Rapid Detection Method for Formation Defect Risk Based on Ground Penetrating Radar

Abstract

This invention relates to the field of geological surveying technology, specifically to a rapid detection method for stratigraphic disease risk based on ground-penetrating radar (GPR). This method solves the technical problem in existing GPR techniques where the inaccurate depth positioning is caused by propagation parameter fluctuations due to the non-uniformity of the underground medium during stratigraphic disease detection. The method includes: acquiring reflected wave signals at each acquisition time using GPR; performing signal decomposition processing on the reflected wave signals to separate the intrinsic mode components characterizing dielectric noise, and extracting features from the intrinsic mode components to determine the dielectric noise intensity index; constructing a dielectric noise intensity sequence based on the dielectric noise intensity index and performing spatiotemporal feature analysis to determine the stratigraphic disease confidence level; constructing a depth correction weight based on the dielectric noise intensity index and the stratigraphic disease confidence level, combined with a pre-calibrated average dielectric constant of the stratigraphy, adaptively correcting the original stratigraphic disease depth, and outputting the corrected stratigraphic disease depth.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of geological survey, in particular to a stratum disease risk rapid detection method based on ground penetrating radar. BACKGROUND

[0002] With the development and utilization of urban underground space, stratum diseases such as road collapse and pipeline leakage of aging underground infrastructure have become an important hidden danger for urban development. The traditional drilling sampling method is low in efficiency and strong in destructiveness. With the development of electromagnetic wave detection theory, the stratum disease detection method based on ground penetrating radar can quickly survey a large area by using the propagation characteristics of electromagnetic waves in underground medium, realize the positioning and risk assessment of stratum diseases.

[0003] However, due to the high spatial heterogeneity and variability of different components in underground medium in terms of density and water content, the propagation parameters such as dielectric constant and conductivity of electromagnetic wave propagation medium fluctuate, thereby affecting the propagation speed of electromagnetic wave. Further, due to the difficulty in accurately calibrating the propagation speed of electromagnetic wave, the positioning of stratum disease is inaccurate, and the reliability of disease identification is reduced. SUMMARY

[0004] In order to solve the technical problem that the fluctuation of propagation parameters caused by the non-uniformity of underground medium in the stratum disease detection by ground penetrating radar makes the depth positioning inaccurate in the prior art, the purpose of the present application is to provide a stratum disease risk rapid detection method based on ground penetrating radar, and the technical solution adopted is as follows:

[0005] In the first aspect, a stratum disease risk rapid detection method based on ground penetrating radar is provided, comprising: collecting signals of stratum by ground penetrating radar to obtain reflected wave signals at each collection time; performing signal decomposition processing on the reflected wave signals to separate out intrinsic mode components representing dielectric noise, and performing feature extraction on the intrinsic mode components to determine dielectric noise intensity index; the dielectric noise intensity index is used to represent the dielectric noise intensity caused by the non-uniformity of underground medium; a dielectric noise intensity sequence is constructed according to the dielectric noise intensity index, and a stratum disease confidence is determined by analyzing the time and space features of the dielectric noise intensity sequence; the stratum disease confidence is used to represent the confidence degree of the existence of real stratum disease under noise interference; a depth correction weight is constructed according to the dielectric noise intensity index and the stratum disease confidence, and the original stratum disease depth measured is adaptively corrected according to the depth correction weight, and the corrected stratum disease depth is output.

[0006] Based on the technical scheme, in the stratum disease risk rapid detection method based on the ground penetrating radar provided by the application, the reflection wave signals of the stratum at each time are collected by the ground penetrating radar to obtain basic detection data, and then the reflection wave signals are subjected to signal decomposition processing, the intrinsic modal component is separated and the features are extracted, so that the dielectric noise intensity caused by the non-uniformity of the underground medium can be effectively quantified, and the problem that the dielectric noise caused by the non-uniformity of the underground medium is difficult to accurately characterize is solved. Then, the sequence is constructed based on the dielectric noise intensity index, and the spatial and temporal feature analysis is performed to determine the stratum disease confidence, so that the credibility of the real stratum disease under the noise interference can be determined, and the misjudgment caused by the confusion of the noise and the real disease signal is avoided. Finally, the original stratum disease depth is adaptively corrected by constructing the deep correction weight, so that the fluctuation error of the electromagnetic wave propagation parameter caused by the non-uniformity of the underground medium can be compensated, the accuracy of the stratum disease depth positioning can be significantly improved, and the corrected stratum disease depth can be output.

[0007] In combination with the first aspect, in a possible implementation manner, the method for determining the stratum disease confidence by performing the spatial and temporal feature analysis on the dielectric noise intensity sequence, specifically includes: determining a window kurtosis value by using a sliding window analysis on the dielectric noise intensity sequence; the window kurtosis value is used to represent the concentration degree of the burst peak noise event in the window; the dielectric noise intensity sequence is clustered to obtain a high noise area and a low noise area, and a noise area separation degree is determined according to the distribution of the high noise area and the low noise area; the noise area separation degree is used to represent the separation degree of the high noise area and the low noise area in the feature space; and the stratum disease confidence is determined according to the window kurtosis value and the noise area separation degree.

[0008] In combination with the first aspect, in a possible implementation manner, the method for determining the noise area separation degree according to the distribution of the high noise area and the low noise area, specifically includes: calculating the Euclidean distance between the cluster centers of the high noise area and the low noise area; calculating a first standard deviation of the data points in the high noise area and a second standard deviation of the data points in the low noise area; and determining the noise area separation degree according to the Euclidean distance, the first standard deviation and the second standard deviation.

[0009] In combination with the first aspect, in a possible implementation manner, the method for determining the stratum disease confidence according to the window kurtosis value and the noise area separation degree, specifically includes: performing accumulation calculation on the window kurtosis values corresponding to a plurality of windows to obtain a kurtosis accumulation value; the kurtosis accumulation value is used to comprehensively represent the cumulative intensity of the burst peak noise in the collection process; and the stratum disease confidence is determined according to the kurtosis accumulation value and the noise area separation degree.

[0010] In a possible implementation manner of the first aspect, the method for performing signal decomposition processing on the reflected wave signal to separate the intrinsic mode component representing the dielectric noise comprises: performing decomposition on the reflected wave signal using a preset signal decomposition algorithm to output an intrinsic mode component set; the intrinsic mode component set comprises a first intrinsic mode component of a high-frequency signal and a second intrinsic mode component of a low-frequency signal; and the second intrinsic mode component is selected from the intrinsic mode component set as the intrinsic mode component representing the dielectric noise.

[0011] In a possible implementation manner of the first aspect, the method for performing feature extraction on the intrinsic mode component to determine the dielectric noise intensity index comprises: performing instantaneous feature analysis on the second intrinsic mode component to obtain instantaneous amplitudes and instantaneous phases of each data point in the second intrinsic mode component; determining an instantaneous amplitude variation coefficient for representing a discrete degree of the instantaneous amplitudes according to the instantaneous amplitudes of the multiple data points in the second intrinsic mode component; determining an instantaneous phase maximum change value for representing an extreme change of the instantaneous phases according to the instantaneous phases of the multiple data points in the second intrinsic mode component; and determining the dielectric noise intensity index according to the instantaneous amplitude variation coefficient and the instantaneous phase maximum change value.

[0012] In a possible implementation manner of the first aspect, the method for constructing the depth correction weight according to the dielectric noise intensity index and the formation disease confidence, and in combination with the pre-labeled formation average dielectric constant comprises: determining a relative depth error caused by dielectric constant fluctuation according to the dielectric noise intensity index, the instantaneous amplitude variation coefficient, and the formation average dielectric constant; and determining the depth correction weight according to the relative depth error and the formation disease confidence; the depth correction weight is inversely correlated with the formation disease confidence, and is used to represent a correction strength for the original formation disease depth.

[0013] In a possible implementation manner of the first aspect, the method for performing self-adaptive correction on the measured original formation disease depth according to the depth correction weight comprises: inputting the depth correction weight into a preset function to obtain a scaling factor; the preset function is used to map the depth correction weight to a preset range, and the preset range ensures smooth transition of the scaling factor; and scaling the original formation disease depth according to the scaling factor to obtain a corrected formation disease depth.

[0014] In a possible implementation manner of the first aspect, the method for scaling the original formation disease depth according to the scaling factor comprises: when the scaling factor is greater than a preset threshold, amplifying the original formation disease depth according to the scaling factor to strengthen a real disease signal; and when the scaling factor is less than the preset threshold, reducing the original formation disease depth according to the scaling factor to compress a false alarm.

[0015] In conjunction with the first aspect above, in one possible implementation, the ground-penetrating radar is equipped with a ground-coupled antenna and is in close contact with the ground surface; the method for acquiring signals from the strata using the ground-penetrating radar and obtaining reflected wave signals at each acquisition moment specifically includes: driving the ground-penetrating radar to operate at a preset center frequency, while simultaneously driving the radar antenna to move at a constant speed along the ground surface of the area to be measured; automatically triggering a transmission and reception cycle at a preset acquisition frequency, acquiring and storing reflected wave signals at each acquisition moment.

[0016] Secondly, a ground-penetrating radar-based rapid detection device for geological defects risk is provided, comprising: a processor and a storage medium; the storage medium includes instructions, and the processor is used to execute the instructions to perform the actions described in the first aspect and any possible implementation thereof. This ground-penetrating radar-based rapid detection device for geological defects risk can be an electronic device or a chip within an electronic device.

[0017] Thirdly, a computer-readable storage medium is provided, in which instructions are stored, which, when executed on a ground-penetrating radar-based rapid detection device for geological defects, cause the ground-penetrating radar-based rapid detection device for geological defects to perform the actions described in the first aspect and any possible implementation thereof.

[0018] Fourthly, a computer program product containing instructions is provided, which, when run on a ground-penetrating radar-based rapid detection device for geological defects, causes the ground-penetrating radar-based rapid detection device for geological defects to perform the actions described in the first aspect and any possible implementation thereof.

[0019] The present invention has the following beneficial effects:

[0020] By acquiring ground-penetrating radar (GPR) to collect reflected wave signals from the strata at various times to obtain basic detection data, and then performing signal decomposition processing on the reflected wave signals to separate intrinsic mode components and extract features, the intensity of dielectric noise caused by subsurface medium inhomogeneity can be effectively quantified, solving the problem of difficulty in accurately characterizing dielectric noise due to subsurface medium inhomogeneity. Subsequently, a sequence is constructed based on the dielectric noise intensity index, and spatiotemporal feature analysis is performed to determine the confidence level of stratum defects, which can clearly define the credibility of the existence of real stratum defects under noise interference and avoid misjudgment caused by the confusion between noise and real defect signals. Finally, by constructing depth correction weights, the original stratum defect depth is adaptively corrected, which can compensate for the fluctuation error of electromagnetic wave propagation parameters caused by subsurface medium inhomogeneity, significantly improve the accuracy of stratum defect depth positioning, and output a precise corrected stratum defect depth. Attached Figure Description

[0021] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can be obtained without creative labor.

[0022] Figure 1 A method flow chart of a stratum disease risk rapid detection method based on ground penetrating radar provided for an embodiment of the present application;

[0023] Figure 2 A method flow chart of another stratum disease risk rapid detection method based on ground penetrating radar provided for an embodiment of the present application;

[0024] Figure 3 A method flow chart of another stratum disease risk rapid detection method based on ground penetrating radar provided for an embodiment of the present application;

[0025] Figure 4 A method flow chart of another stratum disease risk rapid detection method based on ground penetrating radar provided for an embodiment of the present application;

[0026] Figure 5 A hardware structure schematic diagram of a stratum disease risk rapid detection device based on ground penetrating radar provided for an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined purposes, the specific embodiments, structures, features and effects of a stratum disease risk rapid detection method based on ground penetrating radar according to the present application are described in detail as follows in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the present application belongs.

[0029] The specific scheme of a stratum disease risk rapid detection method based on ground penetrating radar provided by the present application is specifically described below in combination with the drawings.

[0030] Please refer to Figure 1It shows a method flow chart of a stratum disease risk rapid detection method based on ground penetrating radar provided by one embodiment of the application. The stratum disease risk rapid detection method based on ground penetrating radar comprises the following steps of:

[0031] S1, collecting signals of the stratum by the ground penetrating radar to obtain reflected wave signals at each collection time.

[0032] In some implementations, the ground penetrating radar can be driven to operate at a preset center frequency, and the radar antenna can be driven to move at a constant speed along the surface of the region to be detected. Then, the transmission and reception cycle is automatically triggered at a preset collection frequency to collect and store the reflected wave signals at each collection time. The ground penetrating radar is provided with a ground coupling antenna and is in close contact with the ground surface.

[0033] Specifically, the design of the ground coupling antenna is adapted to the stratum detection scene, which can reduce the energy loss of electromagnetic waves at the interface between the ground surface and the air. During the device deployment stage, the radar antenna needs to be in close contact with the ground surface of the region to be detected. During the contact process, it is necessary to ensure that the bottom surface of the antenna is not blocked by foreign matter and that the adhesion degree meets the preset requirements (such as the adhesion gap being less than a preset gap threshold of 10 cm) to ensure the ground coupling effect and avoid additional attenuation or waveform distortion of the radar signal in the early stage of propagation due to poor coupling, which affects the accuracy of subsequent signal analysis.

[0034] According to the stratum depth range and geological complexity of the region to be detected, the core collection parameters of the ground penetrating radar are set, including: determining the center frequency of the ground penetrating radar in combination with the size and detection depth requirement of the target disease to ensure that the electromagnetic wave can penetrate to the target stratum depth and has sufficient resolution to identify the disease characteristics. In this embodiment, the center frequency of the ground penetrating radar can be set to 900Mhz (megahertz). The signal collection frequency is set considering the data collection efficiency and data density to ensure that the reflected information of the stratum below each position point can be completely captured during the movement of the radar antenna, and data loss due to too low collection frequency is avoided. In this embodiment, the collection frequency can be set to 1khz.

[0035] After starting the ground penetrating radar, the radar antenna is controlled to move at a uniform speed along the surface of the area to be detected by an automated driving device (or manual assistance). The moving speed needs to match the set collection frequency to ensure that the distance between the positions of the antennas corresponding to adjacent two collection times is uniform, avoiding uneven spatial distribution of data due to fluctuations in moving speed. During the movement of the antenna, the ground penetrating radar host automatically triggers the transmission and reception cycle according to the preset collection frequency. Each time the radar host transmits electromagnetic waves to the underground, it also receives the electromagnetic wave signals reflected back by different medium interfaces in the stratum, i.e., reflected wave signals, and stores the reflected wave signals in association with the current collection time and antenna position information. The reflected wave signals contain echo information of each reflection interface in the stratum, which can reflect the interface distribution characteristics of different underground media (such as soil, rock, disease area, etc.).

[0036] Further, during the signal collection process, the reflected wave signals obtained in real time can also be subjected to basic verification, including whether the signal amplitude is within the normal range, whether the signal waveform is significantly distorted, whether the data storage is complete, etc. If an abnormal signal (such as irregular mutation in the waveform) is found, the collection needs to be paused and the abnormal reason needs to be investigated, including the state of the antenna and the surface, the stability of the device power supply, the surrounding electromagnetic interference, etc. After the abnormality is eliminated, the collection is restarted to ensure that the reflected wave signals obtained at each collection time meet the quality requirements for subsequent analysis.

[0037] In other implementations, in addition to using a ground coupling antenna, an air coupling antenna can also be selected as an alternative. The air coupling antenna does not need to be in direct contact with the ground surface and is suitable for scenarios where there is water, silt or obstacles (such as gravel, vegetation) on the ground surface, which can avoid wear and tear or contamination when the antenna is in contact with the ground surface. When using an air coupling antenna, the fixed height (such as 5 to 15 cm) of the antenna from the ground surface needs to be set, and the signal error caused by the difference in electromagnetic wave propagation path needs to be corrected by the height compensation algorithm of the device to ensure that the collected reflected wave signals can still accurately reflect the characteristics of the underground stratum.

[0038] S2, performing signal decomposition processing on the reflected wave signals, separating out intrinsic mode components representing dielectric noise, and performing feature extraction on the intrinsic mode components to determine a dielectric noise intensity index.

[0039] The dielectric noise intensity index is used to represent the dielectric noise intensity caused by the non-uniformity of the underground medium.

[0040] In some implementations, the pre-processed reflected wave signal can be analyzed in the frequency domain first to determine the dielectric noise characteristic frequency interval, and then the wavelet packet decomposition algorithm is used to decompose the signal to 8 layers and screen the coefficients of the corresponding frequency band to reconstruct the component representing the dielectric noise. Subsequently, the peak frequency, bandwidth (frequency domain feature) and skewness, kurtosis (statistical feature) of the power spectrum of the component are extracted, the multiple features are reduced by principal component analysis, the principal component weights are determined by combining the entropy weight method, and the dielectric noise intensity index is finally obtained by weighted summation and normalization calculation. The core logic is to actively separate noise based on the preset frequency band and fuse the frequency domain and statistical features to quantify the noise intensity.

[0041] In other implementations, the reflected wave signal can also be regarded as a mixed signal of dielectric noise, formation disease, and equipment interference, and the independent component analysis algorithm (independent component analysis, ICA) is used to separate the three independent sources, and the target component is screened out by correlation with known dielectric noise samples. Then the average energy density of the wavelet ridge line of the component, the ridge line duration ratio (time-frequency domain feature) and complexity, and the approximate entropy (nonlinear feature) are extracted. With these features and the noise intensity level labeled by artificial, a support vector machine regression model is trained, the features of the component to be processed are input into the model, and the mapped continuous value is output as the dielectric noise intensity index. The core logic is to trace the noise by blind source separation and rely on machine learning data to drive the quantification of noise intensity.

[0042] S3, constructing a dielectric noise intensity sequence according to the dielectric noise intensity index, and performing spatiotemporal feature analysis on the dielectric noise intensity sequence to determine a formation disease confidence.

[0043] The formation disease confidence is used to represent the credibility of the existence of a real formation disease under noise interference.

[0044] Due to the inhomogeneity of the composition of the underground medium, the ground penetrating radar echo signal will exhibit certain random fluctuations, and electromagnetic wave scattering at the interface of multiple media may also occur during electromagnetic wave propagation. When electromagnetic waves are incident on a medium interface, Mie scattering of electromagnetic waves occurs, causing different pulses to superimpose together, resulting in energy coupling between noise and other geological anomalies, further causing the problem of difficulty in effectively separating noise.

[0045] In some implementations, a two-dimensional dielectric noise intensity sequence containing spatiotemporal information can be constructed first, based on the acquisition time and spatial location. Spatial anomaly areas are located by calculating the spatial correlation coefficient between adjacent spatial locations, and the proportion of spatial anomalies is statistically analyzed to reflect the complexity of spatial noise interference. Then, a network model is used to predict the time subsequences of each spatial location, calculating the mean and standard deviation of the time prediction error to assess the stability of temporal noise fluctuations. Finally, the weights of the spatial anomaly proportion, the mean of the time prediction error, and the standard deviation of the error are determined using the analytic hierarchy process (AHP). The confidence level of the formation disease is calculated by weighted summation and normalization. The core logic is to comprehensively evaluate the credibility of the actual disease signal under noise interference based on spatiotemporal correlation characteristics and prediction errors.

[0046] In other implementations, a one-dimensional sequence of dielectric noise intensity indices can be constructed first, and nonlinear dynamic features such as fractal dimension (reflecting noise variation patterns), maximum Lyapunov exponent (determining chaotic noise characteristics), and recursive quantization parameters (characterizing noise temporal repetition patterns) can be extracted. Then, a hidden Markov model is constructed, using these nonlinear features as observations to decode the hidden state sequence of high and low noise interference and statistically analyze the proportion of high interference. Finally, based on the numerical range of nonlinear features and the proportion of high interference, different intervals are divided and mapped to corresponding confidence baseline values. After normalization, the confidence level of the formation disease is obtained. The core logic is to determine the credibility of the actual disease signal by analyzing the nonlinear chaotic characteristics and evolution state of the noise.

[0047] S4. Based on the dielectric noise intensity index and the confidence level of formation defects, and combined with the pre-calibrated average dielectric constant of the formation, construct a depth correction weight, and adaptively correct the measured original formation defect depth according to the depth correction weight, and output the corrected formation defect depth.

[0048] In some implementations, the dielectric fluctuation coefficient, which quantifies the dielectric instability of the medium, can be determined first based on the dielectric noise intensity index. Then, the dielectric constant deviation rate is calculated by combining real-time dielectric sampling values ​​with a pre-calibrated average formation dielectric constant. Next, the confidence level of formation defects is divided into high, medium, and low levels, and a differentiated formula is used for each level. The dielectric fluctuation coefficient and deviation rate are integrated to construct a correction weight, which is then normalized. Finally, based on different intervals of the normalized weights, the original formation defect depth is slightly amplified, nonlinearly amplified, or appropriately reduced using a piecewise nonlinear function, achieving depth correction based on confidence level grading and nonlinear adjustment.

[0049] In other implementations, the approach can be rooted in the physical mechanism of electromagnetic wave propagation. First, a correlation model between dielectric constant and wave velocity is constructed based on electromagnetic theory, and the dielectric constant error rate is determined using the dielectric noise intensity index. Then, the wave velocity fluctuation range is obtained by combining the average dielectric constant of the formation. Using the confidence level of formation defects as a wave velocity correction weighting factor, the propagation velocity correction coefficient for quantifying wave velocity deviation correction is calculated. Subsequently, the original wave velocity is replaced with the correction coefficient to obtain the preliminary corrected wave velocity. Whether to introduce a depth compensation factor is determined based on the magnitude of the dielectric noise intensity index. Finally, the corrected depth of formation defects is obtained through inversion using the corrected wave velocity, propagation time, and compensation factor. The core principle is to achieve depth correction through corrected wave velocity.

[0050] Based on the above technical solution, ground-penetrating radar (GPR) is used to collect reflected wave signals from the strata at various times to obtain basic detection data. Then, signal decomposition processing is performed on the reflected wave signals to separate intrinsic mode components and extract features, effectively quantifying the dielectric noise intensity caused by the inhomogeneity of the underground medium. This solves the problem of the difficulty in accurately characterizing dielectric noise due to the inhomogeneity of the underground medium. Subsequently, a sequence is constructed based on the dielectric noise intensity index, and spatiotemporal feature analysis is performed to determine the confidence level of stratum defects. This clarifies the credibility of the existence of real stratum defects under noise interference, avoiding misjudgments caused by the confusion between noise and real defect signals. Finally, by constructing depth correction weights, the original stratum defect depth is adaptively corrected, compensating for the fluctuation error of electromagnetic wave propagation parameters caused by the inhomogeneity of the underground medium. This significantly improves the accuracy of stratum defect depth positioning and outputs a precise corrected stratum defect depth.

[0051] In one possible implementation, combining Figure 1 ,like Figure 2 As shown, the method for determining the confidence level of formation defects by performing spatiotemporal feature analysis on the dielectric noise intensity sequence in S3 above can be specifically implemented through the following S31 to S33, which are explained in detail below:

[0052] S31. Use sliding window analysis to determine the window kurtosis value for the dielectric noise intensity sequence.

[0053] Among them, the window kurtosis value is used to characterize the concentration of sudden spike noise events within the window.

[0054] In some implementations, the dielectric noise intensity sequence is used as input, a sliding window of 10ms is used with a sliding step of 1ms, and a sliding window algorithm is used. This ensures that the algorithm can capture local variation features while avoiding making the data too smooth and losing detailed information. The window kurtosis value F in each window is output. Assuming there are N windows in total, F describes the relative concentration of sudden spike noise events in its window time period. The larger the F value, the higher the probability that more sudden spike noise events occurred in this time period.

[0055] S32. Cluster the dielectric noise intensity sequence to obtain high noise region and low noise region, and determine the noise region separation degree based on the distribution of high noise region and low noise region.

[0056] Among them, the noise region separation degree is used to characterize the degree of separation between high noise region and low noise region in feature space.

[0057] In some implementations, methods for clustering dielectric noise intensity sequences may include: using the K-means clustering algorithm to cluster the dielectric noise intensity sequences, setting the number of clusters to 2 (corresponding to high-noise regions and low-noise regions respectively). Algorithm parameters include: a maximum number of iterations set to 100 (to ensure cluster convergence), initializing cluster centers using the K-means++ method (to avoid poor clustering results due to improper initial center selection), and using Euclidean distance as the distance metric (adapting to the numerical characteristics of the dielectric noise intensity index).

[0058] The dielectric noise intensity sequence is input into the K-means algorithm, and the cluster centers are updated iteratively to finally assign each data point to the nearest cluster. Clusters with larger cluster center values ​​are defined as high-noise areas (the dielectric noise intensity in these areas is significant, and the underground medium is highly heterogeneous), while clusters with smaller cluster center values ​​are defined as low-noise areas (the dielectric noise intensity in these areas is weak, and the underground medium is relatively homogeneous).

[0059] In some implementations, the method for determining the noise region separation degree may include: calculating the Euclidean distance between the cluster centers of the high noise region and the low noise region; calculating the first standard deviation of the data points in the high noise region and the second standard deviation of the data points in the low noise region; and determining the noise region separation degree based on the Euclidean distance, the first standard deviation, and the second standard deviation.

[0060] Among them, based on the Euclidean distance between cluster centers First standard deviation Compared with the second standard deviation Determine the separation degree of the noise region. The calculation formula can be:

[0061]

[0062] In the formula, Euclidean distance The distance between the centers of the two types of regions in the feature space is quantified. The greater the distance, the more significant the feature difference between the high and low noise regions.

[0063] First standard deviation Compared with the second standard deviation The dispersion of noise intensity within each region is quantified; the smaller the standard deviation, the more consistent the noise intensity within the region.

[0064] Noise region separation By combining inter-class distance and intra-class compactness, the larger the R value, the higher the degree of separation between high and low noise regions in the feature space, and the better the distinction between noise and real disease signals.

[0065] In other implementations, for scenarios where there are large differences in noise density in the dielectric noise intensity sequence (such as extremely strong and concentrated noise in local areas and weak and dispersed noise in other areas), density-based spatial clustering of applications with noise (DBSCAN) can be used to effectively identify areas with significant density differences, avoiding the limitation of K-means being insensitive to density.

[0066] S33. Determine the confidence level of formation defects based on the window kurtosis value and noise region separation.

[0067] In some implementations, methods for determining the confidence level of formation defects may include: summing the kurtosis values ​​F corresponding to multiple windows to obtain a cumulative kurtosis value. This is used to comprehensively characterize the cumulative intensity of sudden spike noise during the data acquisition process. Among them, This is the window kurtosis value corresponding to the nth window.

[0068] Then according to Separation between noise and high noise regions Determine the confidence level of formation diseases The calculation formula can be:

[0069]

[0070] In the formula, It is the average value of the window kurtosis, which represents the average peak noise intensity. By averaging, we avoid the problem of magnitude imbalance caused by excessively high cumulative values.

[0071] Confidence level of formation disease It is negatively correlated with peak noise intensity and positively correlated with the separability of noise and signal.

[0072] For hyperbolic tangent function, when the domain is greater than zero (corresponding to...) (Always positive), with a range of (0, 1), used to map the result to the interval (0, 1).

[0073] In other implementations, when dealing with mixed distributions of multiple strata (such as clay, sand, and gravel layers) and requiring data-driven adaptation to different strata noise characteristics, a confidence mapping based on a logistic regression model can be used. This specifically involves collecting historical detection data from different strata, with each sample containing a window kurtosis value F and noise region separation. And artificially calibrated confidence levels of stratigraphic diseases (e.g., levels 1-5, with level 1 being the lowest and level 5 the highest); using sample F and sample Using the input features and the labeled confidence levels as the output, a logistic regression model is trained; the F-values ​​to be processed are then... The model is input into a pre-trained model and outputs a confidence score for formation defects. The model can adapt to the mapping relationship between noise and confidence scores in different formations.

[0074] Based on the above technical solution, by analyzing the window kurtosis value, the concentration of sudden peak noise events within different time periods can be accurately captured, clearly quantifying the extreme interference of local noise. Simultaneously, high-noise and low-noise areas are clustered, and the noise region separation degree is determined based on the distribution of these two areas. This clarifies the separation degree of high and low noise areas in the feature space, effectively distinguishing the differences in the intensity distribution of noise interference. Finally, the confidence level of formation defects is determined by combining the window kurtosis value and the noise region separation degree. This comprehensively considers the local suddenness characteristics and overall distribution differences of noise, enabling the formation defect confidence level to accurately characterize the credibility of actual formation defects under noise interference.

[0075] In one possible implementation, combining Figure 1 ,like Figure 3 As shown, the method for performing signal decomposition processing on the reflected wave signal and separating the intrinsic mode components characterizing dielectric noise in S2 above can be specifically implemented through the following S21 to S22, which are explained in detail below:

[0076] S21. Use a preset signal decomposition algorithm to decompose the reflected wave signal and output the set of intrinsic mode components.

[0077] The intrinsic mode component set includes the first intrinsic mode component of the high-frequency signal and the second intrinsic mode component of the low-frequency signal.

[0078] S22. Select the second intrinsic mode component from the set of intrinsic mode components as the intrinsic mode component characterizing dielectric noise.

[0079] The non-uniformity of the underground medium composition can cause high-frequency random fluctuations in the ground penetrating radar signal that are similar to the characteristics of geological defects. The dielectric noise caused by these similar high-frequency random fluctuations can mask the real geological defect signals, making it impossible for traditional methods to effectively separate the two types of characteristic signals.

[0080] In some implementations, the reflected wave signal is used as input, the decomposition stopping condition is set to envelope standard deviation less than 0.2, and the empirical mode decomposition (EMD) algorithm is used to output the set of intrinsic mode function components.

[0081] Among them, high-frequency random noise is separated into the first eigenmode component due to its short-time abrupt change characteristics. Since the scale of the disease is much larger than the size of the medium particles, the dielectric anomaly caused by actual formation disease exhibits a low-frequency, slow-changing characteristic, resulting in the concentration of electromagnetic wave scattered energy in the second intrinsic mode component. In the middle, therefore Dielectric noise characteristics introduced by the dielectric properties of the formation. The larger the value, the more significant the abrupt change in the dielectric constant of the formation due to differences in formation composition.

[0082] Based on the above technical solution, by decomposing the reflected wave signal to obtain the set of intrinsic mode components, and by selecting the second intrinsic mode component from this set as the intrinsic mode component characterizing dielectric noise, accurate separation of dielectric noise and high-frequency interference signals can be achieved. This effectively avoids confusion caused by high-frequency interference signals in dielectric noise identification, and provides an accurate target for subsequent feature extraction of dielectric noise and further determination of the dielectric noise intensity index. This lays the foundation for distinguishing noise from real formation disease signals and ensuring the reliability of formation disease detection and analysis.

[0083] In one possible implementation, combining Figure 1 ,like Figure 3 As shown, the method for extracting features from intrinsic mode components to determine the dielectric noise intensity index in S2 above can be specifically implemented through the following S23 to S26, which are explained in detail below:

[0084] S23. Perform instantaneous characteristic analysis on the second intrinsic mode component to obtain the instantaneous amplitude and instantaneous phase of each data point in the second intrinsic mode component.

[0085] Due to random compositional changes in the underground medium, The study revealed abnormal dielectric properties of the formation, which overlapped with the actual disease in the frequency domain. However, the random component mutation of the underground medium was manifested as random amplitude spikes and instantaneous phase jumps, while the actual disease showed slow amplitude decay and continuous phase distortion.

[0086] Therefore, with For input, Each data point in the sequence is output using a Hilbert transform. The instantaneous amplitude d and instantaneous phase p are defined for each data point. The instantaneous amplitude d represents the local variation of the signal amplitude, and a larger value indicates a more drastic change in the dielectric constant of the geological medium. The instantaneous phase p represents the local evolution of the phase, and a larger value indicates a more rapid spatial variation in the conductivity of the geological medium.

[0087] S24. Based on the instantaneous amplitude of multiple data points in the second intrinsic mode component, determine the instantaneous amplitude variation coefficient, which is used to characterize the degree of dispersion of the instantaneous amplitude.

[0088] In some implementations, the instantaneous amplitude variation coefficient The calculation formula is:

[0089]

[0090] In the formula, Let d be the standard deviation of the instantaneous amplitude d of multiple data points.

[0091] The instantaneous amplitude d of multiple data points is the arithmetic mean. In real-world scenarios, the instantaneous amplitude d reflects a sudden change in the dielectric constant. It is not possible for the instantaneous amplitude of all data points to be 0, so the mean is not zero.

[0092] Standard deviation with the mean The ratio is defined as the instantaneous amplitude variation coefficient. This coefficient is dimensionless; a larger value indicates a higher degree of dispersion in the instantaneous amplitude d, meaning a more random and non-uniform local change in the dielectric constant of the underground medium. For example, the dielectric difference between particles of different sizes in a sandy or gravelly stratum can lead to… Significantly increased.

[0093] S25. Based on the instantaneous phase of multiple data points in the second intrinsic mode component, determine the maximum instantaneous phase change value used to characterize the extreme changes in the instantaneous phase.

[0094] In some implementations, the instantaneous phase change value The calculation formula is:

[0095]

[0096] In the formula, Let t be the instantaneous phase of the data point at time t. This is the instantaneous phase of the data point at the previous time (i.e., time t-1).

[0097] It represents the absolute difference in instantaneous phase between adjacent moments, which can avoid the cancellation of positive and negative values ​​and uniformly represent the magnitude of change. The larger the rate of change, the more violent the phase change between adjacent moments. For example, a sudden change in the water content of underground media can cause a sharp fluctuation in electrical conductivity, which in turn can trigger a phase change.

[0098] Divide by This eliminates dimensions, unifies the dimensional system of multiple parameters, and ensures the physical rationality of subsequent calculations and the accuracy of quantitative analysis.

[0099] Maximum instantaneous phase change It represents the maximum value among the absolute differences of instantaneous phases from all adjacent moments. Focusing on extreme phase changes, it can effectively capture dielectric noise peaks caused by drastic changes in the composition of the medium (such as at the interface between clay and sand).

[0100] S26. Determine the dielectric noise intensity index based on the instantaneous amplitude variation coefficient and the maximum instantaneous phase change value.

[0101] In some implementations, the dielectric noise intensity index The calculation formula is:

[0102]

[0103] In the formula, the instantaneous amplitude variation coefficient Multiplying the value by the maximum instantaneous phase change, and integrating the noise characteristics of amplitude discreteness and extreme phase changes, the larger the value, the more significant the dielectric noise caused by the non-uniformity of the underground medium, the greater the influence of the non-uniformity of the formation composition on the signal, and the higher the overall intensity of the dielectric noise.

[0104] Based on the above technical solution, by conducting instantaneous characteristic analysis on the second eigenmode component characterizing dielectric noise, the randomness of the dielectric constant change and the drasticness of the spatial variation of conductivity in the geological medium can be quantified. From the two key dimensions of the randomness of dielectric constant fluctuation and the extremeness of conductivity variation, the intensity of dielectric noise caused by the non-homogeneity of underground medium can be comprehensively and accurately quantified. This provides a reliable basis for noise quantification for distinguishing dielectric noise from real formation disease signals and improving the accuracy of formation disease detection.

[0105] In one possible implementation, combining Figure 3 ,like Figure 4 As shown, the method for constructing depth correction weights in S4 above, based on the dielectric noise intensity index and formation defect confidence level, combined with the pre-calibrated average dielectric constant of the formation, can be specifically implemented through the following S41 to S42, which are explained in detail below:

[0106] S41. Determine the relative depth error caused by dielectric constant fluctuation based on the dielectric noise intensity index, instantaneous amplitude variation coefficient, and average dielectric constant of the formation.

[0107] Due to the spatial inhomogeneity of the underground medium's composition, the measured dielectric constant of the formation exhibits random variations. This non-stationary characteristic, which changes over time, causes the actual propagation velocity to deviate from the nominal value, resulting in a certain propagation delay difference. This leads to a systematic offset error in the depth of formation defects.

[0108] In some implementations, based on the dielectric noise intensity index Instantaneous amplitude variation coefficient and the average dielectric constant of the formation Determine the relative depth error caused by dielectric constant fluctuations. The calculation formula can be:

[0109]

[0110] In the formula, the average dielectric constant of the formation is... It is a quantitative value of the dielectric properties of the underground medium. In real-world scenarios, the underground medium must possess dielectric properties (e.g., the dielectric constant of soil and rock is greater than 1). .

[0111] By combining the discreteness of dielectric constant abrupt changes with the benchmark effect of the average dielectric constant, the relative deviation of dielectric property fluctuations is quantified.

[0112] The relative depth error caused by the fusion of interference intensity and relative deviation, and the quantization dielectric constant fluctuation. .

[0113] S42. Determine the depth correction weight based on the relative depth error and the confidence level of formation defects.

[0114] Among them, the depth correction weight is inversely correlated with the confidence level of formation disease, and is used to characterize the correction intensity of the original formation disease depth.

[0115] In some implementations, based on relative depth error and confidence level of formation diseases Determine the depth correction weights The calculation formula can be:

[0116]

[0117] In the formula, through Modification intensity and formation disease confidence level An inverse proportional mapping is performed so that the lower the confidence weight of diseases in high strata, the higher the confidence weight of diseases in low strata.

[0118] The required correction amount is quantified by combining the depth error magnitude with the confidence level. The larger the relative depth error and the lower the confidence level, the larger the correction amount.

[0119] Further, using 1 as the base weight and adding correction amounts, we obtain the deep correction weight. .

[0120] Based on the above technical solution, the relative depth error caused by the fluctuation of the dielectric constant of the underground medium can be accurately quantified, and the depth correction weight can be determined by combining the confidence of the formation disease. The depth correction weight is inversely correlated with the confidence of the formation disease, which can be used to match the depth correction requirements under different noise interference scenarios. This provides a scientific correction intensity standard for subsequent adaptive correction of the original formation disease depth based on weight and to improve the accuracy of depth positioning.

[0121] In one possible implementation, combining Figure 3 ,like Figure 4 As shown, the method for adaptively correcting the measured original formation disease depth based on the depth correction weight in S4 above can be specifically implemented through the following S43 to S44, which are explained in detail below:

[0122] S43. Input the depth correction weight into the preset function to obtain the scaling factor.

[0123] The preset function is used to map the depth correction weights to a preset range, and the preset range ensures a smooth transition of the scaling factor.

[0124] The depth correction weight is calculated through the aforementioned steps. Its value range is This characterizes the required correction intensity for the original stratigraphic depth of the defect. When When the value distribution spans a large range, it is necessary to smooth and suppress the scaling requirements of extremely large weights (such as mapping to the (0,1) interval) to avoid scenarios where the scaling factor grows excessively (such as complex strata containing caves).

[0125] In some implementations, the preset function can be any smooth function such as the hyperbolic tangent function or a linear mapping function. Taking the hyperbolic tangent function as an example, the scaling factor... Represented as .

[0126] S44. Scale the original formation disease depth according to the scaling factor to obtain the corrected formation disease depth.

[0127] In some implementations, the method of scaling the original formation disease depth according to the scaling factor may include: when the scaling factor is greater than a preset threshold, increasing the original formation disease depth according to the scaling factor to strengthen the real disease signal; when the scaling factor is less than the preset threshold, decreasing the original formation disease depth according to the scaling factor to compress false alarms.

[0128] Specifically, the formula for calculating the depth of corrected formation defects can be:

[0129]

[0130] In the formula, This represents the original depth of the geological defects. This represents the corrected depth of geological defects. This is an empirical value (usually taken as 0.5).

[0131] Based on scaling factor Depth of the original disease Scaling adjustments are made so that the confidence weights can dynamically control the scaling speed. Taking an empirical value of 0.5 as an example, when... The output depth approaches 1.5 when it is close to 1. To enhance the true disease signals, and conversely when As the depth approaches 0, the output depth approaches 0.5. This reduces false alarms and locations caused by noise interference, thereby enabling adaptive optimization of the depth measurement results of formation defects.

[0132] Based on the above technical solution, a preset range can effectively ensure a smooth transition of the scaling factor, avoiding abrupt changes in depth adjustment due to fluctuations in correction weights, and ensuring the stability and continuity of the depth correction process. Subsequently, the original formation disease depth is scaled according to this scaling factor, enabling precise adaptation and adjustment of the original depth. The final corrected formation disease depth can significantly reduce depth positioning deviations caused by underground medium heterogeneity, reduce false alarms, and improve the accuracy and reliability of formation disease depth detection.

[0133] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0134] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0135] In this embodiment of the invention, the ground-penetrating radar-based rapid detection device for geological defects risk can be divided into functional units according to the above method example. For example, each function can be divided into its own functional unit, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0136] This invention also provides a schematic diagram of the hardware structure of a rapid detection device for geological defects based on ground-penetrating radar, see [link / reference]. Figure 5 The ground-penetrating radar-based rapid detection device for geological disease risks 500 includes a processor 501, and optionally, a memory 502 connected to the processor 501.

[0137] In the first possible implementation, see Figure 5 The rapid detection device 500 for geological defects based on ground-penetrating radar also includes a transceiver 503. The processor 501, memory 502, and transceiver 503 are connected via a bus. The transceiver 503 is used to communicate with other devices or communication networks. Optionally, the transceiver 503 may include a transmitter and a receiver. The device in the transceiver 503 that implements the receiving function can be considered as a receiver, which is used to perform the receiving steps in the embodiments of the present invention. The device in the transceiver 503 that implements the transmitting function can be considered as a transmitter, which is used to perform the transmitting steps in the embodiments of the present invention.

[0138] Based on the first possible implementation method Figure 5 The schematic diagram shown can be used to illustrate the structure of the rapid detection device for geological disease risks based on ground penetrating radar involved in the above embodiments.

[0139] in, Figure 5 The diagram can also illustrate the system chip in a ground-penetrating radar-based rapid detection device for geological defects. In this case, the actions performed by the aforementioned ground-penetrating radar-based rapid detection device for geological defects can be implemented by this system chip. The specific actions performed are described above and will not be repeated here.

[0140] In implementation, each step of the method provided in this embodiment can be completed by integrated logic circuits in the processor or by instructions in software form. The steps of the method disclosed in this embodiment can be directly manifested as being executed by a hardware processor, or being executed by a combination of hardware and software modules in the processor.

[0141] The processor in this invention may include, but is not limited to, at least one of the following: a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a microcontroller unit (MCU), or an artificial intelligence processor, etc., which are various computing devices that run software. Each computing device may include one or more cores for executing software instructions to perform calculations or processing. The processor may be a standalone semiconductor chip or integrated with other circuits into a single semiconductor chip. For example, it may be integrated with other circuits (such as encoding / decoding circuits, hardware acceleration circuits, or various bus and interface circuits) to form a System-on-a-Chip (SoC), or it may be integrated as a built-in processor within an ASIC. The ASIC with the integrated processor may be packaged separately or together with other circuits. In addition to the cores for executing software instructions to perform calculations or processing, the processor may further include necessary hardware accelerators, such as field-programmable gate arrays (FPGAs), programmable logic devices (PLDs), or logic circuits that implement dedicated logic operations.

[0142] The memory in the embodiments of the present invention may include at least one of the following types: read-only memory (ROM) or other types of static storage devices capable of storing static information and instructions; random access memory (RAM) or other types of dynamic storage devices capable of storing information and instructions; or electrically erasable programmable read-only memory (EEPROM). In some scenarios, the memory may also be a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0143] This invention also provides a computer-readable storage medium including instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0144] This invention also provides a computer program product containing instructions that, when run on a computer, cause the computer to perform any of the methods described above.

[0145] This invention also provides a chip, which includes a processor and an interface circuit. The interface circuit is coupled to the processor. The processor is used to run computer programs or instructions to implement the above-described method. The interface circuit is used to communicate with other modules outside the chip.

[0146] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0147] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings and the disclosure, will understand and implement other variations of the disclosed embodiments in carrying out the claimed invention. In this invention, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several of the functions listed in this invention.

[0148] Although the invention has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made therein without departing from the spirit and scope of the invention. Accordingly, this specification and drawings are merely illustrative of the invention and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the invention and its equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A ground stratum disease risk rapid detection method based on ground penetrating radar, characterized in that, The method comprises the following steps: Signal acquisition is performed on the stratum by ground penetrating radar to obtain reflected wave signals at each acquisition time; Signal decomposition processing is performed on the reflected wave signals to separate out intrinsic mode components representing dielectric noise, and feature extraction is performed on the intrinsic mode components to determine a dielectric noise intensity index; the dielectric noise intensity index is used to represent the dielectric noise intensity caused by the non-uniformity of the underground medium; A dielectric noise intensity sequence is constructed according to the dielectric noise intensity index, and spatio-temporal feature analysis is performed on the dielectric noise intensity sequence to determine a stratum disease confidence; the stratum disease confidence is used to represent the credibility of the existence of a real stratum disease under noise interference; According to the dielectric noise intensity index and the stratum disease confidence, a depth correction weight is constructed in combination with a pre-labeled average dielectric constant of the stratum, and the original stratum disease depth measured is adaptively corrected according to the depth correction weight to output a corrected stratum disease depth.

2. The method of claim 1, wherein, The spatio-temporal feature analysis of the dielectric noise intensity sequence to determine the stratum disease confidence comprises: A sliding window analysis is performed on the dielectric noise intensity sequence to determine a window kurtosis value; the window kurtosis value is used to represent the concentration degree of burst peak noise events in the window; The dielectric noise intensity sequence is clustered to obtain a high-noise region and a low-noise region, and a noise region separation degree is determined according to the distribution of the high-noise region and the low-noise region; the noise region separation degree is used to represent the separation degree of the high-noise region and the low-noise region in the feature space; The stratum disease confidence is determined according to the window kurtosis value and the noise region separation degree.

3. The method of claim 2, wherein, The determination of the noise region separation degree according to the distribution of the high-noise region and the low-noise region comprises: The Euclidean distance between the cluster centers of the high-noise region and the low-noise region is calculated; The first standard deviation of the data points in the high-noise region and the second standard deviation of the data points in the low-noise region are calculated; The noise region separation degree is determined according to the Euclidean distance, the first standard deviation and the second standard deviation.

4. The method of claim 2, wherein, The determination of the stratum disease confidence according to the window kurtosis value and the noise region separation degree comprises: The window kurtosis values corresponding to multiple windows are accumulated to obtain a kurtosis accumulation value; the kurtosis accumulation value is used to comprehensively represent the cumulative intensity of burst peak noise in the acquisition process; The stratum disease confidence is determined according to the kurtosis accumulation value and the noise region separation degree.

5. The method of claim 1, wherein, The signal decomposition processing of the reflected wave signals to separate out the intrinsic mode components representing dielectric noise comprises: A preset signal decomposition algorithm is used to decompose the reflected wave signals to output a set of intrinsic mode components; the set of intrinsic mode components includes a first intrinsic mode component of high-frequency signals and a second intrinsic mode component of low-frequency signals; The second intrinsic mode component is selected from the set of intrinsic mode components as the intrinsic mode component representing dielectric noise.

6. The method of claim 5, wherein, The feature extraction of the intrinsic mode components to determine the dielectric noise intensity index comprises: performing instantaneous feature analysis on the second eigenmodal component to obtain an instantaneous amplitude and an instantaneous phase of each data point in the second eigenmodal component; determining, according to the instantaneous amplitudes of the data points in the second eigenmodal component, an instantaneous amplitude coefficient of variation for characterizing a dispersion degree of the instantaneous amplitudes; determining, according to the instantaneous phases of the data points in the second eigenmodal component, an instantaneous phase maximum change value for characterizing an extreme change of the instantaneous phases; determining the dielectric noise intensity index according to the instantaneous amplitude coefficient of variation and the instantaneous phase maximum change value.

7. The method of claim 6, wherein, The constructing of the depth correction weight according to the dielectric noise intensity index and the formation disease confidence, in combination with a pre-calibrated formation average dielectric constant, comprises: determining, according to the dielectric noise intensity index, the instantaneous amplitude coefficient of variation and the formation average dielectric constant, a relative depth error caused by dielectric constant fluctuation; determining, according to the relative depth error and the formation disease confidence, a depth correction weight; the depth correction weight is inversely related to the formation disease confidence and is used to characterize a correction strength of the original formation disease depth.

8. The method of claim 7, wherein, The self-adaptive correction of the measured original formation disease depth according to the depth correction weight comprises: inputting the depth correction weight into a preset function to obtain a scaling factor; the preset function is used to map the depth correction weight into a preset range, and the preset range ensures smooth transition of the scaling factor; scaling the original formation disease depth according to the scaling factor to obtain the corrected formation disease depth.

9. The method of claim 8, wherein, The scaling of the original formation disease depth according to the scaling factor comprises: when the scaling factor is greater than a preset threshold, amplifying the original formation disease depth according to the scaling factor to strengthen a real disease signal; when the scaling factor is less than the preset threshold, reducing the original formation disease depth according to the scaling factor to compress a false alarm.

10. The method of claim 1, wherein, The ground penetrating radar is disposed with a ground coupling antenna and is in close contact with the ground surface. The signal acquisition of the formation by the ground penetrating radar to obtain a reflected wave signal at each acquisition time comprises: driving the ground penetrating radar to operate at a preset center frequency, while driving the radar antenna to move at a constant speed along the ground surface of the area to be measured; automatically triggering a transmission and reception cycle at a preset acquisition frequency to acquire and store the reflected wave signal at each acquisition time.

Images