementing adaptive binarization processing based on local energy entropy in step S4; a wave velocity inversion module for implementing optimal wave velocity inversion based on a joint objective function in step S5; and an anti-multipath weighted orientation module for implementing target azimuth calculation based on phase consistency factor weighting in step S6. The above module division is merely illustrative; those skilled in the art can combine or further split them as needed. Any combination of processor and memory capable of implementing the method steps described in Embodiment 1 falls within the protection scope of this embodiment.

[0086] Example 3 This embodiment provides a borehole radar directional detection device based on a circumferential multi-receiver antenna array. The cross-sectional structure of the circumferential receiving antenna array in the device is shown below. Figure 5 The device includes a borehole radar probe 1, on which a transmitting antenna 2 and a circumferential multi-receiving antenna array are arranged on the same cross-section. The circumferential multi-receiving antenna array consists of at least three receiving antenna units, such as 3-1, 3-2, 3-3, and 3-4, collectively referred to as receiving antenna units 3. They are arranged at equal angles along the circumference of the probe 1, with the main lobe direction 4 of each receiving antenna unit 3 being approximately perpendicular to the probe axis and pointing radially outward toward the borehole wall 5. The device also includes independent receiving channels corresponding to each receiving antenna unit 3, used for synchronously or time-divisionally sampling radar wave fields excited by the same transmitting source, and weighting the azimuth angle of each receiving antenna based on the reflected wave amplitude and phase consistency factor of each receiving antenna to analyze the relative azimuth angle of the borehole periphery anomaly.

[0087] Specifically, the circumferential multi-receiver antenna array arranges multiple receiving antenna units at equal angles along the circumference on the same cross-section of the borehole radar probe, typically three or four units, but can be expanded to more units according to azimuth resolution requirements. Each receiving channel synchronously or time-divisionally samples the radar wave field excited by the same transmitting source. The main lobe direction of each receiving antenna unit is approximately perpendicular to the probe axis and points radially outward, so that the reflected / scattered signals at different azimuths exhibit azimuth-related amplitude differences on each receiving antenna, thereby being used to resolve the relative azimuth angle of borehole periphery anomalies (such as discontinuities, caves, water-rich zones, etc.). Compared to existing single-transmitter-single-receiver or multipole borehole radars, which can only obtain the travel time and amplitude of reflected waves along the logging trajectory and cannot distinguish reflection sources at different azimuths around the borehole, this embodiment obtains borehole periphery azimuth information through circumferential spatial sampling.

[0088] In terms of structure and configuration, the circumferential multi-receiver antenna array has the following characteristics. First, a circumferential equiangular spatial sampling structure: at least three receiving antenna elements 3 are arranged at equal angles along the circumferential direction on the same cross-section of the borehole radar probe 1, forming a circumferential spatial sampling receiving array covering 360° azimuth; the main lobe direction 4 of each receiving antenna element 3 is approximately perpendicular to the probe axis and points radially outward, so that the same reflection / scattering source has a distinguishable incident angle and path length in different receiving channels.

[0089] Secondly, the single-transmitter-multiple-receiver common aperture configuration: a single-transmitter-multiple-receiver (SIMO) common aperture configuration with one transmitting antenna and a circumferential multi-receiver antenna is adopted to achieve multi-directional synchronous or time-division sampling without increasing the borehole diameter and probe outer diameter. Compared with multi-transmitter systems, it can significantly simplify the requirements for radio frequency switching and isolation, and reduce probe power consumption and structural complexity.

[0090] Third, an independent and controllable multi-channel receiving link: each receiving antenna is connected to an independent RF front-end and sampling channel, supporting synchronous sampling or high-precision time-division multiplexing, ensuring the amplitude and phase consistency between channels, so that the same reflection / scattering source can be received by multiple receiving antennas at the same time, providing a multi-channel consistent hardware foundation for subsequent azimuth weighting based on the amplitude and phase consistency factor of the reflected wave of each receiving antenna.

[0091] Fourth, a compact integrated design suitable for small-diameter drilling: the transmitting antenna, the circumferential receiving antenna array and related RF modules are integrated into the same rigid or semi-rigid probe to meet the entry and pushing requirements of engineering drilling (usually less than 110mm in diameter), while ensuring the circumferential symmetry of the circumferential antenna unit and the consistency of the azimuth reference, which facilitates the subsequent azimuth angle calibration and error correction.

[0092] It is important to further explain that the significance of arranging multiple receiving antennas circumferentially along the same cross-section of the probe lies in resolving the azimuth of the borehole's reflection / scattering sources relative to the borehole. Since all receiving antennas in the array are located on the same cross-section, their depth and distance relative to the same target are essentially the same. The only physical quantity that can distinguish each receiving antenna from the others is its azimuth directivity. If each receiving antenna is an omnidirectional antenna, the amplitude and phase received by each antenna on the same cross-section from the same target will be essentially the same, and there will be no comparable differences between the receiving channels. The circumferential arrangement will not provide any azimuth information. Only when each receiving antenna has azimuth directivity, that is, each has a high receiving gain only for waves arriving from a certain angular range directly in front of it, will the same target exhibit azimuth-related amplitude and phase differences on each receiving antenna, thereby allowing the determination of its azimuth.

[0093] In this embodiment, the azimuth directivity is achieved by setting a shielding partition 6 between adjacent receiving antennas. The shielding partition 6 has a shielding effect on electromagnetic waves, so that each receiving antenna mainly receives waves from the sector directly in front of it while suppressing waves from other directions. This constrains the receiving antenna with a wide azimuth response into a receiving unit with azimuth directivity. At the same time, the angular width of the sector directly in front of it is greater than the circumferential angular interval of the adjacent receiving antennas, so that the receiving fields of view of the adjacent receiving antennas overlap. This ensures that the same reflection / scattering source can simultaneously enter multiple adjacent receiving channels, forming multiple signals that can be compared in amplitude and phase. This avoids the situation where only a coarse azimuth at the sector level can be resolved because the fields of view of each antenna do not overlap. As an alternative, the azimuth directivity can also be achieved by a receiving antenna unit that is itself azimuth directive (e.g., an antenna unit with a back reflection or shielding structure). In this case, the shielding partition between adjacent receiving antennas can be reduced or omitted accordingly.

[0094] It should be noted that in this embodiment, each receiving antenna is a directional antenna, with its main lobe pointing to its respective circumferential direction and having a high receiving gain for reflected signals from that direction. Therefore, the same reflection / scattering source exhibits amplitude differences related to its azimuth on different receiving antennas. Simultaneously, due to the aforementioned azimuth directivity of each receiving antenna and the overlapping receiving fields of adjacent antennas, the same reflection / scattering source can be simultaneously received by multiple adjacent receiving antennas with different receiving gains, thus forming multiple comparable signals in each corresponding receiving channel. It is this characteristic of "the same reflection source entering multiple receiving channels" that enables the azimuth weighting based on the phase consistency factor in the method described in Embodiment 1 to be implemented: at the same depth of the target anomalous object, the amplitude and phase of the reflected wave from each receiving antenna are extracted, and the product of the amplitude and the phase consistency factor is used as the weight to weight the azimuth angle of each receiving antenna, thus obtaining the azimuth angle of the target anomalous object. It should be noted that the aforementioned shielding partition is intended to impart azimuth directivity to each receiving antenna, rather than completely isolating them. If the partition prevents the receiving sectors of each antenna from overlapping and the same reflection source falls into only a single sector, it becomes impossible to obtain the amplitude and phase of the same reflection source across multiple channels for comparison. Therefore, the receiving sectors of adjacent receiving antennas should be designed to overlap appropriately. Thus, the circumferential multi-receiving antenna array of this embodiment constitutes the hardware prerequisite for implementing the directional detection method described in Embodiment 1. The two work together to form a complete borehole radar directional detection scheme.

[0095] The workflow of the whole-machine directional detection device described in this embodiment is as follows: First, the borehole radar probe is inserted into the borehole and moves along the borehole axis. The transmitting antenna emits electromagnetic waves around the borehole. Each receiving antenna arranged in a ring simultaneously or in a time-division manner samples the radar wave field excited by the same transmitting source through its own independent receiving channel. At the same time, the moving distance of the probe is recorded by a pulley counter and the azimuth reference of the probe is recorded by an electronic compass, thereby obtaining multi-channel radar profile data corresponding to each receiving antenna and the azimuth angle of each receiving antenna. Subsequently, the processor executes the computer program stored in the memory and sequentially performs steps S1 to S6 of the method described in Embodiment 1 on the above multi-channel data, including range and azimuth angle correction, offset imaging, adaptive partition binarization based on local energy entropy, optimal wave velocity inversion based on joint objective function, and azimuth weighting based on phase consistency factor. Finally, the azimuth angle of each receiving antenna is weighted and averaged using the product of the reflected wave amplitude of each receiving antenna and the phase consistency factor as weights, and the azimuth angle and depth position of the target anomaly are output, thereby realizing the three-dimensional directional detection of the borehole anomaly.

[0096] Through the aforementioned hardware structure, this embodiment transplants the concept of multi-channel azimuth arrays from ground radar to the borehole environment. It introduces the crucial azimuth dimension into the traditional two dimensions of borehole depth and radial distance, achieving circumferential azimuth resolution. The multi-channel radar profile data acquired by the circumferential multi-receiver array can be processed by the data processing method described in Embodiment 1, which is based on adaptive partitioning binarization, joint objective function velocity selection, and phase consistency weighting. This enables three-dimensional directional detection of borehole perimeter anomalies, completing the transition from traditional two-dimensional profile detection to three-dimensional spatial directional detection.

[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0098] The contents not described in detail in this specification are existing technologies known to those skilled in the art.

Claims

1. A borehole radar directional detection method based on a circumferential multi-receiver antenna array, characterized in that, Includes the following steps: The azimuth angles of each receiving antenna in the circumferential multi-receiver antenna array are obtained, as well as multiple radar image maps generated from the radar electromagnetic wave data of each receiving antenna, each corresponding to a specific receiving antenna. A preset medium wave velocity is determined, and the optimal wave velocity is inverted based on a joint objective function: Multiple candidate wave velocities are acquired within a velocity scanning interval determined by the preset medium wave velocity; for each candidate wave velocity, offset imaging processing is performed on multiple radar images to obtain multiple offset images corresponding to each receiving antenna; adaptive binarization processing based on local energy entropy is performed on each offset image to obtain multiple candidate binarized images; the image focus index and the standard deviation of the azimuth angle of outliers are calculated based on the multiple candidate binarized images corresponding to each candidate wave velocity; an objective function is constructed including the product of a first weighting coefficient and the image focus index, and the product of a second weighting coefficient and the standard deviation of the azimuth angle of outliers; the candidate wave velocity that minimizes the objective function is selected as the optimal wave velocity, and the offset imaging processing is performed on multiple radar images using the optimal wave velocity to obtain multiple final offset images corresponding to each receiving antenna; The adaptive binarization process based on local energy entropy is as follows: The local energy entropy of each pixel within a preset local window and the global average entropy of the entire offset image are calculated respectively; when the local energy entropy is less than or equal to the global average entropy, the corresponding pixel is determined to be located in the background area, and the local grayscale mean within the local window is used as the binarization threshold for that pixel; when the local energy entropy is greater than the global average entropy, the corresponding pixel is determined to be located in a suspected abnormal area, and the product of the global grayscale mean of the entire offset image and a magnification factor not less than one is used as the binarization threshold for that pixel; the offset image is binarized point by point according to the determined binarization threshold to obtain the corresponding binarized image; The image focusing index is an index characterizing the degree of dispersion of each suspected anomaly point relative to its spatial centroid. The smaller the image focusing index, the more focused the anomalous energy is in space. The standard deviation of the azimuth angle of the anomaly point is as follows: the pixels in the candidate binarized images corresponding to each receiving antenna that are determined to be located in the suspected anomaly area and whose binarization result is one are identified as suspected anomaly points. The azimuth angle of the receiving antenna to which each suspected anomaly point belongs is taken as the azimuth angle of the suspected anomaly point. The standard deviation of the azimuth angle of all suspected anomaly points relative to the mean of the azimuth angles of all suspected anomaly points is the standard deviation of the azimuth angle of the anomaly point. The target azimuth angle is calculated based on the phase consistency factor weighting: the adaptive binarization process is performed on the final offset image corresponding to each receiving antenna, the suspected abnormal area obtained by the adaptive binarization process is determined as the target abnormal body, and the depth corresponding to the target abnormal body is determined as the depth position of the target abnormal body. At the depth position of the final offset image corresponding to each receiving antenna, the maximum reflected wave amplitude and corresponding phase are extracted respectively, and the phase average value of the corresponding phases of all receiving antennas is calculated; the product of the maximum reflected wave amplitude of each receiving antenna and the phase consistency factor is used as the weighting coefficient of each receiving antenna, and the azimuth angle of each receiving antenna is weighted and averaged using the weighting coefficient to obtain the target azimuth angle of the target anomaly; wherein, the phase consistency factor is an exponential function with the natural constant as the base and the negative of the phase deviation as the exponent, and the phase deviation is the absolute value of the difference between the corresponding phase and the phase average value.

2. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The calculation of the local energy entropy of each pixel within a preset local window includes: calculating the pixel percentage corresponding to each gray value within the local window, accumulating the product of each pixel percentage with its corresponding base-2 logarithm, and taking the negative of the accumulation result as the local energy entropy.

3. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1 or 2, characterized in that, The magnification factor of not less than 1 ranges from 1.0 to 1.5; the size of the preset local window is 5 by 5 pixels.

4. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The image focus index is calculated as follows: the spatial centroid of the lateral distance coordinates and longitudinal depth coordinates of all suspected anomalies is determined, and the sum of the squares of the differences between the lateral distance coordinates and longitudinal depth coordinates of each suspected anomaly relative to the corresponding coordinates of the spatial centroid is averaged over all suspected anomalies. The average value obtained is the image focus index. The standard deviation of the azimuth angle of the anomaly point is the arithmetic square root of the square mean of the difference between the azimuth angle of the receiving antenna corresponding to all suspected anomalies and the mean azimuth angle of all suspected anomalies.

5. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1 or 4, characterized in that, The sum of the first weighting coefficient and the second weighting coefficient is equal to 1; within the velocity scanning interval, multiple discrete velocity values ​​are extracted as each candidate wave velocity according to a preset velocity scanning interval.

6. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The velocity scanning range is a velocity range determined by extending a preset range upwards and downwards to both sides based on the preset medium wave velocity.

7. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The determination of the preset medium wave velocity includes: selecting a region with hyperbolic features in the radar image, extracting the time and spatial coordinates of the hyperbola vertex and at least two reflection points in the region, establishing a wave velocity solution equation based on the geometric path relationship, deriving the preliminary electromagnetic wave propagation speed and using it as the preset medium wave velocity.

8. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The offset imaging processing is a frequency-wavenumber method-based offset imaging processing, including: performing a two-dimensional Fourier transform on the radar image in the lateral range and time directions to obtain wavenumber-frequency domain data; performing wavefield extension on the wavenumber-frequency domain data along the longitudinal depth direction; performing a two-dimensional inverse Fourier transform on the extended data, and taking the imaging time as zero based on the mapping relationship between frequency and wavenumber using the medium wave velocity adopted in the offset imaging processing to obtain the offset image.

9. The borehole radar directional detection method based on a circumferential multi-receiver antenna array according to claim 1, characterized in that, The azimuth angles of each receiving antenna and the multiple radar images corresponding to each receiving antenna are obtained in the following way: the initial azimuth angles and radar electromagnetic wave data of each receiving antenna in the one-transmitter-multiple-receiver antenna array are obtained, the initial azimuth angles corresponding to the electromagnetic wave data of each receiving antenna are corrected by combining the probe movement distance and rotation angle, and multiple radar images corresponding to each receiving antenna are generated based on the corrected azimuth angles and the radar electromagnetic wave data.

10. A borehole radar directional detection device based on a circumferential multi-receiver antenna array, characterized in that, The device includes a borehole radar probe, on which a transmitting antenna and a circumferential multi-receiving antenna array are provided on the same cross-section. The circumferential multi-receiving antenna array is composed of at least three receiving antenna elements arranged at equal angles along the circumference of the probe. The main lobe direction of each receiving antenna element is approximately perpendicular to the probe axis and points radially outward. The probe is also provided with an independent receiving channel corresponding to each receiving antenna element, for synchronous or time-division sampling of radar wave fields excited by the same transmitting source. The method includes a processor and a memory, the memory storing a computer program that, when executed by the processor, implements the steps of the method according to any one of claims 1 to 9.

Images