Multi-source fusion geophysical prospecting method based on directional drilling

Abstract

The invention provides a multisource fusion geophysical prospecting method based on directional drilling, which comprises the following steps: determining a detection range, depth, path and profile, and monitoring earth surface noise to obtain E1; performing numerical simulation inversion to obtain required lowest initial energy E3; a drill bit is provided with a vibration exciter and a radar transmitter and drills to a section along a path, and a receiver array is arranged on the earth surface; the method comprises the following steps: emitting electromagnetic waves, carrying out on-site interpretation imaging, emitting seismic waves, and carrying out deep learning fusion two-wave imaging; repeating section by section, and finally interpolating to construct an underground three-dimensional obstacle model. According to the invention, underground obstacles can be accurately identified in complex environments of urban high noise, limited space and the like, and the detection depth, the resolution and the construction safety are improved.

Description

Technical Field

[0001] This invention relates to the field of urban underground space exploration technology, and more specifically to a multi-source fusion geophysical exploration method based on directional drilling. Background Technology

[0002] In the fields of modern underground engineering and resource exploration, with the acceleration of urbanization and the deepening development and utilization of underground space, accurate detection of underground obstacles has become a key link in ensuring construction safety and improving exploration efficiency. Traditional methods for detecting underground obstacles mainly rely on single geophysical techniques, such as seismic exploration or electromagnetic exploration. Seismic exploration infers the stratigraphic structure by artificially generating seismic waves and receiving their reflection and refraction signals in the underground medium, but its ability to identify small-scale obstacles is limited and it is easily affected by surface noise. Electromagnetic exploration uses the propagation characteristics of electromagnetic waves in the strata to detect electrical differences in the underground medium. Although it has high sensitivity for certain types of obstacles, a single electromagnetic wave signal is difficult to accurately image under complex geological conditions. In addition, most existing geophysical methods rely on sensor arrays deployed on the ground for signal acquisition, lacking the ability to directly detect underground space, which limits the detection accuracy and resolution. In practical applications, single geophysical techniques are often insufficient to meet the detection needs of complex underground environments, especially in scenarios requiring high-precision identification of the location, shape, and properties of underground obstacles, where the limitations of traditional methods become increasingly apparent.

[0003] Single geophysical exploration techniques cannot effectively integrate multiple physical information sources, making it difficult to meet the high-precision exploration requirements under complex geological conditions. Existing methods have insufficient ability to suppress surface noise, resulting in a low signal-to-noise ratio of the detection signal, which affects imaging quality and accuracy. The lack of effective direct underground exploration methods limits the detection accuracy and resolution to the deployment of ground sensors and signal propagation characteristics. In addition, existing technologies are insufficient in multi-source data fusion and 3D modeling, and cannot efficiently integrate and analyze the results of different geophysical exploration methods to form a complete 3D model of underground obstacles, thus failing to meet the requirements of modern underground engineering and resource exploration for accurate understanding of underground space. Summary of the Invention

[0004] This invention provides a multi-source fusion geophysical exploration method based on directional drilling, comprising: S1: Determine the depth, path, and locations of multiple detection profiles based on the area to be explored for underground obstacles; S2: Monitor the surface environmental noise in the area to be detected, and statistically analyze it to obtain the average surface environmental noise energy E1; S3: Simulate the propagation of seismic waves and electromagnetic waves in the strata using numerical simulation technology, analyze and obtain the energy E2 of seismic waves and electromagnetic waves reaching the surface, and obtain the initial energy E3 of seismic waves and electromagnetic waves based on the fact that E2 is greater than E1 through inverse analysis. S4: Install a vibration exciter and a ground-penetrating radar transmitter in the directional drilling bit. The vibration exciter and the ground-penetrating radar transmitter are capable of generating seismic waves and electromagnetic waves with a single energy of not less than E3. S5: Guide the directional drilling rig to the first detection profile position according to the depth and path determined in S1; S6: Deploy a receiver array on the ground at the location of the detection profile; S7: Pause drilling, start the ground-penetrating radar transmitter to emit electromagnetic waves, and the receiver array receives the electromagnetic waves; S8: Deep learning algorithms are used to perform underground space imaging analysis on measured electromagnetic wave signals; S9: If the results of the imaging analysis are unclear or ambiguous, the vibration exciter is activated to emit seismic waves, and the receiver array receives the seismic waves; S10: The deep learning algorithm is used to image the underground space based on the seismic wave signals received by the receiver array, and the imaging analysis results of S8 are combined to form a superimposed image. S11: Repeat steps S5 to S10 until all detection profiles are completed, and interpolate to form a three-dimensional underground obstacle model based on the stratigraphic imaging results of all detection profiles.

[0005] Furthermore, statistical analysis in S2 reveals that the average surface environmental noise energy E1 includes: Multiple noise monitoring points are set up in the area to be detected, and an array of receivers is used to continuously collect environmental noise signals within a preset monitoring time. The collected environmental noise signals are preprocessed to remove abnormal pulse interference; Time-frequency analysis was performed on the preprocessed noise signal to extract the noise energy within the frequency bands of the seismic and electromagnetic waves used in the detection. The average noise energy of all noise monitoring points within the specified frequency band is statistically averaged to calculate the average noise energy E1 of the surface environment.

[0006] Furthermore, in S3, numerical simulation analysis yielded the following energy E2 of seismic waves and electromagnetic waves reaching the Earth's surface: Based on the geological data of the area to be explored, a geological physical model including stratigraphic density, elastic wave velocity, electromagnetic wave velocity, and attenuation coefficient is established. Set the initial emission energy of seismic waves and electromagnetic waves at the drill bit location. ; The wave field of seismic and electromagnetic waves propagating from the drill bit location to the surface receiver array was calculated using the wave equation numerical simulation method, and the energy reaching the surface was extracted. ; Will Compare with E1; if If ≤ E1, then increase And repeat the numerical simulation method using the wave equation; if >E1, then at this time The minimum initial energy E3 is determined.

[0007] Furthermore, S8 employs deep learning algorithms to perform imaging analysis of the strata between the borehole and the surface, including: The electromagnetic wave signals received by the receiver array are preprocessed, including DC component removal, gain restoration, and bandpass filtering. The preprocessed electromagnetic wave signal data is input into a pre-trained deep learning neural network model. The deep learning neural network model is used to extract features from the input signal and perform nonlinear mapping to output an image of the dielectric constant distribution of the strata between the borehole and the surface. Calculate the overall signal-to-noise ratio and local contrast ratio of the dielectric constant distribution image.

[0008] Furthermore, in S9, determining that the results of the imaging analysis are unclear or ambiguous includes: The overall signal-to-noise ratio of the dielectric constant distribution image is compared with a preset signal-to-noise ratio threshold; The local contrast index of the dielectric constant distribution image is compared with a preset contrast threshold. Analyze whether there are any abnormal regions in the dielectric constant distribution image that cannot be classified and interpreted based on the dielectric constant distribution characteristics; When the overall signal-to-noise ratio is lower than a preset signal-to-noise ratio threshold, or the local contrast index is lower than a preset contrast threshold, or there are abnormal areas that cannot be classified or explained, the imaging analysis result is determined to be unclear or ambiguous.

[0009] Furthermore, the overlay analysis of the imaging analysis results in S10 combined with those in S8 includes: Spatial coordinate normalization was performed on the formation dielectric constant distribution map obtained by electromagnetic wave imaging and the formation impedance distribution map obtained by seismic wave imaging, respectively. Based on the signal-to-noise ratio of the two imaging results in their respective data domains, a fusion weight is assigned to each pixel. Based on the assigned weights, pixel-level data fusion is performed on the distribution information of dielectric constant and wave impedance to generate a comprehensive physical property parameter distribution map; The comprehensive physical property parameter distribution map is subjected to image enhancement processing to obtain the stratigraphic imaging results after overlay analysis.

[0010] Furthermore, in S11, the subsurface three-dimensional obstacle model is interpolated based on the stratigraphic imaging results of all probe profiles, including: The stratigraphic imaging results obtained at each detection profile are located at the corresponding position in three-dimensional space; A three-dimensional volume mesh is established within the spatial range formed by all the detection profiles; A spatial interpolation algorithm is used to calculate the physical property parameter values ​​of each grid node in the three-dimensional volume mesh, generating three-dimensional physical property parameter volume data; Set a threshold for physical property parameters, and identify regions in the three-dimensional physical property parameter volume data that are higher or lower than the threshold as potential obstacle regions; Three-dimensional region growth and boundary tracking are performed on the identified potential obstacle areas to determine the spatial shape, location and size of each obstacle.

[0011] Furthermore, the receiver array deployed on the ground at the detection profile location in S6 includes: The surface projection point of the drill bit at the location of the probe profile is determined as the center point of the array; Multiple receiving survey lines are laid out along the direction perpendicular to the horizontal drilling direction; Receivers are deployed at preset intervals on each receiving test line, the preset intervals being determined based on the detection depth; The number of receivers is determined based on the detection range and accuracy requirements; The array of receivers covers a detection depth area that is a preset multiple on both sides of the detection profile.

[0012] Furthermore, it also includes the following guidance and positioning steps: Positioning signals are continuously transmitted via a guide signal transmitter integrated within the drill bit; The positioning signal is received by multiple directional signal receivers deployed on the ground surface; Based on the signal parameters received by the guide signal receiver, a positioning algorithm is used to calculate the real-time three-dimensional coordinates and attitude of the drill bit; The calculated real-time position of the drill bit is compared in real time with the planned drilling path and the position of the detection profile in S1. Adjust the drilling direction based on the comparison results to ensure that the drill bit travels along the predetermined path to the target detection profile position.

[0013] Furthermore, the activation of the ground-penetrating radar transmitter in S7 to emit electromagnetic waves includes: Set the transmission parameters of the ground penetrating radar transmitter, including transmission energy, center frequency, and pulse width; The operation of the ground-penetrating radar transmitter is controlled by a pulse triggering mechanism; Record the timestamp of electromagnetic wave emission; The activation of the vibration exciter in S9 to emit seismic waves includes: Set the excitation parameters of the vibration exciter, including excitation energy, dominant frequency, and pulse width; The vibration exciter is controlled by a triggering mechanism. Record the timestamp of the seismic wave emission; The ground-penetrating radar transmitter and the vibration exciter operate in a time-division multiplexing mode, exciting at different time sequences.

[0014] The embodiments of the present invention have at least the following beneficial effects: 1. This invention combines seismic wave and electromagnetic wave geophysical exploration techniques and utilizes deep learning algorithms for imaging analysis of the detected signals, effectively solving the problem of insufficient detection accuracy of single geophysical techniques under complex geological conditions. Seismic waves can detect the macroscopic structure of strata and large-scale obstacles, while electromagnetic waves are more sensitive to changes in the electrical characteristics of strata and can identify small-scale and special-material obstacles. Combining these two techniques and performing joint imaging analysis using deep learning algorithms can more comprehensively reflect the distribution of underground obstacles, improving the accuracy and reliability of underground obstacle detection and providing more accurate geological information for underground engineering construction and resource exploration.

[0015] 2. This invention employs numerical simulation technology to simulate the propagation characteristics of seismic waves and electromagnetic waves, and obtains the initial energy based on back-analysis of surface environmental noise energy, thereby optimizing the excitation parameters of the detection signal. This technique effectively improves the signal-to-noise ratio of the detection signal and reduces the interference of surface noise on the detection results. By precisely controlling the energy of the detection signal, it ensures that it can effectively penetrate the strata during propagation and be clearly received by the receiver array, thus improving the quality and accuracy of imaging analysis. Furthermore, the optimized detection signal can better adapt to different geological conditions, enhancing the applicability and stability of this invention in complex environments.

[0016] 3. This invention generates a three-dimensional model of underground obstacles through interpolation and, combined with a guiding and positioning step, ensures that the drill bit travels along a predetermined path to the target detection profile, achieving three-dimensional visualization and precise positioning of underground obstacles. The three-dimensional obstacle model can intuitively display the spatial shape, location, and size of underground obstacles, providing important decision-making basis for underground engineering planning and construction. Simultaneously, the guiding and positioning technology ensures the accuracy and efficiency of the detection process, avoiding detection errors and repetitive work caused by the drill bit deviating from the predetermined path, significantly improving detection efficiency and project progress. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. The following drawings are not intentionally drawn to scale to actual size; their focus is on illustrating the main points of this disclosure.

[0018] Figure 1 This is a flowchart illustrating a multi-source fusion geophysical exploration method based on directional drilling provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of a longitudinal section of the detector provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a cross-sectional view of the detector provided in an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments.

[0020] The following is for reference. Figure 1 , Figure 1 This is a schematic flowchart of a multi-source fusion geophysical exploration method based on directional drilling provided in an embodiment of the present invention. Figure 1 As shown, a multi-source fusion geophysical exploration method based on directional drilling includes: S1: Determine the depth, path, and locations of multiple detection profiles based on the area to be explored for underground obstacles; S2: Monitor the surface environmental noise in the area to be detected, and statistically analyze it to obtain the average surface environmental noise energy E1; S3: Simulate the propagation of seismic waves and electromagnetic waves in the strata using numerical simulation technology, analyze and obtain the energy E2 of seismic waves and electromagnetic waves reaching the surface, and obtain the initial energy E3 of seismic waves and electromagnetic waves based on the fact that E2 is greater than E1 through inverse analysis. S4: Install a vibration exciter and a ground-penetrating radar transmitter in the directional drilling bit. The vibration exciter and the ground-penetrating radar transmitter are capable of generating seismic waves and electromagnetic waves with a single energy of not less than E3. S5: Guide the directional drilling rig to the first detection profile position according to the depth and path determined in S1; S6: Deploy a receiver array on the ground at the location of the detection profile; S7: Pause drilling, start the ground-penetrating radar transmitter to emit electromagnetic waves, and the receiver array receives the electromagnetic waves; S8: Deep learning algorithms are used to perform underground space imaging analysis on measured electromagnetic wave signals; S9: If the results of the imaging analysis are unclear or ambiguous, the vibration exciter is activated to emit seismic waves, and the receiver array receives the seismic waves; S10: The deep learning algorithm is used to image the underground space based on the seismic wave signals received by the receiver array, and the imaging analysis results of S8 are combined to form a superimposed image. S11: Repeat steps S5 to S10 until all detection profiles are completed, and interpolate to form a three-dimensional underground obstacle model based on the stratigraphic imaging results of all detection profiles.

[0021] This invention combines seismic waves and electromagnetic waves to achieve high-precision detection of underground obstacles. Directional drilling, a drilling technique that follows a predetermined trajectory, is commonly used in underground engineering and resource exploration. Its key feature is the ability to control the drill bit's direction through a guidance system. Multi-source fusion geophysical exploration combines multiple physical detection signals, such as seismic waves and electromagnetic waves, using data fusion technology to improve detection accuracy. The detection source depth, path, and multiple detection profile locations refer to the pre-planned drilling depth, drilling trajectory, and vertical cross-sectional locations to be detected based on the detection target and geological conditions before conducting underground exploration. These parameters are determined based on a preliminary understanding of the geological structure of the target area and the specific requirements of the exploration mission.

[0022] The detection source depth refers to the underground depth that the drill bit needs to reach during drilling. It is usually set based on the expected burial depth of underground obstacles. For example, in urban underground pipeline detection, the depth is between a few meters and tens of meters; while in deep geological exploration, the depth may reach hundreds of meters. The path refers to the trajectory of the drill bit underground, which can be controlled by a guidance system to avoid known obstacles or to conduct detection according to specific geological profiles. Multiple detection profile locations refer to selecting multiple sections perpendicular to the drilling direction for detailed detection during the drilling process. The distribution of these sections can be determined according to the size and complexity of the detection area; for example, a detection profile may be set at regular intervals.

[0023] The average noise energy E1 of the surface environment refers to the average energy level of background noise generated by natural and human factors in the surface environment of the exploration area. It is statistically derived by deploying multiple noise monitoring points in the exploration area and collecting noise signals. The energy E2 of seismic waves and electromagnetic waves reaching the surface refers to the energy of the simulated seismic waves and electromagnetic waves propagating from the drill bit location to the surface receiver array during numerical simulation. This parameter is used to evaluate the signal attenuation during propagation, and based on this, the initial energy E3 is obtained through inverse analysis. This is the minimum energy required for the signal to effectively penetrate the strata and be clearly received at the surface.

[0024] In some embodiments, the statistical analysis in S2 to obtain the average surface environmental noise energy E1 includes: Multiple noise monitoring points are set up in the area to be detected, and an array of receivers is used to continuously collect environmental noise signals within a preset monitoring time. The collected environmental noise signals are preprocessed to remove abnormal pulse interference; Time-frequency analysis was performed on the preprocessed noise signal to extract the noise energy within the frequency bands of the seismic and electromagnetic waves used in the detection. The average noise energy of all noise monitoring points within the specified frequency band is statistically averaged to calculate the average noise energy E1 of the surface environment.

[0025] The number and location of noise monitoring points should be determined based on the size and complexity of the detection area. For example, in a large detection area, multiple evenly distributed noise monitoring points are needed to ensure the representativeness of the collected noise signals. The receiver array needs to be selected based on the detection target and signal type. For instance, in electromagnetic wave detection, the receiver array includes multiple high-sensitivity electromagnetic sensors to collect electromagnetic wave signals. In the preprocessing stage, removing anomalous pulse interference refers to identifying and eliminating anomalous signal pulses caused by non-geological factors, such as human interference or instrument malfunction, using signal processing algorithms. In time-frequency analysis, extracting noise energy within the frequency bands of the seismic and electromagnetic waves used in the detection involves decomposing the collected noise signal into different frequency components using methods such as Fourier transform and calculating the energy distribution within the target frequency band. Finally, by statistically averaging the noise energy of all noise monitoring points, the average surface environmental noise energy E1 is obtained. This parameter will be used for subsequent signal energy assessment and noise suppression.

[0026] Noise monitoring points can be deployed in a grid pattern, for example, placing a monitoring point at regular intervals within the detection area, such as 10 or 20 meters, to ensure coverage of the entire detection area. The selection of the receiver array can be determined based on the detection depth and accuracy requirements, specifying the number and spacing of receivers. For example, in shallow detection, the receiver spacing can be set to approximately 1 meter, while in deep detection, the spacing can be appropriately increased. In the preprocessing stage, an adaptive filtering algorithm can be used to remove abnormal pulse interference. This algorithm can automatically identify and eliminate signal pulses that differ significantly from the background noise characteristics. In time-frequency analysis, methods such as Short-Time Fourier Transform (STFT) or Wavelet Transform can be used. These methods can provide the energy distribution of the signal at different times and frequencies. Through these methods, the noise energy within the target frequency band can be extracted more accurately, and the average noise energy E1 of the surface environment can be calculated, thus providing reliable data support for subsequent signal processing and imaging analysis.

[0027] In some embodiments, the energy E2 of seismic waves and electromagnetic waves reaching the Earth's surface obtained by numerical simulation in S3 includes: Based on the geological data of the area to be explored, a geological physical model including stratigraphic density, elastic wave velocity, electromagnetic wave velocity, and attenuation coefficient is established. Set the initial emission energy of seismic waves and electromagnetic waves at the drill bit location. ; The wave equation numerical simulation method was used to calculate the wave fields of seismic waves and electromagnetic waves propagating from the drill bit location to the surface receiver array, and the energy reaching the surface was extracted. ; Will Compare with E1; if If ≤ E1, then increase And repeat the above simulation process; if >E1, then at this time The minimum initial energy E3 is determined.

[0028] The geological physical model is constructed based on geological data of the area to be explored, and includes parameters such as stratigraphic density, elastic wave velocity, electromagnetic wave velocity, and attenuation coefficient. Stratigraphic density refers to the mass of rock per unit volume of the stratigraphic layer; elastic wave velocity refers to the propagation speed of seismic waves in the stratigraphic layer; electromagnetic wave velocity refers to the propagation speed of electromagnetic waves in the stratigraphic layer; and the attenuation coefficient indicates the degree of energy attenuation during signal propagation in the stratigraphic layer. The determination of these parameters requires combining geological exploration reports and existing geological data. For example, elastic wave velocity can be obtained through experiments on the propagation of seismic waves in rock, and the attenuation coefficient can be derived by analyzing the signal attenuation characteristics of known stratigraphic layers. The wave equation numerical simulation method is a mathematical method used to calculate wave propagation in a medium. By setting initial emission energy and stratigraphic parameters, it simulates the propagation process of seismic and electromagnetic waves and extracts the energy E2 reaching the surface. If E2 is less than or equal to E1, the initial emission energy needs to be increased and the simulation repeated until E2 is greater than E1.

[0029] During implementation, the construction of the stratigraphic physical model can be achieved by collecting geological exploration data of the area to be explored, including information on rock type and stratigraphic structure, and combining this data with laboratory tests and field monitoring data to determine model parameters. For example, the elastic wave velocity setting can be adjusted based on the rock type and density, referencing existing rock physical property databases. In the numerical simulation of the wave equation, numerical calculation methods such as the finite difference method or the finite element method can be used. The stratigraphic physical model parameters and initial emission energy are input to calculate the propagation process of seismic waves and electromagnetic waves. During the simulation, the initial emission energy needs to be adjusted gradually. Each adjustment can be set empirically as a small increment, such as 10% of the initial energy, to ensure the accuracy of the simulation results. In this way, the required minimum initial energy E3 can be efficiently determined, providing accurate energy parameters for subsequent exploration processes and ensuring that the detection signal can effectively penetrate the strata and be clearly received at the surface.

[0030] In some embodiments, the imaging analysis of the strata between the borehole and the surface using a deep learning algorithm in S8 includes: The electromagnetic wave signals received by the receiver array are preprocessed, including DC component removal, gain restoration, and bandpass filtering. The preprocessed electromagnetic wave signal data is input into a pre-trained deep learning neural network model. The deep learning neural network model is used to extract features from the input signal and perform nonlinear mapping to output an image of the dielectric constant distribution of the strata between the borehole and the surface. Calculate the overall signal-to-noise ratio and local contrast ratio of the dielectric constant distribution image.

[0031] Specifically, preprocessing refers to a series of processing operations performed on the acquired electromagnetic wave signals to remove noise and interference and enhance the useful information of the signal. For example, removing the DC component means eliminating the constant offset in the signal, making the signal centered at zero, which facilitates subsequent processing; gain restoration means amplifying or reducing the signal to restore its original amplitude; bandpass filtering means removing high-frequency and low-frequency noise from the signal through a filter, retaining only the signal components within the target frequency band. For deep learning neural network models, which are multi-layered artificial neural networks, the mapping relationship between the signal and the target output is learned through a large amount of training data. The model's input is the preprocessed electromagnetic wave signal data volume, and the output is an image of the dielectric constant distribution of the strata. During model training, labeled training datasets are needed. These datasets contain electromagnetic wave signals of known strata structures and their corresponding dielectric constant distributions. The network parameters are adjusted through the backpropagation algorithm, enabling the model to accurately extract features from the input signal and generate the target image.

[0032] The construction of deep learning neural network models can employ a Convolutional Neural Network (CNN) architecture, which is particularly suitable for processing spatially correlated signal data, such as electromagnetic wave signals. During model construction, multiple parameters need to be input, including the number of network layers, kernel size, and activation function type. For example, a network structure containing multiple convolutional and pooling layers can be set up. The kernel size can be selected based on the frequency characteristics of the signal, and the ReLU activation function can be used to increase the model's nonlinear fitting capability. During signal processing, the preprocessed electromagnetic wave signal can be normalized to a range between 0 and 1, facilitating neural network training and convergence. Furthermore, to evaluate the quality of the imaging results, the overall signal-to-noise ratio and local contrast ratio of the dielectric constant distribution image can be calculated. These indicators reflect the image's sharpness and detail, providing a reference for subsequent imaging analysis.

[0033] like Figure 2 As shown, in some embodiments, determining in S9 that the imaging analysis result is unclear or ambiguous includes: The overall signal-to-noise ratio of the dielectric constant distribution image is compared with a preset signal-to-noise ratio threshold; The local contrast index of the dielectric constant distribution image is compared with a preset contrast threshold. Analyze whether there are any abnormal regions in the dielectric constant distribution image that cannot be classified and interpreted based on the dielectric constant distribution characteristics; When the overall signal-to-noise ratio is lower than a preset signal-to-noise ratio threshold, or the local contrast index is lower than a preset contrast threshold, or there are abnormal areas that cannot be classified or explained, the imaging analysis result is determined to be unclear or ambiguous.

[0034] The determination is made by comparing the overall signal-to-noise ratio (SNR) and local contrast ratio of the dielectric constant distribution image with preset thresholds, and analyzing whether there are any unclassifiable anomalous regions in the image. The overall SNR refers to the ratio of the intensity of the effective signal to the noise signal in the image, reflecting the overall image quality; the local contrast ratio refers to the contrast difference between different regions in the image, used to measure the clarity of image details. Through a comprehensive evaluation of these indicators, it can be determined whether the imaging analysis results meet the requirements for further processing. If the imaging results are unclear or ambiguous, it is necessary to activate a vibration exciter to emit seismic waves to obtain more geological information, thereby improving the accuracy and reliability of the imaging.

[0035] The preset signal-to-noise ratio (SNR) and contrast thresholds are reference values ​​set based on actual detection needs and experience, used to evaluate the quality of imaging results. The SNR threshold is typically determined based on the accuracy requirements and noise level of the detection task. For example, in high-precision detection tasks, the SNR threshold is set higher to ensure image quality. The local contrast threshold is set based on the complexity of the geological structure and the imaging target; a higher contrast threshold helps to identify subtle geological differences.

[0036] Anomaly regions refer to areas in a dielectric constant distribution image that cannot be explained based on known geological features. These areas may be caused by noise interference or complex geological structures leading to signal distortion. The identification process requires detailed image analysis to pinpoint these anomaly regions and a comprehensive evaluation combining signal-to-noise ratio and contrast metrics.

[0037] During implementation, automated image analysis algorithms can be used to calculate the overall signal-to-noise ratio (SNR) and local contrast ratio. For example, the overall SNR can be obtained by calculating the ratio of signal intensity to background noise intensity in the image, while the local contrast ratio can be calculated by analyzing the differences between adjacent pixels in the image. For the identification of anomalous regions, image segmentation algorithms can be used to divide the image into different regions, and the dielectric constant distribution characteristics of each region can be analyzed. If the characteristics of a certain region differ significantly from the surrounding regions and cannot be explained by known stratigraphic models, it is marked as an anomalous region. When judging whether the imaging results meet the requirements, if the overall SNR is lower than a preset threshold, or the local contrast ratio is lower than a preset threshold, or there are anomalous regions that cannot be classified and explained, the imaging analysis results are determined to be unclear or ambiguous. In this case, a vibration exciter is activated to emit seismic waves to obtain more geological information, thereby improving the accuracy and reliability of the imaging.

[0038] In some embodiments, the overlay analysis in S10, combining the imaging analysis results from S8, includes: Spatial coordinate normalization was performed on the formation dielectric constant distribution map obtained by electromagnetic wave imaging and the formation impedance distribution map obtained by seismic wave imaging, respectively. Based on the signal-to-noise ratio of the two imaging results in their respective data domains, a fusion weight is assigned to each pixel. Based on the assigned weights, pixel-level data fusion is performed on the distribution information of dielectric constant and wave impedance to generate a comprehensive physical property parameter distribution map; The comprehensive physical property parameter distribution map is subjected to image enhancement processing to obtain the stratigraphic imaging results after overlay analysis.

[0039] Spatial coordinate normalization refers to unifying the coordinate systems of the formation dielectric constant distribution map obtained from electromagnetic wave imaging and the formation impedance distribution map obtained from seismic wave imaging. This process requires standardizing parameters such as geographic coordinates and depth coordinates of the two imaging results to ensure they correspond accurately in space. For example, interpolation methods can be used to align the grid points of the two imaging results, giving them the same resolution and coordinate range. Fusion weights are dynamically assigned based on the signal-to-noise ratio (SNR) of each imaging result in its respective data domain. A higher SNR indicates higher reliability of the imaging result, and therefore, a larger weight is assigned. For example, if the SNR of electromagnetic wave imaging is higher than that of seismic wave imaging, the electromagnetic wave imaging result will have a greater weight during the fusion process. Pixel-level data fusion refers to integrating the two imaging results at the pixel level of the image. Methods such as weighted summation are used to fuse the distribution information of dielectric constant and impedance together to generate a comprehensive physical property parameter distribution map. This process requires calculating the physical property parameters of each pixel to ensure that the fusion result accurately reflects the true characteristics of the formation.

[0040] In some embodiments, spatial coordinate normalization can be achieved by establishing a unified three-dimensional coordinate system, mapping the coordinate data of the two imaging results to this system. For example, geographic coordinates, longitude, latitude, and depth coordinates can be used to define the position of each pixel, and interpolation algorithms, such as bilinear interpolation or cubic spline interpolation, can be used to resample the two imaging results to ensure they have the same resolution and grid distribution. Regarding the allocation of fusion weights, the weight values ​​can be dynamically adjusted based on the signal-to-noise ratio (SNR) of each imaging result. For example, the SNR of each imaging result can be calculated, normalized to between 0 and 1, and then weights can be allocated based on the normalized SNR. During pixel-level data fusion, a weighted average method can be used to sum the dielectric constant and wave impedance of each pixel to generate comprehensive physical property parameters. Finally, image enhancement processing, such as contrast stretching or filtering, is applied to the comprehensive physical property parameter distribution map to highlight the detailed features of the stratigraphic structure, thereby obtaining clearer and more accurate three-dimensional stratigraphic imaging results.

[0041] In some embodiments, the process of interpolating and forming a three-dimensional underground obstacle model based on the stratigraphic imaging results of all probe profiles in S11 includes: The stratigraphic imaging results obtained at each detection profile are located at the corresponding position in three-dimensional space; A three-dimensional volume mesh is established within the spatial range formed by all the detection profiles; A spatial interpolation algorithm is used to calculate the physical property parameter values ​​of each grid node in the three-dimensional volume mesh, generating three-dimensional physical property parameter volume data; Set a threshold for physical property parameters, and identify regions in the three-dimensional physical property parameter volume data that are higher or lower than the threshold as potential obstacle regions; Three-dimensional region growth and boundary tracking are performed on the identified potential obstacle areas to determine the spatial shape, location and size of each obstacle.

[0042] A three-dimensional volumetric mesh is constructed by dividing the detection area into grids in three-dimensional space. The grid resolution can be adjusted according to the required detection accuracy and computational resources. For example, for high-precision detection tasks, a smaller grid size, such as 1 meter × 1 meter × 1 meter, can be used. The threshold values ​​for physical property parameters need to be determined based on the detection target and geological background. For example, when detecting underground pipelines, the threshold can be set based on the differences between the pipeline material's physical property parameters, such as dielectric constant or wave impedance, and the surrounding soil. Spatial interpolation algorithms are used to calculate the physical property parameter values ​​of each grid node in the three-dimensional volumetric mesh. Common interpolation methods include linear interpolation and Kriging interpolation. Through interpolation algorithms, discrete data from the detection profile can be extended to the entire three-dimensional space to generate continuous three-dimensional volumetric data of physical property parameters. In addition, three-dimensional region growing is an image segmentation method based on seed points. It starts from the identified obstacle region and gradually expands to adjacent similar regions to ultimately determine the boundary of the obstacle.

[0043] In some embodiments, the construction of the 3D volumetric mesh can be optimized by combining the geological model of the detection area and the distribution of the detection profile. For example, based on the topography and geological structure of the detection area, an appropriate mesh shape and size can be selected to improve the accuracy of interpolation. When setting the threshold for physical property parameters, the optimal threshold range can be determined by analyzing known geological data or conducting small-scale experiments. For example, by comparing the distribution of physical property parameters of known obstacle areas and background areas, a threshold that can effectively distinguish between the two can be determined. In the spatial interpolation process, a multi-step interpolation strategy can be adopted, first performing coarse interpolation to generate preliminary 3D volumetric data, and then improving the accuracy of key areas through local refinement interpolation. For 3D region growth and boundary tracking, machine learning algorithms can be combined to learn the morphological features of obstacles, thereby more accurately identifying and describing the spatial morphology of obstacles. For example, by training a classifier to identify the boundary features of obstacles, the accuracy and reliability of the model can be improved.

[0044] like Figure 3 As shown, in some embodiments, deploying a receiver array on the ground at the detection profile location in step S6 includes: The surface projection point of the drill bit at the location of the probe profile is determined as the center point of the array; Multiple receiving survey lines are laid out along the direction perpendicular to the horizontal drilling direction; Receivers are deployed at preset intervals on each receiving test line, the preset intervals being determined based on the detection depth; The number of receivers is determined based on the detection range and accuracy requirements; The array of receivers covers a detection depth area that is a preset multiple on both sides of the detection profile.

[0045] Specifically, the surface projection point is determined by measuring the real-time position of the drill bit and projecting it onto the surface. This position is typically obtained using a high-precision positioning system, such as GPS or a total station. After determining the projection point, the receiving survey lines are laid out perpendicular to the horizontal drilling direction, as this direction maximizes the reception of signals emitted from the borehole. Receivers on each receiving survey line are arranged at preset intervals, the size of which depends on the detection depth and accuracy requirements. For example, for shallow exploration, the interval is set to approximately 1 meter; for deep exploration, the interval can be appropriately increased. Furthermore, the number of receivers is determined based on the detection range and accuracy requirements, typically needing to cover a certain multiple of the detection depth area on both sides of the detection profile to ensure complete signal reception.

[0046] In some embodiments, the deployment of the receiver array needs to be optimized in conjunction with the specific detection task. For example, for complex underground environments or high-precision detection requirements, the number of receivers can be increased and the spacing reduced to improve signal resolution. When determining the surface projection point, multi-point positioning technology can be used, where multiple positioning devices simultaneously measure the drill bit position and then average the results to improve positioning accuracy. The layout of the receiving survey lines can be adjusted according to the terrain and geological conditions of the detection area. For example, in irregular terrain, the survey lines can be laid out as broken lines or curves. During signal acquisition, the signals acquired by the receivers can be monitored in real time and preliminarily processed, such as removing noise and abnormal signals, to ensure data quality. Furthermore, to improve signal coverage and acquisition efficiency, multiple receiving survey lines can be deployed on both sides of the detection profile to form a fan-shaped or rectangular receiver array layout, thereby achieving omnidirectional reception of underground signals.

[0047] In some embodiments, the following guiding and positioning steps are also included: Positioning signals are continuously transmitted via a guide signal transmitter integrated within the drill bit; The positioning signal is received by multiple directional signal receivers deployed on the ground surface; Based on the signal parameters received by the guide signal receiver, a positioning algorithm is used to calculate the real-time three-dimensional coordinates and attitude of the drill bit; The calculated real-time position of the drill bit is compared in real time with the planned drilling path and the position of the detection profile in S1. Adjust the drilling direction based on the comparison results to ensure that the drill bit travels along the predetermined path to the target detection profile position.

[0048] Specifically, the guide signal transmitter typically uses electromagnetic or acoustic signals as the carrier, and the frequency and intensity of the emitted signal can be adjusted according to the exploration depth and geological conditions. For example, in shallow exploration, a higher frequency signal can be used to improve positioning accuracy; in deep exploration, the signal frequency needs to be reduced to enhance penetration. The guide signal receiver needs to be selected according to the type of emitted signal. For example, for electromagnetic signals, a high-sensitivity electromagnetic induction sensor can be used; for acoustic signals, an acoustic sensor is required. The positioning algorithm calculates the real-time three-dimensional coordinates and attitude of the drill bit based on the signal parameters received by the receiver, such as signal strength and phase difference. For example, by measuring the time difference of the signal arriving at different receivers, the position of the drill bit can be calculated using triangulation. In addition, the drill bit's attitude includes parameters such as its inclination and azimuth in three-dimensional space, which are crucial for ensuring that the drill bit travels along a predetermined path.

[0049] The signal frequency of the guide signal transmitter can be optimized based on the geological characteristics of the detection area. For example, the optimal frequency range of the signal under specific geological conditions can be determined experimentally to ensure that the signal can propagate effectively in the strata without excessive attenuation. The placement of the guide signal receivers needs to be rationally planned according to the drilling path and the terrain of the detection area. Typically, multiple receivers need to be placed on both sides and in front of the drilling path to form a three-dimensional positioning network. The positioning algorithm can employ advanced multi-parameter fusion algorithms, such as combining signal strength, phase difference, and time difference for comprehensive calculation, to improve the accuracy and reliability of positioning. Furthermore, to adjust the drilling direction in real time, the calculated drill bit position can be compared with the predetermined path, and the automatic control system can fine-tune the drill bit's travel direction to ensure that the drill bit accurately reaches the target detection profile position.

[0050] In some embodiments, activating the ground-penetrating radar transmitter to emit electromagnetic waves in S7 includes: Set the transmission parameters of the ground penetrating radar transmitter, including transmission energy, center frequency, and pulse width; The operation of the ground-penetrating radar transmitter is controlled by a pulse triggering mechanism; Record the timestamp of electromagnetic wave emission; The activation of the vibration exciter in S9 to emit seismic waves includes: Set the excitation parameters of the vibration exciter, including excitation energy, dominant frequency, and pulse width; The vibration exciter is controlled by a triggering mechanism. Record the timestamp of the seismic wave emission; The ground-penetrating radar transmitter and the vibration exciter operate in a time-division multiplexing mode, exciting at different time sequences.

[0051] Transmission energy refers to the amount of energy used by a ground-penetrating radar (GPR) transmitter and vibration exciter when emitting signals. This parameter directly affects the signal's propagation distance and penetration capability. For example, higher transmission energy allows the signal to be detected in deeper strata, but it also increases the equipment's energy consumption. The center frequency is the frequency of the electromagnetic waves emitted by the GPR transmitter, typically selected based on the depth of the target and the characteristics of the geological formation. For instance, high-frequency signals are suitable for shallow detection, providing higher resolution; low-frequency signals are suitable for deep detection, offering stronger penetration. Pulse width refers to the duration of the transmitted signal; a narrower pulse width can improve signal resolution but may reduce signal energy. For vibration exciters, the dominant frequency refers to the main frequency of the seismic wave, usually selected based on the elastic characteristics of the target strata to ensure effective signal propagation within the formation. These parameters need to be optimized based on the specific detection task and geological conditions.

[0052] In some embodiments, the transmission parameters of the ground-penetrating radar transmitter can be adjusted according to the specific needs of the detection mission. For example, for urban underground pipeline detection, a higher center frequency, such as 100MHz, can be selected to obtain higher resolution, and the pulse width can be set to a narrower value, such as 10 nanoseconds, to improve signal clarity. The excitation parameters of the vibration exciter can also be optimized according to the detection depth and target characteristics. For example, in deep geological exploration, the dominant frequency can be set to a lower value, such as 10Hz, to enhance the signal penetration ability, and the excitation energy can be appropriately increased to ensure that the signal can be effectively detected in deep strata.

[0053] The foregoing description is illustrative of the invention and should not be construed as limiting it. Although several exemplary embodiments of the invention have been described, those skilled in the art will readily understand that many modifications can be made to the exemplary embodiments without departing from the novel teachings and advantages of the invention. Therefore, all such modifications are intended to be included within the scope of the invention as defined in the claims. It should be understood that the foregoing description is illustrative of the invention and should not be construed as limiting it to the specific embodiments disclosed, and modifications to the disclosed embodiments and other embodiments are intended to be included within the scope of the appended claims. The invention is defined by the claims and their equivalents.

Claims

1. A directional drilling based multi-source fusion method for prospecting, characterized in that, The method comprises the following steps: S1: determining the depth, path and multiple detection profile positions of the detection source according to the range of the underground obstacle to be detected; S2: monitoring the surface environmental noise of the detection area to be detected and statistically analyzing the average noise energy E1 of the surface environment; S3: simulating the propagation of seismic waves and electromagnetic waves in the stratum by numerical simulation technology, analyzing the energy E2 of the seismic waves and electromagnetic waves reaching the surface, and inversely analyzing the initial energy E3 of the seismic waves and electromagnetic waves according to E2 being greater than E1; S4: installing a vibration exciter and a ground penetrating radar transmitter in the directional drilling head, wherein the vibration exciter and the ground penetrating radar transmitter can generate seismic waves and electromagnetic waves with single energy not less than E3; S5: guiding drilling to the first detection profile position by the directional drilling machine at the depth and path determined in S1; S6: arranging a receiver array on the surface at the detection profile position; S7: pausing drilling, starting the ground penetrating radar transmitter to emit electromagnetic waves, and the receiver array receiving the electromagnetic waves; S8: using a deep learning algorithm to perform underground space imaging analysis on the measured electromagnetic wave signals; S9: if the imaging analysis result is unclear or ambiguous, starting the vibration exciter to emit seismic waves, and the receiver array receiving the seismic waves; S10: using a deep learning algorithm to image the underground space according to the seismic wave signals received by the receiver array, and superimposing the imaging analysis result of S8; S11: repeating steps S5 to S10 until all detection profiles are completed, and interpolating a three-dimensional underground obstacle model according to the stratum imaging results of all detection profiles.

2. The method of claim 1, wherein, The statistical analysis in S2 to obtain the average noise energy E1 of the surface environment comprises: arranging multiple noise monitoring points in the detection area, using a receiver array to continuously collect environmental noise signals within a preset monitoring time; preprocessing the collected environmental noise signals to remove abnormal pulse interference; performing time-frequency analysis on the preprocessed noise signals to extract noise energy within the frequency band range of the detection seismic waves and electromagnetic waves; statistically averaging the noise energy of all noise monitoring points within the frequency band range to calculate the average noise energy E1 of the surface environment.

3. The method of claim 1, wherein, The analysis in S3 to obtain the energy E2 of the seismic waves and electromagnetic waves reaching the surface by numerical simulation technology comprises: based on the geological data of the detection area, establishing a stratum physical model containing stratum density, elastic wave velocity, electromagnetic wave velocity and attenuation coefficient; Setting initial launch energy of seismic and electromagnetic waves at drill bit location ; The wave field of seismic and electromagnetic waves propagating from the drill bit location to the surface receiver array is calculated using a numerical simulation method of wave equation, and the energy reaching the surface is extracted ; Will Compare with E1; if If ≤ E1, then increase And repeat the numerical simulation method using the wave equation; if >E1, then at this time The minimum initial energy E3 is determined.

4. The method of claim 1, wherein, The imaging analysis in S8 of the stratum between the borehole and the surface by a deep learning algorithm comprises: preprocessing the electromagnetic wave signals received by the receiver array, including removing direct current components, restoring gain and band-pass filtering; inputting the preprocessed electromagnetic wave signal data into a pre-trained deep learning neural network model; performing feature extraction and nonlinear mapping on the input signals by the deep learning neural network model to output a dielectric constant distribution image of the stratum between the borehole and the surface; calculating the overall signal-to-noise ratio and local contrast index of the dielectric constant distribution image.

5. The method of claim 4, wherein, The judgment in S9 that the imaging analysis result is unclear or ambiguous comprises: comparing the overall signal-to-noise ratio of the permittivity distribution image with a preset signal-to-noise ratio threshold value; comparing the local contrast index of the permittivity distribution image with a preset contrast threshold value; analyzing whether there is an abnormal area in the permittivity distribution image that cannot be classified and explained according to the permittivity distribution characteristics; when the overall signal-to-noise ratio is lower than the preset signal-to-noise ratio threshold value, or the local contrast index is lower than the preset contrast threshold value, or there is an abnormal area that cannot be classified and explained, determining that the imaging analysis result is unclear or ambiguous.

6. The method of claim 1, wherein, The superimposed analysis in S10 includes: respectively performing spatial coordinate normalization processing on the stratigraphic permittivity distribution image obtained by electromagnetic wave imaging and the stratigraphic wave impedance distribution image obtained by seismic wave imaging; based on the signal-to-noise ratios of the two imaging results in their respective data domains, assigning a fusion weight to each pixel point; according to the assigned weight, performing pixel-level data fusion on the distribution information of permittivity and wave impedance to generate a comprehensive physical property parameter distribution image; performing image enhancement processing on the comprehensive physical property parameter distribution image to obtain the stratigraphic imaging result after superimposed analysis.

7. The method of claim 1, wherein, The formation of the underground three-dimensional obstacle model in S11 includes: positioning the stratigraphic imaging result obtained at each detection profile to the corresponding position in the three-dimensional space; establishing a three-dimensional body grid within the spatial range formed by all the detection profiles; using a spatial interpolation algorithm to calculate the physical property parameter value of each grid node in the three-dimensional body grid to generate three-dimensional physical property parameter body data; setting a physical property parameter threshold value, and identifying the area higher or lower than the threshold value in the three-dimensional physical property parameter body data as a potential obstacle area; performing three-dimensional region growing and boundary tracking on the identified potential obstacle area to determine the spatial form, position and size of each obstacle.

8. The method of claim 1, wherein, The surface layout of the receiver array at the detection profile position in S6 includes: determining the surface projection point of the drill bit at the detection profile position as the array center point; laying out multiple receiving lines perpendicular to the horizontal drilling direction; laying out receivers on each receiving line at a preset interval, the preset interval being determined based on the detection depth; determining the number of receivers according to the detection range and accuracy requirements; the layout range of the receiver array covers a detection depth area with a preset multiple on both sides of the detection profile.

9. The method of claim 1, wherein, Further comprising the following guiding and positioning steps: continuously emitting positioning signals through the guiding signal emitter integrated in the drill bit; receiving the positioning signals through the multiple guiding signal receivers laid out on the ground surface; based on the signal parameters received by the guiding signal receivers, using a positioning algorithm to solve the real-time three-dimensional coordinates and attitude of the drill bit; real-time comparing the solved real-time position of the drill bit with the drilling path planned in S1 and the detection profile position; adjusting the drilling direction according to the comparison result to ensure that the drill bit travels along the predetermined path to the target detection profile position.

10. The method of claim 1, wherein, The launching of the ground penetrating radar transmitter to emit electromagnetic waves in S7 includes: setting the emission parameters of the ground penetrating radar transmitter, including emission energy, center frequency and pulse width; controlling the ground penetrating radar transmitter to work using a pulse trigger mechanism; recording the timestamp of the electromagnetic wave emission; The starting of the vibration exciter in S9 to emit seismic waves comprises: Setting the excitation parameters of the vibration exciter, including excitation energy, main frequency and pulse width; Using a trigger mechanism to control the vibration exciter to work; Recording the time stamp of the emission of the seismic wave; Wherein, the ground penetrating radar transmitter and the vibration exciter adopt a time-sharing working mode and are excited in different time sequences respectively.

Images