Method and system for detecting soil body around pipeline based on ground penetrating radar

Abstract

The invention relates to the technical field of data processing, and discloses a method and a system for detecting a soil body around a pipeline based on a ground penetrating radar. The method comprises the following steps: performing generalized reflection coefficient modeling on soil layer distribution data to obtain reflection coefficient model data, and calculating the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; performing feature parameter extraction on the dielectric feature data of the soil body to obtain feature vector data of the soil body, and performing anomaly identification on the feature vector data of the soil body to obtain data of an abnormal region of the soil body; and performing fuzzy hierarchy analysis on the soil body abnormal region data to obtain risk index data, and performing correlation analysis on the risk index data and the pipeline position feature data to obtain risk assessment data of the soil body around the pipeline. Based on the obtained quantification result of the dielectric parameters of the soil body, the accuracy of abnormal region identification is improved, and the accuracy of detection of the soil body around the pipeline by the ground penetrating radar is further improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method and system for detecting soil around a pipeline based on ground penetrating radar. Background Art

[0002] Ground penetrating radar technology has been widely used in the field of underground pipeline detection due to its advantages such as non-destructive detection, rapid detection and high resolution. At present, ground penetrating radar is mainly used for pipeline location positioning, pipeline damage detection and pipeline surrounding environment survey. Traditional ground penetrating radar detection methods usually use a single-frequency radar system for detection, and detect underground pipelines and their surrounding environment by processing and analyzing radar echo signals. At the same time, with the development of computer technology, digital signal processing technology, image processing technology and artificial intelligence technology have also been gradually introduced into ground penetrating radar data processing, which has improved the accuracy and reliability of detection results.

[0003] However, the existing ground penetrating radar detection methods still have some shortcomings in assessing the state of soil around pipelines. First, it is difficult for a single-frequency radar system to take into account both detection depth and resolution, which affects the detection effect; second, traditional data processing methods often only focus on signal amplitude information, and do not fully utilize phase information and spectrum characteristics, resulting in low accuracy in identifying soil anomalies; third, existing anomaly assessment methods are mostly based on empirical judgment, lacking quantitative assessment standards, and it is difficult to objectively evaluate the safety state of soil around pipelines. In addition, due to the complexity of the underground environment, traditional methods also have obvious deficiencies in dealing with multi-scale feature fusion and uncertainty analysis. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the present application provides a method and system for detecting soil around pipelines based on ground penetrating radar. Based on the quantitative results of the soil dielectric parameters obtained, the accuracy of abnormal area identification is improved, so as to further improve the accuracy of the existing method for detecting soil around pipelines.

[0005] In the first aspect, the present application provides a method for detecting soil around a pipeline based on a ground penetrating radar, and the method for detecting soil around a pipeline based on a ground penetrating radar comprises: performing a multi-frequency ground penetrating radar scan on the area around the pipeline to obtain original radar signal data, and spatially positioning the original radar signal data to obtain positioning marked radar data; performing time zero point correction on the positioning marked radar data to obtain zero point correction data, and performing energy correction on the zero point correction data through an FK domain offset algorithm to obtain energy corrected radar data; performing hyperbola feature extraction on the energy corrected radar data to obtain pipeline feature data, and performing multi-scale feature fusion on the pipeline feature data The algorithm performs spatial positioning to obtain pipeline location characteristic data and soil layer distribution data; performs generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculates the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; extracts characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and performs anomaly identification on the soil characteristic vector data to obtain soil abnormal area data; performs fuzzy hierarchical analysis on the soil abnormal area data to obtain risk index data, and performs correlation analysis on the risk index data and the pipeline location characteristic data to obtain soil risk assessment data around the pipeline.

[0006] In a second aspect, the present application provides a pipeline surrounding soil detection device based on ground penetrating radar, and the pipeline surrounding soil detection device based on ground penetrating radar includes: The scanning module is used to perform a multi-frequency ground penetrating radar scan on the area surrounding the pipeline to obtain raw radar signal data, and to perform spatial positioning marking on the raw radar signal data to obtain positioning marking radar data; A correction module is used to perform time zero point correction on the positioning mark radar data to obtain zero point correction data, and perform energy correction on the zero point correction data through an FK domain migration algorithm to obtain energy correction radar data; An extraction module is used to perform hyperbolic feature extraction on the energy-corrected radar data to obtain pipeline feature data, and to perform spatial positioning on the pipeline feature data through a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data; A modeling module is used to perform generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculate the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; An identification module is used to extract characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and to identify abnormalities from the soil characteristic vector data to obtain soil abnormal area data; The analysis module is used to perform fuzzy hierarchical analysis on the soil abnormal area data to obtain risk index data, and to perform correlation analysis on the risk index data and the pipeline location feature data to obtain soil risk assessment data around the pipeline.

[0007] In the technical solution provided by the present application, by performing multi-frequency ground penetrating radar scanning on the area around the pipeline, a combination of antennas with three different frequencies of 200MHz, 400MHz and 900MHz is used to achieve collaborative detection at different depths, effectively improving the depth range and resolution of the detection. The spatial positioning marking technology of positioning and marking radar data, combined with the RTK-GPS real-time positioning system, enables the detection data to have accurate spatial position information, providing a reliable spatial reference for subsequent analysis. The FK domain offset algorithm is used to perform energy correction on the zero-point correction data, effectively eliminating the position offset of the inclined reflection surface and improving the imaging quality. The pipeline position is accurately located by hyperbolic feature extraction and multi-scale feature fusion algorithm, which not only ensures the accuracy of pipeline positioning, but also realizes the continuous tracking of soil layer distribution. The generalized reflection coefficient model and Bayesian inversion framework are used to invert the dielectric characteristics of the soil, and the quantitative results of the dielectric parameters of the soil are obtained, which improves the reliability of anomaly detection. The soil feature vector data is used for anomaly identification, and the accuracy of abnormal area identification is enhanced by combining multiple feature parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0009] Figure 1 This is a schematic diagram of an embodiment of a method for detecting soil around a pipeline based on a ground penetrating radar in an embodiment of the present application; Figure 2 It is a schematic diagram of an embodiment of a pipeline surrounding soil detection device based on ground penetrating radar in an embodiment of the present application. DETAILED DESCRIPTION

[0010] The embodiment of the present application provides a method and system for detecting soil around a pipeline based on ground penetrating radar. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0011] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiment of the present application, an embodiment of the method for detecting soil around a pipeline based on ground penetrating radar includes: Step S101, performing a multi-frequency ground penetrating radar scan on the area surrounding the pipeline to obtain original radar signal data, and performing spatial positioning marking on the original radar signal data to obtain positioning marking radar data; Step S102, performing time zero point correction on the positioning mark radar data to obtain zero point correction data, and performing energy correction on the zero point correction data through the FK domain migration algorithm to obtain energy correction radar data; Step S103, performing hyperbolic feature extraction on the energy-corrected radar data to obtain pipeline feature data, and spatially locating the pipeline feature data using a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data; Step S104, performing generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculating the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; Step S105, extracting characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and performing abnormality identification on the soil characteristic vector data to obtain soil abnormal area data; Step S106: perform fuzzy hierarchical analysis on the soil abnormal area data to obtain risk index data, and perform correlation analysis on the risk index data and pipeline location feature data to obtain soil risk assessment data around the pipeline.

[0012] It is understandable that the execution subject of the present application may be a pipe surrounding soil detection device based on ground penetrating radar, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0013] Specifically, a multi-frequency ground penetrating radar system is used to collect data in the area around the pipeline. By laying out survey lines at intervals of 0.5 meters on both sides of the pipeline, antennas with three frequencies of 200MHz, 400MHz and 900MHz are used for scanning. Among them, the 200MHz antenna is mainly used for deep detection, with a detection depth of up to 6 meters, the 400MHz antenna is used for mid-layer detection, with a depth range of 3-4 meters, and the 900MHz antenna is used for shallow high-precision detection, with a detection depth of 1-2 meters. During the data collection process, real-time positioning is performed through RTK-GPS, and the sampling interval is set to 5 cm. At the same time, environmental parameters such as soil moisture content and ground temperature are recorded for subsequent data correction. The collected raw radar data needs to be processed by time zero point correction. First, the DC component is removed to eliminate signal drift, and then the time reference is established through time window analysis to eliminate the influence of antenna coupling effect on the signal. After amplitude compensation and noise filtering, the FK domain offset algorithm is used to relocate the energy. The algorithm performs phase correction on the reflection signals at different depths through frequency-wavenumber domain transformation, which effectively improves the imaging quality of underground targets.

[0014] Hyperbolic features are extracted from the corrected radar data, and the position of the pipeline is determined by identifying the reflected hyperbolic features at the top of the pipeline. The coordinates of the waveform vertices are extracted during the calculation of the hyperbolic parameters, and the spatial positioning data of the pipeline is established by combining pipeline trajectory fitting and spatial curve smoothing. Multi-scale decomposition is performed using wavelet transform to achieve feature fusion of data of different frequencies, and soil layer distribution information is obtained through layer tracking technology. The generalized reflection coefficient modeling method is used to process soil layer distribution data. First, the propagation path and incident angle of electromagnetic waves in different soil layers are calculated, and the interface reflection coefficient is calculated through wave impedance analysis. Based on the Bayesian inversion framework, the prior probability distribution and likelihood function are established, and the dielectric characteristics of the soil are obtained through parameter estimation to achieve quantitative inversion of the dielectric parameters of the soil layer.

[0015] The waveform characteristics, spectrum characteristics, phase characteristics and attenuation characteristics of the acquired soil dielectric characteristic data are extracted, and the characteristic vector is constructed for abnormal identification. The abnormal threshold is determined by cluster analysis, and the spatial range and attribute labeling of the abnormal area are performed. Finally, risk assessment is performed based on the fuzzy hierarchical analysis method, and the judgment matrix is ​​established to calculate the weight index, and the risk zoning is performed in combination with the pipeline location characteristic data.

[0016] For example, in the soil detection around a gas pipeline, this method was used for detection and analysis. The pipeline burial depth was determined to be 4.2 meters through 200MHz antenna data, and the 400MHz and 900MHz antenna data focused on analyzing the soil state within 1-4 meters around the pipeline. During the data processing, the initial offset time was corrected from 12ns to 0ns through time zero correction, and the FK domain offset algorithm adjusted the spatial position of the reflected energy at the top of the pipeline upward by 0.3 meters. In the feature extraction stage, it was found that the soil dielectric constant at 2.5 meters to the right of the pipeline and 3.1 meters deep was abnormal. After wavelet multi-scale decomposition, the area showed obvious attenuation anomalies in the three frequency bands. The water content of the soil at this location reached 38% through Bayesian inversion, which was much higher than the 15-20% in the surrounding areas. The risk level of the abnormal area was determined to be medium risk through fuzzy hierarchical analysis, and continuous monitoring and early warning were required.

[0017] In the embodiment of the present application, by performing multi-frequency ground penetrating radar scanning on the area around the pipeline, a combination of antennas with three different frequencies of 200MHz, 400MHz and 900MHz is used to achieve collaborative detection at different depths, effectively improving the depth range and resolution of the detection. The spatial positioning marking technology of positioning and marking radar data, combined with the RTK-GPS real-time positioning system, enables the detection data to have accurate spatial position information, providing a reliable spatial reference for subsequent analysis. The FK domain offset algorithm is used to perform energy correction on the zero-point correction data, effectively eliminating the position offset of the inclined reflection surface and improving the imaging quality. The pipeline position is accurately located by hyperbolic feature extraction and multi-scale feature fusion algorithm, which not only ensures the accuracy of pipeline positioning, but also realizes the continuous tracking of soil layer distribution. The generalized reflection coefficient model and Bayesian inversion framework are used to invert the dielectric characteristics of the soil, and the quantitative results of the dielectric parameters of the soil are obtained, which improves the reliability of anomaly detection. The soil feature vector data is used for abnormal identification, and the accuracy of abnormal area identification is enhanced by combining multiple feature parameters.

[0018] In a specific embodiment, the process of executing step S101 may specifically include the following steps: (1) Lay out survey lines at intervals of 0.5 meters in the pipeline area to obtain pipeline inspection survey line data, and collect real-time coordinates of the pipeline inspection survey line data to obtain survey line coordinate data; (2) Scan the survey line coordinate data using three-band radars at 200 MHz, 400 MHz, and 900 MHz to obtain multi-band radar echo data, and perform trigger sampling on the multi-band radar echo data at 5 cm intervals to obtain radar sampling data; (3) Perform ground temperature compensation on the radar sampling data to obtain temperature correction data, and perform soil moisture compensation on the temperature correction data to obtain environmental correction data; (4) Depth calibration is performed on the environmental correction data to obtain depth marking data, and pipeline axial positioning marking is performed on the depth marking data to obtain axial positioning data; (5) Calculate the lateral displacement of the axial positioning data to obtain lateral offset data, and perform coordinate mapping on the lateral offset data to obtain spatial coordinate data; (6) Performing data association on the spatial coordinate data and the environmental correction data to obtain associated marker data, and performing spatial interpolation processing on the associated marker data to obtain positioning marker radar data.

[0019] Specifically, in the process of detecting the soil around underground pipelines, the layout of survey lines is the primary link of data collection. First, with the center line of the pipeline as the reference, parallel survey lines are laid on both sides of the pipeline, with a survey line interval of 0.5 meters. The layout range is determined according to the buried depth of the pipeline, usually 3-5 meters on both sides of the pipeline. After the survey line is laid, the RTK-GPS receiver is used to collect real-time coordinates of each survey line to obtain the spatial position information of the survey line. The coordinate accuracy is better than 2 cm, and the coordinates of the starting and ending points and feature points of the survey line are recorded at the same time.

[0020] Based on the survey line coordinate data, a ground penetrating radar device equipped with three frequency antennas is used for scanning. Among them, the 200MHz antenna is mainly used for deep detection, with a detection depth of up to 6 meters, mainly to obtain the deep soil information under the pipeline; the 400MHz antenna is used for mid-layer detection, with a detection depth of 3-4 meters, focusing on obtaining the soil state around the pipeline; the 900MHz antenna is used for shallow detection, with a detection depth of 1-2 meters, to obtain high-precision shallow soil information. During the radar scanning process, the equidistant trigger mode is adopted, and the trigger interval is set to 5 cm to ensure the continuity and integrity of data collection. Environmental parameter compensation is performed on the collected radar sampling data. First, temperature compensation is performed based on the soil temperature data measured by the geothermal sensor to eliminate the influence of temperature changes on the radar signal. Then, based on the moisture content data measured by the soil moisture sensor, the radar signal is compensated for moisture content to establish environmental correction data. In the compensation process, a layered compensation method is adopted, and different compensation parameters are used for soil layers of different depths.

[0021] The environmental correction data is processed by depth calibration to establish the corresponding relationship between the radar signal time and the actual depth. The depth reference point is obtained by burying the calibration body at a known position, and the depth value of each sampling point is determined in combination with the propagation speed of electromagnetic waves in the soil. The pipeline axial positioning mark is then used to mark the depth mark data, and the intersection position of the survey line and the pipeline is determined according to the direction of the pipeline to establish the axial positioning benchmark. The axial positioning data needs to be calculated for lateral displacement, considering the relative position relationship between the survey line and the pipeline, and calculating the lateral offset distance of each measuring point relative to the center line of the pipeline. The measuring point data in the local coordinate system is mapped to a unified spatial coordinate system through coordinate transformation to establish complete three-dimensional spatial positioning data.

[0022] Finally, the spatial coordinate data and environmental correction data are correlated and analyzed to establish the corresponding relationship between the measuring point position and the radar signal. The Kriging spatial interpolation method is used to interpolate the discrete measuring point data to generate continuous positioning mark radar data.

[0023] For example, in the inspection of a certain section of direct buried gas pipeline, the pipeline is buried about 3 meters deep, and 6 survey lines are laid on both sides of the pipeline, with a total length of 100 meters. RTK-GPS is used to collect survey line coordinates, with an average coordinate accuracy of 1.5 cm. Three frequencies of radar antennas collect a total of 20,000 measurement point data at 5 cm intervals. Environmental parameters show that the soil temperature is 22°C and the average moisture content is 18%. After environmental compensation and depth calibration, it is determined that the propagation speed of electromagnetic waves in the soil is 0.09 m / ns. Through coordinate transformation and spatial interpolation, a three-dimensional detection data volume with a resolution of 0.1 m × 0.1 m × 0.05 m is generated.

[0024] In a specific embodiment, the process of executing step S102 may specifically include the following steps: (1) Remove the DC component of the positioning marker radar data to obtain the de-drift radar data, and perform time window analysis on the de-drift radar data to obtain the time reference data; (2) Eliminate the antenna coupling effect of the time reference data to obtain coupling correction data, and calculate the time zero point offset of the coupling correction data to obtain zero point position data; (3) Performing time axis calibration on the zero point position data to obtain calibration time data, and performing amplitude compensation on the calibration time data to obtain amplitude balanced data; (4) Perform high-frequency noise filtering on the amplitude equalization data to obtain noise-reduced radar data, and perform low-frequency interference elimination on the noise-reduced radar data to obtain zero-point correction data; (5) performing FK domain transformation on the zero-point correction data to obtain frequency-wavenumber domain data, and performing phase correction on the frequency-wavenumber domain data to obtain phase-corrected data; (6) The phase-corrected data is energy relocated using the FK domain migration algorithm to obtain the migration-corrected data, and the migration-corrected data is inversely transformed to obtain the energy-corrected radar data.

[0025] Specifically, the DC component of the positioning marker radar data is removed, and the DC drift of the signal is eliminated by subtracting the average value of each data. In the original radar data, the DC component manifests itself as an offset phenomenon of the overall signal. The signal after the DC component is removed is more suitable for subsequent processing. The data after drift removal is analyzed in time window. According to the detection depth of antennas with different frequencies, a 500ns time window is set for the 200MHz antenna, a 300ns time window is set for the 400MHz antenna, and a 150ns time window is set for the 900MHz antenna to establish the time reference data. The elimination of antenna coupling effect is an important signal processing step. Direct coupling waves and air coupling waves will interfere with the target reflection signal. By identifying and extracting the coupling waveform at the front end of the time reference data, the coupling effect is removed by using the subtraction elimination method to obtain the coupling correction data. Subsequently, the time zero offset of the coupling correction data is calculated, and the actual time zero point is determined by identifying the position of the surface reflection wave, and the establishment of the zero point position data is completed.

[0026] The zero-point position data is processed by time axis calibration to align the time axes of different channels and eliminate the slight differences in the sampling triggering time. The calibrated time data is amplitude compensated, and the segmented gain control method is used to perform different degrees of gain compensation on signals in different depth ranges to compensate for the attenuation of deep signals and obtain amplitude-equalized radar data. The amplitude-equalized data is processed by high-frequency noise filtering, and a bandpass filter design is used. A 100-300MHz filter is used for 200MHz antenna data, a 300-500MHz filter is used for 400MHz antenna data, and a 700-1100MHz filter is used for 900MHz antenna data to obtain noise-reduced radar data. Subsequently, low-frequency interference elimination is performed to remove 50Hz power frequency interference and slowly varying interference signals, and zero-point correction data is established.

[0027] The zero-point correction data is converted to the FK domain (frequency-wavenumber domain) through Fourier transform to obtain frequency-wavenumber domain data. Phase correction is performed in the FK domain to correct the phase distortion caused by surface unevenness and antenna coupling to obtain phase correction data. The phase correction data is energy relocated through the FK domain offset algorithm to focus the energy of the inclined reflector to the true position, and finally the data is converted back to the time domain through inverse transformation processing to obtain energy correction radar data.

[0028] For example: In a certain section of pipeline detection, the raw radar data collected by the 400MHz antenna is processed. First, the DC component is removed, and the average amplitude of the raw data is reduced from 156mv to nearly 0mv. The time window is set to 300ns, and the number of sampling points is 512. By eliminating the coupling effect, the amplitude of the coupled wave in the first 50ns is reduced from the original 800mv to below 100mv. The time zero position is adjusted from the original 12ns to 0ns, and the signal strength at a depth of 4 meters after amplitude compensation is increased from the original 50mv to 200mv. After bandpass filtering, the signal-to-noise ratio is increased from the initial 6dB to 15dB. FK domain processing successfully focuses the energy of the 45-degree inclined reflection surface, and finally forms a clear reflection waveform at the pipeline position. The reflection wave amplitude reaches 350mv and the time resolution is better than 1ns.

[0029] In a specific embodiment, the process of executing step S103 may specifically include the following steps: (1) Calculate the hyperbola parameters on the energy-corrected radar data to obtain the hyperbola equation data, and extract the vertices on the hyperbola equation data to obtain the vertex coordinate data; (2) Fitting the vertex coordinate data to the pipeline trajectory to obtain pipeline trajectory data, and performing spatial curve smoothing on the pipeline trajectory data to obtain smooth trajectory data; (3) Perform multi-scale decomposition of the smooth trajectory data through wavelet transform to obtain scale decomposition data, and extract feature points from the scale decomposition data to obtain feature point data; (4) Performing feature fusion processing on the feature point data to obtain fused feature data, and performing spatial coordinate mapping on the fused feature data to obtain pipeline position feature data; (5) Calibrate the pipeline location characteristic data with reference to the layer to obtain the layer benchmark data, and track the soil layer interface of the layer benchmark data to obtain the soil layer distribution data.

[0030] Specifically, the hyperbolic parameters of the energy-corrected radar data are calculated, and the hyperbolic features in the echo data are identified by the Hough transform method. For a cylindrical pipe, its radar reflection waveform presents a standard hyperbola on the time-position profile, and the hyperbolic equation is calculated as follows: ; in, are the coordinates of the hyperbola vertices, and is the shape parameter. The hyperbola vertices are extracted by extreme point detection to obtain the initial coordinates of the pipeline position. The pipeline trajectory is fitted to the acquired vertex coordinate data, and the spatial curve equation is established using the cubic spline interpolation method. : ; in is the cubic B-spline basis function, The fitted pipeline trajectory data is smoothed, and the Savitzky-Golay filter is used to eliminate the local fluctuation of the trajectory curve. The filter window length is set according to the change of pipeline curvature. The smoothed trajectory data is more suitable for subsequent multi-scale analysis.

[0031] The smooth trajectory data is decomposed at multiple scales using wavelet transform, and the Daubechies wavelet basis function is selected. The decomposition scale is set to 4 layers, corresponding to the characteristic scales of antenna data at different frequencies. The wavelet transform decomposes the signal into different frequency components. The high-frequency coefficient reflects the local detail characteristics, and the low-frequency coefficient reflects the overall contour characteristics. Feature points are extracted from wavelet coefficients of different scales, including local extreme points, inflection points, and mutation points. The feature point data is fused and the weighted average method is used to combine the feature information of data at different frequencies. The weight coefficient is dynamically adjusted according to the signal-to-noise ratio and the detection depth. The shallow layer is mainly based on 900MHz antenna data, and the deep layer is mainly based on 200MHz antenna data. The fused feature data is mapped to a unified spatial coordinate system through coordinate transformation to establish the precise spatial positioning data of the pipeline.

[0032] The reference layer is calibrated based on the pipeline location feature data, and the reflection at the top of the pipeline is used as the reference layer, and multiple reference layers are established by extending upward and downward. The phase tracking technology is used to track the soil layer interface of the layer reference data to achieve continuous identification of the soil layer interface, and finally obtain complete soil layer distribution data.

[0033] For example, in the detection of a gas pipeline section, the top position of the pipeline was identified by calculating the hyperbolic parameters. The coordinates of the hyperbolic vertex were located at (x=15.3m, t=68ns), corresponding to an actual depth of 3.2 meters. The pipeline trajectory was fitted using 50 control points, and the Savitzky-Golay filter window length was set to 11 points. After wavelet decomposition, obvious soil interface features were found in the coefficients of the second and third layers. Through feature fusion, it was determined that there were two main soil interfaces at 2.1 meters and 2.8 meters above the pipeline. Soil interface tracking showed that the thickness of the upper soil layer was 0.7 meters, the thickness of the middle soil layer was 1.1 meters, and the thickness of the soil layer around the pipeline was 1.4 meters, forming a complete soil layer distribution structure. When fusion was performed, the weight of 900MHz data was 0.6, the weight of 400MHz data was 0.3, and the weight of 200MHz data was 0.1. The final positioning accuracy was better than 5 cm.

[0034] In a specific embodiment, the process of executing step S104 may specifically include the following steps: (1) Calculate the electromagnetic wave propagation path of the soil layer distribution data to obtain the wave propagation path data, and analyze the incident angle of the wave propagation path data to obtain the incident angle data; (2) Calculate the wave impedance of the incident angle data to obtain wave impedance data, and calculate the interface reflection coefficient of the wave impedance data to obtain reflection coefficient data; (3) performing amplitude attenuation analysis on the reflection coefficient data to obtain attenuation coefficient data, and generating reflection coefficient model data on the attenuation coefficient data to obtain reflection coefficient model data; (4) Calculating the prior probability distribution of the reflection coefficient model data to obtain prior distribution data, and constructing a likelihood function for the prior distribution data to obtain likelihood function data; (5) The likelihood function data is estimated through the Bayesian inversion framework to obtain the posterior distribution data, and the maximum probability solution is extracted from the posterior distribution data to obtain the soil dielectric characteristic data.

[0035] Specifically, the electromagnetic wave propagation characteristics are analyzed based on the soil layer distribution data. First, the propagation path of the electromagnetic wave in the layered medium is calculated. The propagation path calculation is based on the Fermat principle, and the electromagnetic wave propagation time satisfies the equation: ; in is the propagation distance in the i-th layer, is the propagation speed in this layer, is the number of soil layers. The wave path is determined by minimizing the propagation time, and then the incident angle of each interface is calculated. The incident angle is calculated using Snell's law: ; in , are the angles of incidence and refraction, , is the wave velocity in the adjacent layer medium.

[0036] The wave impedance is calculated based on the incident angle data and is expressed as: ; in is the vacuum permeability, is the angular frequency, is the complex propagation constant of the qth layer. The interface reflection coefficient is calculated using wave impedance data. The calculation of the reflection coefficient takes into account the superposition effect of multiple layers of media, and is calculated layer by layer from deep to shallow through a recursive algorithm. The reflection coefficient data is subjected to an amplitude attenuation analysis, and the attenuation coefficient data is established by considering the two main factors of geometric diffusion loss and dielectric absorption loss. The generation of reflection coefficient model data is based on the generalized reflection coefficient theory, taking into account the influence of multiple reflections and transmission. By establishing a priori probability distribution, physical constraints on soil parameters are introduced, such as the range of dielectric constants and spatial continuity characteristics. The construction of the likelihood function is based on the degree of fit between the observed data and the forward model, and the Gaussian error model is used to describe the observation noise.

[0037] Finally, the parameters are estimated through the Bayesian inversion framework, and the posterior probability density function is sampled through the Markov chain Monte Carlo method. The maximum probability solution is extracted from the posterior distribution as the optimal estimate of the dielectric characteristics of the soil. For example: in a certain section of pipeline detection, the soil layer is divided into three layers, the first layer has a depth of 0-1.5 meters, the second layer has a depth of 1.5-3.0 meters, and the third layer has a depth of 3.0-4.5 meters. The Fermat principle is used to calculate that the incident angle of the electromagnetic wave on the first interface is 32 degrees, and the incident angle of the second interface is 28 degrees. The wave impedance calculation shows that the wave impedance values ​​of the three layers of soil are 265Ω, 312Ω and 378Ω, respectively, and the interface reflection coefficients are 0.08 and 0.10, respectively. The attenuation coefficient analysis shows that the signal attenuates by about 6dB per meter during propagation. After Bayesian inversion, the relative dielectric constants of the three layers of soil are 4.2±0.3, 6.8±0.4 and 8.5±0.5, respectively, where the error range represents the standard deviation of the posterior distribution. The maximum probability solution was extracted using 100,000 MCMC iterations, and the convergence criterion was set to a parameter variation coefficient of less than 1%. The inversion results showed that the moisture content of the third layer of soil was significantly higher than that of the upper layer, which was consistent with the actual drilling verification results.

[0038] In a specific embodiment, the process of executing step S105 may specifically include the following steps: (1) Performing waveform feature extraction on the soil dielectric characteristic data to obtain waveform characteristic data, and performing spectrum analysis on the waveform characteristic data to obtain spectrum characteristic data; (2) extracting phase features from the frequency spectrum feature data to obtain phase feature data, and calculating attenuation features from the phase feature data to obtain attenuation feature data; (3) Performing feature vector combination on the attenuation feature data to obtain feature vector data, and performing cluster analysis on the feature vector data to obtain cluster center data; (4) Determine the abnormal threshold of the cluster center data to obtain threshold judgment data, and identify the abnormal area of ​​the threshold judgment data to obtain abnormal identification data; (5) The spatial scope of the abnormal identification data is defined to obtain the abnormal range data, and the attributes of the abnormal range data are annotated to obtain the soil abnormal area data.

[0039] Specifically, the analysis is carried out through three main aspects: amplitude characteristics, waveform symmetry and waveform width. The amplitude characteristics include peak value, mean value and variance. The waveform symmetry is described by skewness and kurtosis. The waveform width is calculated by the half-peak full width. The obtained waveform characteristic data is subjected to spectrum analysis. The short-time Fourier transform method is used. The window width is set to 128 sampling points and the overlap rate is 50% to obtain the time-frequency spectrum characteristic data. The phase characteristics are extracted from the spectrum characteristic data, and the instantaneous phase and phase difference are calculated, mainly including two indicators: phase change rate and phase continuity. The phase change rate reflects the speed of change of the dielectric properties of the soil, and the phase continuity characterizes the uniformity of the soil structure. The attenuation characteristics are analyzed in combination with the phase characteristics, and the frequency-related attenuation coefficient and the law of attenuation change with depth are calculated to establish the attenuation characteristic data.

[0040] The attenuation feature data is combined with the previously extracted features to construct a 35-dimensional feature vector, including 15 dimensions of waveform features, 10 dimensions of spectrum features, 5 dimensions of phase features, and 5 dimensions of attenuation features. The K-means clustering algorithm is used to classify the feature vector data. The number of clusters is determined by the silhouette coefficient and the interval index, which is generally set to 3-5 categories to obtain cluster centers representing different soil states. Based on the cluster center data, anomaly judgment criteria are established. The Mahalanobis distance is used to calculate the distance between the measuring point and the center of each category. The confidence level is set to 95% and the anomaly threshold is calculated. The areas exceeding the threshold are marked to form preliminary anomaly identification data. The scattered anomaly points are connected and filtered through morphological processing, and isolated anomaly areas with an area of ​​less than 50 sampling points are eliminated.

[0041] The spatial scope of the screened abnormal area is defined to determine the depth range and horizontal extension range of the abnormal area. At the same time, the attributes of the abnormal area are annotated based on the main manifestations of the abnormal characteristics, including water-rich anomalies, loose anomalies, and void anomalies.

[0042] For example: In a pipeline detection section, the abnormal area at a depth of 3.2 meters was analyzed. The waveform characteristics showed that the amplitude value there was 85% higher than the surrounding area, and the waveform width increased by 45%. Spectral analysis showed that the energy attenuation in the 200-300MHz frequency band was abnormal, and the attenuation rate reached 12dB / m, which was much higher than the 6dB / m in the normal area. Phase analysis found that the phase change rate reached 0.8rad / ns in the abnormal area, while it was only 0.3rad / ns in the normal area. K-means clustering analysis classified the area as a high attenuation category, and the Mahalanobis distance calculation result was 2.8, which exceeded the set threshold of 2.2. After spatial scope definition, the horizontal range of the abnormal area was 2.8-4.2 meters, the vertical range was 3.0-3.5 meters, and the area was about 1.4 square meters. According to the results of the feature combination analysis, the area was marked as a water-rich anomaly and required key monitoring.

[0043] In a specific embodiment, the process of executing step S106 may specifically include the following steps: (1) The abnormality grade of the soil abnormal area data is classified to obtain grade evaluation data, and the grade evaluation data is hierarchically constructed to obtain hierarchical structure data; (2) Construct a judgment matrix for the hierarchical data to obtain judgment matrix data, and perform weight calculation on the judgment matrix data to obtain weight index data; (3) Perform consistency check on the weight index data to obtain consistency data, and perform fuzzy evaluation on the consistency data to obtain risk index data; (4) Spatial mapping of risk indicator data and pipeline location feature data to obtain location-related data, and risk area division of the location-related data to obtain risk zoning data; (5) Determine the risk level of the risk zoning data to obtain risk level data, and conduct a safety assessment on the risk level data to obtain risk assessment data of the soil around the pipeline.

[0044] Specifically, the abnormality is graded according to its type, scope and degree. The abnormality level includes three levels: slight abnormality, moderate abnormality and severe abnormality. Each level is subdivided according to the quantitative indicators of abnormal characteristics. The grade assessment data is based on this to establish a three-layer hierarchy: the first layer is the overall risk target, the second layer is the evaluation criteria (including three factors: abnormality range, abnormality type and abnormality degree), and the third layer is the specific evaluation indicators (including horizontal range, vertical range, abnormality intensity, persistence, etc.).

[0045] When the hierarchical data is converted into a judgment matrix, the 1-9 scale method is used to construct the judgment matrix to express the relative importance of each factor. The judgment matrix satisfies the formula: ; in For elements Relative to The importance ratio of , is the importance of each factor. The judgment matrix is ​​normalized and the weight is calculated using the characteristic root method to satisfy the characteristic equation: ; in, is the judgment matrix, is the weight vector, is the largest characteristic root. Perform consistency check on the weight index data and calculate the consistency ratio: ; in, , is the consistency index, is the random consistency indicator, is the matrix order. After the inspection is passed, the fuzzy comprehensive evaluation method is used to calculate the risk index, and the membership function is established to describe the fuzziness of each evaluation index. The risk index data is spatially associated with the pipeline location characteristic data to establish a risk impact area centered on the pipeline. According to the distance attenuation principle, the risk area is divided into the immediate area (0-1 meter), the near area (1-2 meters) and the far area (2-3 meters). Each area determines the risk partition according to the risk index value and spatial position to form the risk partition data.

[0046] The risk zoning data is comprehensively analyzed to determine the final risk level. The risk level is divided into four levels: safe (green), alert (yellow), dangerous (orange) and emergency (red). According to the risk level data, the corresponding safety assessment conclusions and prevention and control recommendations are formulated.

[0047] For example, an abnormal area was found in a section of pipeline detection, located 1.8 meters to the right of the pipeline and 3.2 meters deep. The abnormal range extended 1.4 meters horizontally and 0.5 meters vertically. Through hierarchical analysis, an evaluation system consisting of 3 criterion layers and 8 indicator layers was established. The judgment matrix calculation obtained the weight of the abnormal range of 0.35, the weight of the abnormal type of 0.40, and the weight of the abnormal degree of 0.25. The consistency ratio CR=0.046, which is less than 0.1, meets the consistency requirements. The risk index value of the abnormal area obtained by fuzzy evaluation is 0.72. Considering that the anomaly is located in the near area and is in the key part of the pipeline operation, the area is determined as an orange risk area through spatial correlation analysis and risk zoning, which needs to be monitored once a week and an emergency plan is prepared. The risk assessment data shows that the soil safety in this section reaches an 85% confidence level, and it is recommended to strengthen the inspection and monitoring frequency.

[0048] The above describes the pipeline surrounding soil detection method based on ground penetrating radar in the embodiment of the present application. The following describes the pipeline surrounding soil detection device based on ground penetrating radar in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of a pipeline surrounding soil detection device based on ground penetrating radar includes: The scanning module 201 is used to perform a multi-frequency ground penetrating radar scan on the area surrounding the pipeline to obtain raw radar signal data, and to perform spatial positioning marking on the raw radar signal data to obtain positioning marking radar data; The correction module 202 is used to perform time zero point correction on the positioning mark radar data to obtain zero point correction data, and perform energy correction on the zero point correction data through the FK domain migration algorithm to obtain energy correction radar data; The extraction module 203 is used to perform hyperbolic feature extraction on the energy-corrected radar data to obtain pipeline feature data, and to perform spatial positioning on the pipeline feature data through a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data; A modeling module 204 is used to perform generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculate the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; The identification module 205 is used to extract characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and to perform abnormality identification on the soil characteristic vector data to obtain soil abnormal area data; The analysis module 206 is used to perform fuzzy hierarchy analysis on the soil abnormal area data to obtain risk index data, and to perform correlation analysis on the risk index data and the pipeline location feature data to obtain soil risk assessment data around the pipeline.

[0049] Through the cooperation of the above components, by scanning the surrounding area of ​​the pipeline with multi-frequency ground penetrating radar, using antenna combinations of 200MHz, 400MHz and 900MHz, collaborative detection at different depths is achieved, effectively improving the depth range and resolution of detection. The spatial positioning and marking technology of positioning and marking radar data, combined with the RTK-GPS real-time positioning system, enables the detection data to have accurate spatial position information, providing a reliable spatial reference for subsequent analysis. The FK domain offset algorithm is used to perform energy correction on the zero-point correction data, effectively eliminating the position offset of the inclined reflection surface and improving the imaging quality. The pipeline position is accurately located by hyperbolic feature extraction and multi-scale feature fusion algorithm, which not only ensures the accuracy of pipeline positioning, but also realizes the continuous tracking of soil layer distribution. The generalized reflection coefficient model and Bayesian inversion framework are used to invert the dielectric characteristics of the soil, and the quantitative results of the dielectric parameters of the soil are obtained, which improves the reliability of anomaly detection. The soil feature vector data is used for anomaly recognition, and the accuracy of abnormal area recognition is enhanced by combining multiple feature parameters.

[0050] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting soil around a pipeline based on ground penetrating radar, characterized in that: The pipeline surrounding soil detection method based on ground penetrating radar includes: Perform multi-frequency ground penetrating radar scanning on the area around the pipeline to obtain raw radar signal data, and perform spatial positioning marking on the raw radar signal data to obtain positioning marking radar data; Performing time zero point correction on the positioning mark radar data to obtain zero point correction data, and performing energy correction on the zero point correction data through a FK domain migration algorithm to obtain energy correction radar data; Performing hyperbolic feature extraction on the energy-corrected radar data to obtain pipeline feature data, and spatially locating the pipeline feature data using a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data; Performing generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculating the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; Extracting characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and identifying anomalies from the soil characteristic vector data to obtain soil abnormal area data; Perform fuzzy hierarchical analysis on the soil abnormal area data to obtain risk index data, and perform correlation analysis on the risk index data and the pipeline location feature data to obtain soil risk assessment data around the pipeline.

2. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The method of scanning the surrounding area of ​​the pipeline with a multi-frequency ground penetrating radar to obtain raw radar signal data and spatially marking the raw radar signal data to obtain positioning mark radar data includes: Laying out survey lines at intervals of 0.5 meters in the pipeline area to obtain pipeline detection survey line data, and collecting real-time coordinates of the pipeline detection survey line data to obtain survey line coordinate data; Performing a three-band radar scan of 200 MHz, 400 MHz and 900 MHz on the survey line coordinate data to obtain multi-band radar echo data, and performing 5 cm interval trigger sampling on the multi-band radar echo data to obtain radar sampling data; Performing ground temperature compensation processing on the radar sampling data to obtain temperature correction data, and performing soil moisture content compensation on the temperature correction data to obtain environment correction data; Performing depth calibration on the environmental correction data to obtain depth marking data, and performing pipeline axial positioning marking on the depth marking data to obtain axial positioning data; Performing lateral displacement calculation on the axial positioning data to obtain lateral offset data, and performing coordinate mapping on the lateral offset data to obtain spatial coordinate data; The spatial coordinate data and the environmental correction data are data associated to obtain associated mark data, and the associated mark data are spatially interpolated to obtain positioning mark radar data.

3. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The method of performing time zero point correction on the positioning mark radar data to obtain zero point correction data, and performing energy correction on the zero point correction data through a FK domain offset algorithm to obtain energy correction radar data, includes: removing a DC component from the positioning marker radar data to obtain de-drift radar data, and performing a time window analysis on the de-drift radar data to obtain time reference data; Eliminating antenna coupling effect on the time reference data to obtain coupling correction data, and performing time zero point offset calculation on the coupling correction data to obtain zero point position data; Performing time axis calibration on the zero point position data to obtain calibration time data, and performing amplitude compensation on the calibration time data to obtain amplitude balanced data; Performing high-frequency noise filtering on the amplitude equalization data to obtain noise reduction radar data, and performing low-frequency interference elimination on the noise reduction radar data to obtain zero-point correction data; Performing FK domain transformation on the zero-point correction data to obtain frequency-wavenumber domain data, and performing phase correction on the frequency-wavenumber domain data to obtain phase correction data; The phase correction data is energy relocated by using an FK domain offset algorithm to obtain offset correction data, and the offset correction data is inversely transformed to obtain energy correction radar data.

4. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The method of extracting hyperbolic features from the energy-corrected radar data to obtain pipeline feature data, and spatially locating the pipeline feature data using a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data includes: Performing hyperbola parameter calculation on the energy-corrected radar data to obtain hyperbola equation data, and performing vertex extraction on the hyperbola equation data to obtain vertex coordinate data; Performing pipeline trajectory fitting on the vertex coordinate data to obtain pipeline trajectory data, and performing spatial curve smoothing on the pipeline trajectory data to obtain smooth trajectory data; Performing multi-scale decomposition on the smooth trajectory data through wavelet transform to obtain scale decomposition data, and performing feature point extraction on the scale decomposition data to obtain feature point data; Performing feature fusion processing on the feature point data to obtain fused feature data, and performing spatial coordinate mapping on the fused feature data to obtain pipeline position feature data; The pipeline position characteristic data is calibrated with reference to the layer to obtain layer reference data, and the layer interface is tracked on the layer reference data to obtain soil layer distribution data.

5. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The generalized reflection coefficient modeling is performed on the soil layer distribution data to obtain reflection coefficient model data, and the reflection coefficient model data is calculated through a Bayesian inversion framework to obtain soil dielectric characteristic data, including: Performing electromagnetic wave propagation path calculation on the soil layer distribution data to obtain wave propagation path data, and performing incident angle analysis on the wave propagation path data to obtain incident angle data; Performing wave impedance calculation on the incident angle data to obtain wave impedance data, and performing interface reflection coefficient calculation on the wave impedance data to obtain reflection coefficient data; Performing amplitude attenuation analysis on the reflection coefficient data to obtain attenuation coefficient data, and generating reflection coefficient model data on the attenuation coefficient data to obtain reflection coefficient model data; Performing a priori probability distribution calculation on the reflection coefficient model data to obtain priori distribution data, and constructing a likelihood function on the priori distribution data to obtain likelihood function data; The likelihood function data are parameter estimated through a Bayesian inversion framework to obtain posterior distribution data, and the posterior distribution data are subjected to maximum probability solution extraction to obtain soil dielectric characteristic data.

6. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The step of extracting characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and performing abnormality identification on the soil characteristic vector data to obtain soil abnormal area data includes: Performing waveform feature extraction on the soil dielectric feature data to obtain waveform feature data, and performing spectrum analysis on the waveform feature data to obtain spectrum feature data; Performing phase feature extraction on the frequency spectrum feature data to obtain phase feature data, and performing attenuation feature calculation on the phase feature data to obtain attenuation feature data; Performing feature vector combination on the attenuation feature data to obtain feature vector data, and performing cluster analysis on the feature vector data to obtain cluster center data; Determining an abnormal threshold value for the cluster center data to obtain threshold judgment data, and marking an abnormal region for the threshold judgment data to obtain abnormal marking data; The abnormal identification data is spatially scoped to obtain abnormal scope data, and the abnormal scope data is attributed to obtain soil abnormal area data.

7. The method for detecting soil around a pipeline based on ground penetrating radar according to claim 1, characterized in that: The fuzzy hierarchical analysis is performed on the soil abnormal area data to obtain risk index data, and the risk index data and the pipeline location feature data are correlated and analyzed to obtain soil risk assessment data around the pipeline, including: Performing abnormal grade classification on the soil abnormal area data to obtain grade evaluation data, and constructing a hierarchical structure on the grade evaluation data to obtain hierarchical structure data; Constructing a judgment matrix for the hierarchical structure data to obtain judgment matrix data, and performing weight calculation on the judgment matrix data to obtain weight index data; Performing consistency check on the weight index data to obtain consistency data, and performing fuzzy evaluation on the consistency data to obtain risk index data; Performing spatial mapping on the risk indicator data and the pipeline location characteristic data to obtain location association data, and dividing the location association data into risk areas to obtain risk partition data; The risk level of the risk zoning data is determined to obtain risk level data, and a safety assessment is performed on the risk level data to obtain risk assessment data of soil around the pipeline.

8. A pipeline surrounding soil detection device based on ground penetrating radar, used to implement the pipeline surrounding soil detection method based on ground penetrating radar as described in any one of claims 1 to 7, characterized in that: The pipeline surrounding soil detection device based on ground penetrating radar includes: The scanning module is used to perform a multi-frequency ground penetrating radar scan on the area surrounding the pipeline to obtain raw radar signal data, and to perform spatial positioning marking on the raw radar signal data to obtain positioning marking radar data; A correction module is used to perform time zero point correction on the positioning mark radar data to obtain zero point correction data, and perform energy correction on the zero point correction data through an FK domain migration algorithm to obtain energy correction radar data; An extraction module is used to perform hyperbolic feature extraction on the energy-corrected radar data to obtain pipeline feature data, and to perform spatial positioning on the pipeline feature data through a multi-scale feature fusion algorithm to obtain pipeline position feature data and soil layer distribution data; A modeling module is used to perform generalized reflection coefficient modeling on the soil layer distribution data to obtain reflection coefficient model data, and calculate the reflection coefficient model data through a Bayesian inversion framework to obtain soil dielectric characteristic data; An identification module is used to extract characteristic parameters from the soil dielectric characteristic data to obtain soil characteristic vector data, and to identify abnormalities from the soil characteristic vector data to obtain soil abnormal area data; The analysis module is used to perform fuzzy hierarchical analysis on the soil abnormal area data to obtain risk index data, and to perform correlation analysis on the risk index data and the pipeline location characteristic data to obtain soil risk assessment data around the pipeline.

Images