An underground cable detection method based on compression imaging technology
By reconstructing two-dimensional images of underground cables using ground-penetrating radar and compressed imaging technology, the problems of low accuracy, slow speed, and high cost in existing technologies have been solved, achieving efficient and low-cost underground cable detection.
Patent Information
- Application Number
- CN202211701980.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-12-29
- Publication Date
- 2026-02-06
- Estimated Expiration
- 2042-12-29
AI Technical Summary
Existing technologies for underground cable detection suffer from low accuracy, slow speed, and high cost, especially the fixed-distance excavation method, which has large errors and low efficiency of manual detection.
By combining ground-penetrating radar with compressed imaging technology, the echo signals acquired by ground-penetrating radar are preprocessed and then reconstructed into two-dimensional images of underground cables using compressed sensing methods. The target space is then reconstructed using an orthogonal matching pursuit algorithm, which reduces dimensionality and improves imaging accuracy.
It achieves high-precision, fast, and low-cost underground cable detection with small errors, reducing detection costs and improving detection efficiency.
Smart Images
Figure CN116106975B_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of power setting detection maintenance, and particularly to an underground cable detection method based on compression imaging technology. BACKGROUND
[0002] Power infrastructure is the support for maintaining modern industrialized life. Nowadays, in order to save urban land and achieve aesthetic effect, more power facilities adopt underground cable setting. Compared with overground cable laying, underground cable occupies small space and has high stability. However, due to the incoordination of urban construction, underground cable obviously faces aging problems in different degrees, and may also encounter power outage, fire and other accidents if not timely repaired and replaced.
[0003] At present, the methods for solving this problem mainly include:
[0004] 1. fixed-distance excavation according to the data at the time of construction;
[0005] 2. manual detection of the ground using a detection instrument to find the cable position.
[0006] However, the above two methods have their own defects. The fixed-distance excavation method often has errors or omissions due to the error of the saved data at the time of construction, which may cause power outage accidents and affect normal production and life. The manual detection method using a detection instrument is slow, has poor detection accuracy, and has high labor costs.
[0007] Therefore, there is an urgent need for a non-excavation detection method with high precision, fast speed and low cost. SUMMARY
[0008] The present application aims to overcome the shortcomings of the prior art and provide an underground cable detection method based on compression imaging technology with high precision, fast speed and low cost.
[0009] To achieve the above-mentioned purpose, the technical solution provided by the present application is:
[0010] An underground cable detection method based on compression imaging technology, comprising:
[0011] using a ground penetrating radar to detect a target area to obtain a ground penetrating radar echo signal;
[0012] preprocessing the echo signal;
[0013] analyzing the sparse representation of the preprocessed echo signal by a compression sensing method, creating a dictionary matrix, so as to convert the echo signal from a K·L matrix form to a column vector form;
[0014] mapping the echo signal in the column vector form into a low-dimensional measurement vector y;
[0015] According to the measurement vector y, the target space b is reconstructed, and the reconstructed target space b is converted into a K·L matrix, that is, the target space imaging result.
[0016] Further, the signal obtained by detecting the target area by the ground penetrating radar is blocked, the space is changed from three dimensions to separated two dimensions, and the ground penetrating radar two-dimensional echo signal is obtained, and the process includes:
[0017] The time window is selected, L=2*(f s / f c ), f s / f c is the ratio of the ground penetrating radar sampling rate and the center frequency of the transmitted signal, L is used as the standard for dividing the two time windows, the media with small difference in reflection coefficient underground is approximately regarded as homogeneous medium, it is considered that the ground penetrating waves reaching the same plane have the same loss, and the echoes of signals on the same horizontal plane are approximately regarded as the same time window, and the received signals are divided into λ=M / L approximate planes; a signal plane is selected, that is, the ground penetrating radar two-dimensional echo signal is a matrix , wherein M and N A represent the number of sampling points and the number of scanning channels respectively.
[0018] Further, the echo signal is preprocessed, including:
[0019] The mean value method is used for preprocessing to eliminate direct waves and coupled waves and highlight target signals; wavelet is used to filter noise; and hyperbolic curves are identified to extract target hyperbolic curves.
[0020] Further, by using the compressive sensing method, the sparse representation of the preprocessed echo signal is analyzed, a dictionary matrix corresponding to each aperture is created, and the i-th dictionary matrix is , wherein s(t) represents the transmitted signal, t n represents the time of the n-th iteration, τ i (π j ) represents the two-way delay of the i-th antenna to the target space grid π j , and there are N A apertures. The dictionary matrix corresponding to each aperture is arranged longitudinally to obtain a composite dictionary R as Then the echo signal X=R·b, wherein b is the reflection coefficient at each grid of the target space, is converted from the K·L matrix form to the column vector form.
[0021] Further, the echo signal in the column vector form is mapped to a low-dimensional measurement vector y, including:
[0022] A measurement matrix Q is established, the measurement matrix needs to satisfy the RIP characteristic, a Gaussian random matrix is used, and the measurement matrix of the m-th aperture is N Q ·N AGaussian random matrix q(i,j)~N(0,1), the measurement matrix corresponding to each aperture is placed on the diagonal to form a composite matrix The echo signal X is mapped to a low-dimensional measurement vector y.
[0023] Further, the target space b is reconstructed according to the measurement vector y, and the reconstructed target space b is converted into a K·L matrix, including:
[0024] Using the orthogonal matching pursuit algorithm OMP, the column with the largest inner product value of the linear measurement value y and the recovery matrix Q·R is selected by greedy iteration, the residual control is used to make the selected column have the largest inner product value with the remaining recovery matrix array vector in each iteration, the column is removed from the recovery matrix after each selection, and the iteration stops when the iteration number meets the residual threshold; after the reconstruction of the target space b is completed, it is rewritten into a K·L matrix, that is, the target space imaging result.
[0025] Compared with the prior art, the principles and advantages of the present scheme are as follows:
[0026] The present scheme can automatically identify underground cables by using ground penetrating radar, the application of the compression sensing algorithm can effectively reconstruct the cross-sectional image of the underground situation, the focusing effect is good, the error is much smaller than the previous excavation method, and the detection method, and the application of the compression sensing algorithm reduces the difficulty of reconstructing the image, reduces the cost of underground cable detection, and has high detection efficiency. BRIEF DESCRIPTION OF DRAWINGS
[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the services needed to be used in the embodiments or the prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.
[0028] Figure 1 The present application is a kind of principle flow chart of underground cable detection method based on compression imaging technology. DETAILED DESCRIPTION
[0029] The present application will be further described below in combination with specific embodiments:
[0030] As shown in the figure, the underground cable detection method based on compression imaging technology described in the present embodiment includes the following steps: Figure 1
[0031] S1, using ground penetrating radar to detect target area, and obtaining the echo signal of ground penetrating radar;
[0032] In this step, the signal obtained by ground penetrating radar detection target area is blocked, the space is changed from three-dimensional to separated two-dimensional, and the ground penetrating radar two-dimensional echo signal is obtained, the process includes:
[0033] Select time window, L = 2 * (f s / f c ), f s / f c The ratio of the ground penetrating radar sampling rate and the center frequency of the transmitted signal, using L as the standard for dividing two time windows, the reflection coefficient of the medium with small difference in the ground is approximated as a homogeneous medium, it is considered that the ground wave loss reaching the same plane is the same, the echo of the signal on the same horizontal plane is approximated as the same time window, and the received signal is divided into λ = M / L approximate planes; Select a signal plane, that is, the ground penetrating radar two-dimensional echo signal is a matrix Where M and N A Indicate the number of sampling points and the number of scanning channels respectively.
[0034] S2, pre-processing the echo signal;
[0035] In this step, the mean method is used for pre-processing, the direct wave and coupling wave are eliminated, and the target signal is highlighted; The wavelet is used to filter noise; The hyperbolic curve is identified, and the target hyperbolic curve is extracted.
[0036] S3, analyze the sparse representation of the pre-processed echo signal by the compressive sensing method, create a dictionary matrix, and thus convert the echo signal from K·L matrix form to column vector form;
[0037] In this step, the sparse representation of the pre-processed echo signal is analyzed by the compressive sensing (CS) imaging method, and a dictionary matrix corresponding to each aperture is created, and the i-th dictionary matrix is Where s(t) represents the transmitted signal, t n represents the time of the n-th iteration, τ i (π j ) represents the two-way delay of the i-th antenna to the target space grid π j , there are N A apertures, the dictionary matrix corresponding to each aperture is arranged longitudinally, and the composite dictionary R is Then the echo signal X = R·b, where b is the reflection coefficient of each grid in the target space, is converted from K·L matrix form to column vector form.
[0038] S4, map the echo signal in column vector form to a low-dimensional measurement vector y, the process is as follows:
[0039] Establish a measurement matrix Q, the measurement matrix needs to satisfy the RIP characteristic, use a Gaussian random matrix, and the measurement matrix of the m-th aperture isQ ·N A Gaussian random matrix q(i,j)~N(0,1), the measurement matrix corresponding to each aperture is placed on the diagonal to form a composite matrix The echo signal X is mapped to a low-dimensional measurement vector y.
[0040] S5, reconstruct the target space b according to the measurement vector y, and convert the reconstructed target space b into a K*L matrix, which is the target space imaging result.
[0041] In this step, the process of reconstructing the target space b according to the measurement vector y is as follows:
[0042] Using the orthogonal matching pursuit algorithm OMP, the column with the largest inner product value of the linear measurement value y and the recovery matrix Q*R is selected through greedy iteration, and the residual control is used to make the selected column have the largest inner product value with the remaining recovery matrix array vector in each iteration. After each selection, the column is removed from the recovery matrix, and the iteration stops when the number of iterations meets the residual threshold.
[0043] In this embodiment, the ground penetrating radar is used to automatically identify the underground cable, and the application of the compressive sensing algorithm can effectively reconstruct the cross-sectional image of the underground situation, has good focusing effect, and the error is much smaller than the previous digging method and detection method. At the same time, the application of the compressive sensing algorithm reduces the difficulty of reconstructing the image, reduces the cost of underground cable detection, and has high detection efficiency.
[0044] The above-mentioned embodiments are only the preferred embodiments of the present application, and do not limit the scope of the present application. Any changes made according to the shape and principle of the present application should be covered within the protection scope of the present application.
Claims
1. A method for detecting underground cables based on compressed imaging technology, characterized in that, include: Ground penetrating radar is used to detect the target area and obtain the echo signal of the ground penetrating radar; Preprocess the echo signal; By using compressed sensing, the sparse representation of the preprocessed echo signal is analyzed, and a dictionary matrix is created, thereby converting the echo signal from K·L1 matrix form to column vector form. Map the echo signal in column vector form to a low-dimensional measurement vector y; The target space b is reconstructed based on the measurement vector y, and the reconstructed target space b is converted into a K·L1 matrix, which is the target space imaging result; The signal obtained by ground-penetrating radar from the target area is divided into blocks, transforming the three-dimensional space into a separate two-dimensional space, thus obtaining the two-dimensional echo signal of the ground-penetrating radar. The process includes: Select the time window, L = 2*(f) s / f c ), f s / f c The ratio of the ground-penetrating radar sampling rate to the center frequency of the transmitted signal is used, with L as the standard for dividing the two time windows. Subsurface media with small differences in reflection coefficients are approximated as homogeneous media. It is assumed that the ground-penetrating wave loss reaching the same plane is the same, and the echoes of the signal on the same horizontal plane are approximated as the same time window. The received signal is divided into λ = M / L approximate planes. A signal plane is selected, i.e., the two-dimensional echo signal of the ground-penetrating radar is a matrix. Where M and N A These represent the number of sampling points and the number of scanning channels, respectively. Preprocessing of the echo signal includes: Preprocessing is performed using the mean method to eliminate direct and coupled waves and highlight the target signal; noise is filtered out using wavelets; hyperbolas are identified and the target hyperbola is extracted. The target space b is reconstructed based on the measurement vector y, and the reconstructed target space b is converted into a K·L1 matrix, including: The Orthogonal Matching Pursuit (OMP) algorithm is used to greedily iterate and select the column with the largest dot product between the linear measurement value y and the restoration matrix Q·R. Through residual control, the dot product between the selected column and the remaining column vectors of the restoration matrix is maximized in each iteration. After each selection, that column is removed from the restoration matrix until the number of iterations meets the residual threshold and the iteration stops. After the target space b is reconstructed, it is rewritten as a K·L1 matrix, which is the target space imaging result.
2. The method for detecting underground cables based on compressed imaging technology according to claim 1, characterized in that, By using compressed sensing, the sparse representation of the preprocessed echo signal is analyzed, and a dictionary matrix corresponding to each aperture is created. The i-th dictionary matrix is... Where s(t) represents the transmitted signal, t n τ represents the time of the nth iteration. i (π j ) represents the distance from the i-th antenna to the target space grid π. j The two-way delay is N. A Given several apertures, arrange the dictionary matrix corresponding to each aperture vertically to obtain a composite dictionary R. The echo signal X = R·b, where b is the reflection coefficient at each grid point in the target space, which is converted from K·L1 matrix form to column vector form.
3. The method for detecting underground cables based on compressed imaging technology according to claim 1, characterized in that, Mapping the echo signal in column vector form to a low-dimensional measurement vector y includes: Establish a measurement matrix Q, which must satisfy the RIP property. Use a Gaussian random matrix. The measurement matrix for the m-th aperture is of size N. Q ·N A Gaussian random matrix The measurement matrices corresponding to each aperture are placed on the diagonal to form a composite matrix. The echo signal X is mapped to a low-dimensional measurement vector y.