Method for designing three-dimensional visualized gas drainage borehole construction scheme

Abstract

The present application relates to the field of coal mine gas drainage. Disclosed is a method for designing a three-dimensional visualized gas drainage borehole construction scheme. The method comprises: acquiring coal-rock occurrence information and geological anomaly information by means of high-precision geological exploration; using a convolutional neural network to generate a virtual three-dimensional coal seam model; performing measurement on a coal seam, a rock stratum, a geological anomaly and a drilling tool, binding measurement information to the virtual three-dimensional coal seam model, inputting drilling site information, and performing calculation on the basis of a preset formula to generate a borehole trajectory and construction parameters; and extracting an attachment area of a borehole deviation trajectory from the virtual three-dimensional coal seam model, performing several coal seam structure simulation tests in the attachment area on the basis of the convolutional neural network, outputting relationships between deviation magnitudes and deviation probabilities at different borehole depths, and then outputting a drainage range and a drainage blank zone range of an overall borehole scheme. By using the technical solution of the present application, a three-dimensional stereogram of a drilling site, a borehole and a coal seam can be obtained in combination with three-dimensional technology, such that a borehole construction scheme can be visually displayed.

Description

A 3D Image-Based Design Method for Gas Drainage Borehole Construction Scheme

[0001] This application claims priority to Chinese Patent Application No. 202410914388.X, filed on July 9, 2024, entitled “A Three-Dimensional Image-Based Design Method for Gas Drainage Drilling Construction Scheme”, the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of coal mine gas extraction, and in particular to a three-dimensional image-based design method for gas extraction borehole construction schemes. Background Technology

[0003] Coal is a primary energy source among mineral resources and an important industrial raw material for industries such as power, steel, and chemicals, holding a crucial strategic position in economic development. Statistics show that coal accounts for over 75% of primary energy consumption, and with the global energy crisis, its status as a primary energy source is unlikely to change for a considerable period. Meanwhile, coal-bearing methane, a byproduct of coal mining, is considered a clean energy source and has received widespread attention from major coal-producing countries. Major coal-producing nations worldwide are actively engaged in methane exploration and development. However, methane is also one of the major hazards hindering safe production in coal mines. Therefore, efficient methane extraction is a fundamental measure for achieving mine disaster management and resource utilization. Among these methods, borehole drainage technology for coal seam methane extraction has shown significant effectiveness in mine methane extraction and management. In practical applications, accurately determining borehole layout parameters based on the regional characteristics of coal seam methane occurrence is a crucial technical step in achieving efficient coal seam methane extraction.

[0004] In some cases, the design of coal mine drainage boreholes is primarily based on the experience of drainage department designers or the use of prepared design templates. However, this type of drainage borehole design is only suitable for near-horizontal coal seams with stable occurrence and minimal changes in dip, and for drainage boreholes with less stringent design requirements. It is less adaptable to drainage borehole designs in unstable coal seams with large dip angle variations, and for outburst-prone coal seams with strict requirements for borehole control range. It is difficult to achieve borehole designs that strictly comply with regulations and eliminate drainage blank zones. Furthermore, the existing graphical representation of drainage borehole designs consists of only two two-dimensional drawings (plan and profile), and does not graphically represent the borehole offset trajectory, the target layer drainage range, and the blank zone. Therefore, given the generally low educational level of frontline coal mine workers, the existing drainage borehole design methods are not intuitive enough, are difficult to understand, and result in low construction efficiency. Summary of the Invention

[0005] To address the aforementioned problems, the purpose of this application is to provide a three-dimensional image-based design method for gas drainage borehole construction, comprising: Step 1: collecting information from the coal seam through several well logs to obtain coal-rock occurrence information and geological anomaly information, and marking coal-rock sections; Step 2: preparing a trained convolutional neural network, which is trained based on a coal-rock stratum structure training model, and used to generate a virtual three-dimensional coal seam model based on the marked coal-rock occurrence information and geological anomaly information; Step 3: conducting physical strength and mechanical property tests on the coal seam, rock strata, geological anomalies, and drilling tools, and collecting coal seam data. Step 4: Bind the detected physical strength, mechanical properties, and gravitational acceleration to the virtual 3D coal seam model. Input drilling site information, and the virtual 3D coal seam model calculates based on preset formulas to generate borehole trajectories and construction parameters. Step 5: Extract the attachment area of ​​the borehole offset trajectory of the virtual 3D coal seam model. Based on a convolutional neural network, conduct several coal seam structure simulation tests in the attachment area. Generate borehole offset trajectories under different coal seam structure simulations with the same construction parameters. Compare the simulation test results with the original trajectory to show the relationship between the offset amount and the offset probability at different borehole depths.

[0006] The above solution achieved the following technical effects:

[0007] This application utilizes a virtual 3D coal seam model for borehole simulation to achieve the effect of simulating real borehole trajectories. The establishment of the 3D model provides more parameters and information for simulating curved trajectories, effectively avoiding abnormal terrain such as collapse columns and water bubbles, thus reducing drilling risks. Furthermore, by leveraging the random filling characteristic of convolutional neural networks, its simulation capabilities are broadened. Multiple iterations simultaneously address refinement defects and trajectory deviation randomness issues, providing the ability to formulate drilling plans with high success rates, stability, and fault tolerance. By combining 3D technology, this application obtains a 3D stereoscopic view of the drilling site, borehole, and coal seam, intuitively displaying the drilling construction plan and solving the problems of cumbersome design of extraction borehole parameters and the lack of intuitiveness of 2D drawings. Attached Figure Description

[0008] Figure 1 is a flowchart illustrating a three-dimensional image-based gas extraction borehole construction scheme design method provided in an embodiment of this application.

[0009] Figure 2 is a flowchart illustrating step four of a three-dimensional image-based gas extraction borehole construction scheme design method provided in an embodiment of this application.

[0010] Figure 3 is a flowchart illustrating step five of a three-dimensional image-based gas extraction borehole construction scheme design method provided in an embodiment of this application. Detailed Implementation

[0011] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0013] Example 1, as shown in Figures 1-3, provides a three-dimensional image-based design method for gas drainage borehole construction, including:

[0014] Step 1: Use seismic waves to detect the coal seam thickness distribution, plan well logging, and collect information from several wells. Mark areas where drilling occurs (sticking, rock breakage, gas eruptions, water production) as geological anomaly areas, obtaining coal-rock occurrence information and geological anomaly information, and labeling coal-rock sections. Seismic waves can obtain underground geological distribution information, facilitating well logging location planning and collecting accurate coal seam geological information based on the well logs. The information acquisition process utilizes seismic detectors.

[0015] In one implementation, information about the coal seam needs to be collected before constructing the 3D model, thereby enabling the 3D model to have a certain degree of reconstruction capability. Seismic waves are used to detect the path, time, and wave field of seismic waves excited by artificial sources in underground rock strata, probing the geometric and physical properties of the underground rock strata, such as burial depth and shape. Subsequently, drilling is conducted at several points through well logging to obtain detailed information, improving the model's reconstruction accuracy.

[0016] Step 2: Prepare the trained convolutional neural network. The convolutional neural network is trained based on the coal-rock strata structure training model. The convolutional neural network is used to generate a virtual three-dimensional coal seam model based on the labeled coal-rock occurrence information and geological anomaly information.

[0017] Among them, the coal-rock strata structure training model serves as a geological prior knowledge base. It employs an adversarial generative network architecture to decouple features from a large amount of coal seam borehole data, establishing an implicit mapping relationship between the morphological parameters and mechanical properties of the coal-rock interface. The convolutional neural network serves as a dynamic inference engine. Through a transfer learning strategy, it inherits the parameterized geological laws of the training model. Its multi-scale convolutional kernels simultaneously extract local features and regional tectonic trends from well logging annotation data in the spatial dimension, and perform geological semantic completion for missing areas based on an attention mechanism. The final generated virtual three-dimensional coal seam model adopts an unstructured tetrahedral mesh representation. Each unit node not only contains coal / rock classification labels but also integrates the equivalent Young's modulus and permeability tensor verified by Monte Carlo inversion.

[0018] In one implementation, although a large amount of relevant information can be obtained after the detection in step one, it is still impossible to restore the geological details in constructing a virtual three-dimensional coal seam model. Convolutional neural networks are one of the methods for generating models in three-dimensional geological modeling. By mining the geological patterns hidden in geological big data through geoscience artificial intelligence technology, geological details can be supplemented, the refinement of the model can be improved, and thus a basis for subsequent trajectory simulation can be provided.

[0019] Step 3: Conduct physical strength and mechanical property tests on the coal seam, rock strata, geological anomalies, and drilling tools, and collect gravity acceleration data for the coal seam area. The physical strength and mechanical property tests, as well as the gravity acceleration data, are all accurate to 10. -3 Simulations require high precision, but excessively high precision leads to a significant increase in computational burden. Therefore, an appropriate precision setting can achieve a balance between computational power and requirements.

[0020] In one embodiment, well logging is used to collect samples from the coal seam, rock strata, and geological anomalies. These samples are then sent to a laboratory to test their physical strength and mechanical properties, in order to input simulation parameters into a virtual three-dimensional coal seam model. The same tests are performed on the drilling tools in the borehole to complete the collection of simulation parameters.

[0021] Step 4: Bind the detected physical strength, mechanical properties, and gravitational acceleration to the virtual 3D coal seam model, reconstruct the well logging trajectory, and correct the locations of the marked coal-rock occurrence information and geological anomaly information based on the output well logging trajectory, updating the virtual 3D coal seam model. Input drilling site information; the virtual 3D coal seam model calculates based on preset formulas to generate the borehole trajectory and construction parameters. Construction parameters include: borehole location, borehole azimuth, dip angle, depth of coal penetration, depth of penetration through coal, and borehole depth. Generating construction parameters facilitates construction deployment and reduces the intensity of manual calculations for related parameters.

[0022] In practical applications, the borehole trajectory and construction parameters are calculated using a virtual 3D coal seam model and preset formulas. These preset formulas are trajectory parameter formulas that take into account borehole offset.

[0023] In one implementation, the detected data is bound to a virtual 3D coal seam model. First, a well logging inversion is performed. Although well logging serves as the standard for basic information acquisition, the drilling trajectory is also curved, which can cause deviations in the actual data annotation points. Therefore, inversion is necessary after binding physical strength and mechanical properties to restore the true well logging trajectory, thereby correcting the virtual 3D coal seam. Then, based on the corrected virtual 3D coal seam, borehole simulation is performed to plan the construction process. Mechanical simulation can be performed to simulate the borehole trajectory, thus restoring the true borehole trajectory. During drilling, the actual borehole trajectory is curved. Due to factors such as friction between the drill bit and the coal and rock strata, and gravity, the borehole deviates, and this deviation becomes more pronounced with increasing depth and torque. Because of different geological compositions, the amount of deviation produced by the drill bit in different geological environments varies. Therefore, binding physical strength and mechanical properties to the virtual 3D coal seam model can generate a simulation trajectory that more closely matches the actual borehole trajectory.

[0024] In addition to assisting users in formulating construction plans, virtual 3D coal seam models, taking into account the generally low educational level of frontline coal mine workers, can also be used to demonstrate plans to them. The 3D graphical construction plans reduce the difficulty of understanding for workers and improve construction efficiency.

[0025] In one embodiment, during borehole trajectory generation, the drilling process is divided into several length nodes, and the borehole trajectory is generated and spliced ​​sequentially using these length nodes as units. The boundaries of coal seams, rock strata, and geological anomalies are all located at the ends of the length nodes. During borehole trajectory generation, an interpolation algorithm is used to correct the borehole trajectory. The generated trajectory consists of several micro-trajectories that are not interconnected, resulting in low smoothness. The interpolation algorithm can correct the trajectory, improve the continuity of the overall trajectory, and even improve its accuracy. The interpolation algorithm consists of a trained BP network, trained based on a borehole trajectory training model. The BP network is used to add interpolation points within the length nodes, with the coordinates of the interpolation points as follows:

[0026] Where i is the i-th interpolation point, Z a Z is the starting point of the trajectory within the current length node. b α1 is the trajectory endpoint within the current length node, β1 is the projection inclination angle of the xz plane trajectory within the current length node, β2 is the projection inclination angle of the yz plane trajectory within the current length node, and n is the total number of interpolation points.

[0027] The BP network-based interpolation algorithm can simulate the trajectory through machine learning. Compared with ordinary interpolation algorithms, it uses an empirical model for correction, allowing the correction results to overcome the limitations of the minimum step distance and achieve better accuracy. The interpolation points are set within the length nodes using a formula to optimize the borehole trajectory curve.

[0028] In one implementation, since the drill bit's deviation during drilling alters the geological contact and stress conditions, a nodal calculation method is employed. The trajectory calculation within each length node begins at the end of the previous node to ensure simulation accuracy. The boundaries of coal seams, rock strata, and geological anomalies are all located at the ends of length nodes, reducing calculation errors at single length nodes and simplifying the computation.

[0029] In one embodiment, the interpolation algorithm is a method used in CNC machining to reduce approximation errors. By adding interpolation points to the middle section of the cut surface during the formation of the arc, the approximation error is corrected, thus reducing the approximation error. When applied to drilling trajectory correction, it can reduce the error caused by the node degree limitation of length nodes, making the trajectory smoother and closer to the actual trajectory. Conventional interpolation algorithms have certain limitations, requiring high computing power and having low smoothness. However, the interpolation algorithm trained based on a BP network can adapt to complex curve models and has a stronger ability to correct interpolation errors. It can effectively solve the problem of discontinuity between adjacent nodes caused by segmented interpolation, and has a lower computing power burden, thus shortening the interpolation cycle. Placing the interpolation points in the length nodes can improve the amplitude in a single length node, thereby improving the trajectory fitting degree.

[0030] Step 5: Extract the attachment area of ​​the borehole offset trajectory of the virtual 3D coal seam model, and conduct several coal seam structure simulation tests in the attachment area based on a convolutional neural network. Under different coal seam structure simulations, borehole offset trajectories are generated with the same construction parameters. The simulation test results are compared with the original trajectory to show the relationship between the offset amount and the offset probability at different borehole depths.

[0031] In one implementation, discrepancies exist between the simulation and reality, particularly in the accuracy of the virtual 3D coal seam construction and the simulation of borehole trajectories. The accuracy of the virtual 3D coal seam construction inevitably deviates, and the friction between the drill bit and the coal and rock strata in the borehole trajectory is random, making it impossible to definitively determine the direction of deviation. Therefore, to enhance the implementation, accuracy, and practicality of the solution, the probabilistic randomness of convolutional neural networks can be utilized to construct multiple simulation models based on collected information. Multiple borehole trajectory simulation tests are then conducted to obtain the probability distribution of different borehole trajectories. These test results demonstrate the trajectory distribution of multiple boreholes under different refined geological structures, enabling users to perform trajectory analysis based on probability distributions and develop high-success-rate, stable, and fault-tolerant borehole solutions. Extracting the attachment area of ​​the borehole offset trajectory from the virtual 3D coal seam model reduces the computational requirements of the convolutional neural network and simplifies implementation.

[0032] In step four, after generating multiple borehole trajectories, the volume of the extraction blank zone between the standard borehole trajectories is calculated based on the corresponding single-hole extraction range bound to the borehole trajectories.

[0033] In one implementation, gas drainage is carried out using multiple boreholes to ensure that the drainage gap is reduced, thereby increasing the gas drainage volume. As borehole density increases, the drainage gap decreases, but the workload also increases significantly. Therefore, a reasonable density and borehole planning can achieve the best results by balancing the drainage gap and workload. Based on the single-hole drainage radius and simulation in a virtual 3D coal seam model, the size of the drainage gap between borehole trajectories can be calculated, assisting users in planning reasonable borehole density and borehole trajectories.

[0034] Step 5 involves extracting the directional offset and offset probability of the sampling blank zone, and calculating the expected volume of the sampling blank zone.

[0035] Because practical applications differ from simulations in virtual 3D coal seam models, multiple calculations are performed using simulation tests to determine the variation relationship of the extraction blank zone based on borehole offset and offset probability, thereby calculating the expected volume of the extraction blank zone. Users can then perform borehole planning based on the expected volume of the extraction blank zone.

[0036] In step five, the distribution probability of drilled products at each borehole depth is output based on the simulation test.

[0037] In one embodiment, after outputting the distribution probability of drilling products at each borehole depth, during construction, the construction personnel can analyze the accuracy of the simulation in the virtual three-dimensional coal seam model based on the borehole products, and promptly reflect the fit between the construction and the simulation. This allows for timely detection of construction errors or simulation distortions, enabling timely modifications to reduce losses.

[0038] In step five, when outputting the distribution probability of the drilled products at each borehole depth, the time node for the drilled products to be transported to the ground is determined based on the depth inversion.

[0039] In one embodiment, during drilling, rock cuttings and other materials generated by the drill bit return to the ground through the annular space outside the drill rod. However, there is a certain time difference in transportation. Therefore, when outputting drilling product information, the time difference in return to the ground needs to be added so that construction personnel can better understand and compare the fit between construction and simulation. The fit can be verified directly by comparing the products at time points during construction without additional calculations.

[0040] After the construction plan is designed, a probability distribution map of the overall extraction range and blank zone is generated to help users evaluate the overall design effect of the construction plan, and to facilitate users to compare the advantages and disadvantages of different construction plans, thereby optimizing and adjusting the construction plan or selecting the best construction plan.

[0041] The above solution achieved the following technical effects:

[0042] 1. Conduct well logging of the coal seam to collect information. Acquire information on coal-rock occurrence and geological anomalies. Coal-rock occurrence information includes the distribution, dip angle, dip direction, and thickness of coal and rock strata, providing a direct reflection of the geological conditions of the work area.

[0043] 2. Information collection has blind spots and lacks precision; therefore, using convolutional neural networks for machine learning can compensate for the lack of geological information and the resulting refinement issues. A virtual 3D coal seam model is constructed as the basis for the scheme design.

[0044] 3. Physical strength and mechanical properties are tested on the coal seam, rock strata, geological anomalies, and drilling tools respectively. This allows for the binding of a 3D physical model to the virtual 3D coal seam, enabling mechanical simulation and borehole trajectory simulation to recreate the actual borehole trajectory. During drilling, the actual borehole trajectory is a curve. Due to friction between the drill bit and the coal and rock strata, gravity, and other factors, the borehole deviates, and this deviation becomes more pronounced with increasing depth and torque. Because of differences in geological composition, the amount of deviation varies depending on the geological environment in which the drill bit operates. Therefore, binding physical strength and mechanical properties to the virtual 3D coal seam model allows for the generation of a simulated trajectory that more closely resembles the actual borehole trajectory.

[0045] 4. Users can generate borehole trajectories by inputting drilling site information, enabling them to analyze design schemes and assist in scheme design. However, there are discrepancies between the simulation process and reality, specifically in the accuracy of the virtual 3D coal seam construction and the simulation of borehole trajectories. The accuracy of the virtual 3D coal seam construction will inevitably deviate, and the friction in the borehole trajectory caused by the friction between the drill bit and the coal and rock layers is random, making it impossible to completely determine the direction of deviation. Therefore, to improve the execution, accuracy, and practicality of the scheme, the probabilistic randomness of convolutional neural networks can be utilized to construct multiple simulation models based on the collected information, and multiple borehole trajectory simulation tests can be conducted to obtain the probability distribution of different borehole trajectories. The test results can reflect the trajectory distribution of multiple boreholes under different refined geological structures, allowing users to perform trajectory analysis based on probability distributions and thus formulate drilling schemes with high success rates, stability, and fault tolerance. Extracting the attachment area of ​​the borehole offset trajectory in the virtual 3D coal seam model can reduce the computational power requirements of the convolutional neural network and reduce the difficulty of implementation.

[0046] 5. In addition to assisting users in formulating construction plans, the virtual 3D coal seam model, considering the generally low educational level of frontline coal mine workers, can also demonstrate the plans to them. The 3D graphical construction plan reduces the difficulty of understanding for workers and improves construction efficiency.

[0047] Compared to existing technologies, this method solves the problems of cumbersome borehole parameter design and the lack of intuitiveness in two-dimensional drawings. By using a virtual three-dimensional coal seam model for borehole simulation, it achieves the effect of simulating the actual borehole trajectory. Past design methods based on two-dimensional drawings simplified the borehole trajectory to a straight line, failing to simulate curved borehole trajectories and thus introducing significant uncertainty. The establishment of a three-dimensional model provides more parameters and information for curved trajectory simulation, effectively avoiding abnormal terrain such as collapse columns and water bubbles, reducing drilling risks. Furthermore, by utilizing the random filling characteristic of convolutional neural networks, its simulation capabilities are broadened. Multiple iterations simultaneously address refinement defects and trajectory deviation randomness issues, providing the ability to formulate borehole plans with high success rates, stability, and fault tolerance.

[0048] In one embodiment, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described three-dimensional image-based gas extraction borehole construction scheme design method.

[0049] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0050] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A three-dimensional image-based design method for gas extraction borehole construction schemes, characterized in that: include: Step 1: Collect information from the coal seam through several well logs to obtain coal-rock occurrence information and geological anomaly information, and mark the coal-rock sections; Step 2: Prepare the trained convolutional neural network. The convolutional neural network is trained based on the coal-rock strata structure training model. The convolutional neural network is used to generate a virtual three-dimensional coal seam model based on the labeled coal-rock occurrence information and geological anomaly information. Step 3: Conduct physical strength and mechanical property tests on the coal seam, rock strata, geological anomalies, and drilling tools, and collect the gravity acceleration data of the coal seam area; Step 4: Bind the detected physical strength, mechanical properties and gravitational acceleration to the virtual 3D coal seam model, input drilling site information, and the virtual 3D coal seam model calculates based on preset formulas to generate drilling trajectory and construction parameters; Step 5: Extract the attachment area of ​​the borehole offset trajectory of the virtual 3D coal seam model, and conduct several coal seam structure simulation tests in the attachment area based on a convolutional neural network. Under different coal seam structure simulations, borehole offset trajectories are generated with the same construction parameters. The simulation test results are compared with the original trajectory to show the relationship between the offset amount and the offset probability at different borehole depths.

2. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: In step one, seismic waves are first used to detect the coal seam thickness distribution, and well logging is planned. Areas where stuck drill, rock breakout, gas eruption, and water production occur during well logging and drilling are marked as geological anomaly areas.

3. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: In step three, the physical strength, mechanical properties, and gravitational acceleration are all measured to an accuracy of 10. -3 .

4. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: In step four, after binding the virtual three-dimensional coal seam model, the well logging trajectory is restored, and the locations of the marked coal-rock occurrence information and geological anomaly information are corrected based on the output well logging trajectory, thus updating the virtual three-dimensional coal seam model.

5. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: The construction parameters in step four include any one of the following: hole location, drilling azimuth, inclination angle, depth of the hole where coal is encountered, depth of the hole through the coal, and hole depth.

6. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: During the drilling trajectory generation process, the drilling process is divided into several length nodes, and the drilling trajectory is generated and spliced ​​sequentially in units of length nodes.

7. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 6, characterized in that: The boundaries between coal seams, rock strata, and geological anomalies are all located at the end of length nodes.

8. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: The drilling trajectory is corrected based on an interpolation algorithm during the drilling trajectory generation process.

9. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 8, characterized in that: The interpolation algorithm consists of a trained backpropagation (BP) network, which is trained based on a borehole trajectory training model. The BP network is used to add interpolation points within the length nodes, and the coordinates of the interpolation points are: Where i is the i-th interpolation point, Z a Z is the starting point of the trajectory within the current length node. b α1 is the trajectory endpoint within the current length node, β1 is the projection inclination angle of the xz plane trajectory within the current length node, β2 is the projection inclination angle of the yz plane trajectory within the current length node, and n is the total number of interpolation points.

10. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: In step four, after generating multiple borehole trajectories, the volume of the extraction blank zone between the standard borehole trajectories is calculated.

11. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 10, characterized in that: Step 5 involves extracting the directional offset and offset probability of the sampling blank zone, and calculating the expected volume of the sampling blank zone.

12. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 1, characterized in that: In step five, the distribution probability of drilled products at each borehole depth is output based on the simulation test.

13. The three-dimensional image-based gas extraction borehole construction scheme design method according to claim 12, characterized in that: In step five, when outputting the distribution probability of the drilled products at each borehole depth, the time node for the drilled products to be transported to the ground is determined based on the depth inversion.

Images