Method for inverting internal fracture parameters of rock mass based on ground penetrating radar image

Abstract

The invention relates to a method for inverting internal fracture parameters of a rock mass based on a ground penetrating radar image. The method comprises the following steps: acquiring the ground penetrating radar image of the rock mass; based on the ground penetrating radar B-scan image, through a CNN-based image identification model, identifying a crack on the image as one of preset crack distribution forms; and based on the distribution form of the fracture, determining a geometric parameter type corresponding to the distribution form, and inverting the geometric parameter of the corresponding type of the fracture in combination with the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image. According to the method, the distribution form of the fracture on the image is identified based on the B-scan image, then the parameter type corresponding to the fracture distribution form is determined, the geometric parameters of the corresponding type of the fracture are inverted in combination with the image features on the B-scan image and the A-scan image of the ground penetrating radar, and then the length, the inclination angle, the position parameters and the like of the fracture are determined. And a more reliable basis is provided for rock mass stability analysis. The method is suitable for the field of rock mass fracture detection.

Description

Technical Field

[0001] The invention relates to a method for inverting internal crack parameters of a rock mass based on a ground penetrating radar image, and is applicable to the field of rock mass crack detection. Background Art

[0002] Under the influence of long-term natural factors such as weathering, groundwater action, earthquakes, etc., cracks of different sizes and shapes will form inside the slope rock mass. Cracks are discontinuous surfaces inside the rock mass. The existence of cracks will affect the integrity and mechanical properties of the rock mass, and they are easy to become channels for rainwater infiltration under rainfall. They are an important factor causing rock slope collapse and sliding. Therefore, rock mass cracks are a factor that needs to be considered in slope stability analysis. Accurately describing the distribution, development and expansion laws of rock mass cracks can provide a scientific basis for slope stability evaluation and prevention and control engineering design.

[0003] At present, the main methods used to obtain rock fracture parameters include photogrammetry, drilling method, and adit method.

[0004] Photogrammetry can obtain a large amount of parameter information of outcrop cracks, and then infer the parameters of internal cracks, but it cannot directly obtain the parameters of internal cracks in the rock mass. However, the geometric parameters of the slope are different from the surface to the inside of the slope. For example, in deep valleys, due to the cutting of rivers, the slope will form unloading areas of different strengths, resulting in the length and width of the cracks near the slope surface being larger than the internal cracks. It is not reliable to infer the geometric and position parameters of internal cracks only through outcrops.

[0005] The drilling method requires manual drilling into the slope, which has the disadvantages of large workload, low efficiency, and high risk, and it is difficult to obtain the distribution of cracks in the entire slope.

[0006] The tunnel method requires the excavation of a tunnel within the slope, and then the geologists manually measure the geometric information of the cracks inside the tunnel. However, tunnel excavation is very expensive, and similar to the drilling method, it can only count the crack information in a certain area near the tunnel. It is obviously not feasible to obtain the distribution of cracks inside the entire slope. Summary of the invention

[0007] The technical problem to be solved by the present invention is: in view of the above-mentioned existing problems, a method for inverting the internal fracture parameters of a rock mass based on ground penetrating radar images is provided.

[0008] The technical solution adopted by the present invention is: a method for inverting internal fracture parameters of rock mass based on ground penetrating radar images, characterized in that it includes:

[0009] Acquire ground penetrating radar images of rock masses;

[0010] Based on the ground penetrating radar B-scan image, the cracks on the image are identified as one of the preset crack distribution forms through the CNN-based image recognition model;

[0011] Based on the distribution form of the cracks, the type of geometric parameters required to characterize this type of cracks is determined, and the geometric parameter values ​​of the corresponding type of cracks are inverted by combining the image features Ⅰ of the ground penetrating radar B-scan image and the image features Ⅱ of the ground penetrating radar A-scan image;

[0012] The image feature I includes the number of acquisition channels corresponding to the cracks in the image; the image feature II includes the time difference between the peaks in the image.

[0013] The method of identifying the cracks on the image as one of the preset crack distribution forms based on the ground penetrating radar B-scan image includes:

[0014] The ground penetrating radar B-scan image is input into the CNN-based image recognition model to identify the crack distribution pattern corresponding to the cracks in the B-scan image.

[0015] The method of determining the geometric parameters required to characterize the cracks based on the distribution form of the cracks includes:

[0016] Based on the distribution form of the cracks, the geometric parameters required to characterize the cracks are determined by comparing the preset distribution form and geometric parameter correspondence table.

[0017] The preset crack distribution forms include: a single crack parallel to the detection surface, a single crack perpendicular to the detection surface, and a single crack inclined to the detection surface;

[0018] Also, a pair of parallel cracks parallel to the detection surface, a pair of parallel cracks perpendicular to the detection surface, a pair of parallel cracks inclined to the detection surface, a pair of orthogonal cracks, a V-shaped crack opening upward, and a V-shaped crack opening downward.

[0019] Based on the distribution form of the cracks, the geometric parameters required to characterize the cracks are determined, and the geometric parameter values ​​corresponding to the cracks are inverted by combining the image features I of the ground penetrating radar B-scan image and the image features II of the ground penetrating radar A-scan image, as follows:

[0020] For a pair of parallel cracks parallel to the detection surface, the corresponding geometric parameters include: the horizontal coordinate x1 of the left end point of the first crack, the vertical coordinate y1 of the left end point of the first crack, the length l1 of the first crack, the horizontal coordinate x2 of the left end point of the second crack, the distance Δd between the two cracks, and the length l2 of the second crack;

[0021]

[0022] l1=Δn1×Δl

[0023] l2=Δn2×Δl

[0024] x1=n1×Δl+x i

[0025] x2=n2×Δl+x i

[0026] Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the rock matrix, t1 is the time difference between the ground direct wave and the top of the first bowl-shaped curve, Δt is the time difference between the tops of the two bowl-shaped curves, n1 and n2 are the acquisition channel numbers corresponding to the left ends of the two bowl-shaped hyperbolic straight segments in the image, Δl is the channel spacing, and x i is the initial position of the transmitting antenna.

[0027] For a pair of orthogonally distributed cracks, the corresponding geometric parameters include: the horizontal coordinate x1 of the left end point of the horizontal crack, the vertical coordinate y1 of the left end point of the horizontal crack, the length l1 of the horizontal crack, the horizontal coordinate x2 of the top of the vertical crack, the vertical coordinate y2 of the top of the vertical crack, and the length l2 of the vertical crack;

[0028]

[0029] l1=Δn1×Δl

[0030] x1=n1×Δl+x i

[0031] x2=n2×Δl+x i

[0032] l2=Δn2×Δl

[0033] Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the rock matrix, t1 is the time difference between the ground direct wave and the top of the bowl-shaped curve in the image, t2 is the time difference between the ground direct wave and the top of the high and narrow hyperbola in the image, Δn1 is the number of acquisition channels spanned by the straight section of the bowl-shaped hyperbola in the image, Δl is the channel spacing, n1 is the number of acquisition channels corresponding to the left end of the straight section of the bowl-shaped hyperbola in the image, and n2 is the number of acquisition channels corresponding to the top of the high and narrow hyperbola in the image; x i is the initial position of the transmitting antenna, and Δn2 represents the number of acquisition channels spanned by the horizontal segment of the bowl-shaped hyperbola in the B-scan image in the other direction perpendicular to the original detection plane.

[0034] For a V-shaped crack opening upward or a V-shaped crack opening downward, the corresponding geometric parameters include: the abscissa x1 of the end point A, the abscissa x2 of the end point B, the abscissa x3 of the vertex C, the ordinate y1 of the end point A, the ordinate y2 of the end point B, and the ordinate y3 of the vertex C;

[0035] x1=n1×Δl+x i

[0036] x2=n2×Δl+x i

[0037] x3=n3×Δl+x i

[0038]

[0039] Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the rock matrix, n1, n2, n3 are the acquisition channel numbers corresponding to the three arc-shaped turning points in the image, Δl is the channel spacing, x i is the initial position of the transmitting antenna, and t1, t2, and t3 are the time differences between the three bending points in the image and the direct wave from the ground.

[0040] A device for inverting internal fracture parameters of a rock mass based on ground penetrating radar images, characterized by comprising:

[0041] A radar image acquisition module, used to acquire ground penetrating radar images of rock masses;

[0042] A distribution form identification module, for identifying the corresponding fracture crack as one of the preset fracture distribution forms based on the ground penetrating radar B-scan image;

[0043] The geometric parameter inversion module is used to determine the geometric parameters required to characterize this type of cracks based on the distribution form of the cracks, and to invert the corresponding geometric parameter values ​​of this type of cracks in combination with the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image;

[0044] The image feature I includes the number of acquisition channels corresponding to the cracks in the image; the image feature II includes the time difference between the peaks in the image.

[0045] A storage medium stores a computer program executable by a processor, wherein the computer program implements the steps of the method when executed.

[0046] A data processing device comprises a memory and a processor, wherein the memory stores a computer program executable by the processor, and is characterized in that the steps of the method are implemented when the computer program is executed.

[0047] The beneficial effects of the present invention are as follows: the present invention identifies the distribution form of cracks on the image based on the B-scan image, and then determines the parameter type corresponding to the crack distribution form, combines the image features on the ground penetrating radar B-scan image and the A-scan image, and inverts the geometric parameters of the corresponding type of the crack, and then determines the crack length, inclination angle and position parameters, etc., to provide a more reliable basis for rock stability analysis.

[0048] The present invention identifies the distribution form of fractures, determines the parameters required to characterize such fractures, and then inverts the corresponding geometric parameter values, thereby reducing the amount of calculation and improving data processing efficiency. It can realize rapid data collection and processing, greatly improving on-site work efficiency, and is particularly suitable for large-area geological surveys.

[0049] The ground penetrating radar technology in the present invention is a non-destructive detection method. Compared with the traditional geological exploration method, it does not require excavation or destruction of the ground, thus reducing environmental damage and engineering costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 This is a GPR image of a single crack.

[0051] Figure 2 A pair of GPR images of cracks.

[0052] Figure 3 Schematic diagram of the parameters of a pair of parallel cracks.

[0053] Figure 4 Schematic diagram of the parameters of a pair of orthogonal cracks.

[0054] Figure 5 Schematic diagram of the parameters of a pair of V-shaped cracks.

[0055] Figure 6 Schematic diagram of ground penetrating radar imaging principle.

[0056] Figure 7 This is a case model of a pair of horizontal cracks.

[0057] Figure 8 A pair of GPR images of horizontal crack cases.

[0058] Fig. 9 This is a case model for a pair of orthogonal cracks.

[0059] Fig.10 GPR images of a pair of orthogonal crack cases.

[0060] Fig.11 This is a pair of V-shaped crack case models.

[0061] Fig.12GPR images of a pair of V-shaped crack cases. DETAILED DESCRIPTION

[0062] Embodiment 1: This embodiment is a method for inverting internal fracture parameters of a rock mass based on a ground penetrating radar image, which specifically includes the following steps:

[0063] S1. Obtain the ground penetrating radar image of the rock mass. Use the ground penetrating radar (GPR) to detect the surface of the rock mass and obtain the original ground penetrating radar image. The obtained original ground penetrating radar image is preprocessed, including static correction removal, DC drift removal, gain, horizontal interference removal, bandpass filtering, sliding average and other processing.

[0064] S2. Based on the ground penetrating radar B-scan image, the cracks on the image are identified as one of the preset crack distribution forms through the CNN-based image recognition model.

[0065] The preset crack distribution forms in this embodiment include a single crack parallel to the detection surface, a single crack perpendicular to the detection surface, and a single crack inclined to the detection surface; as well as a pair of parallel cracks parallel to the detection surface, a pair of parallel cracks perpendicular to the detection surface, a pair of parallel cracks inclined to the detection surface, a pair of orthogonally distributed cracks, a V-shaped crack opening upward, and a V-shaped crack opening downward.

[0066] The criteria for identifying a single crack in this example are: a single crack parallel to the detection surface produces a strong signal at the top of the radar image, with a clear horizontal section, and both ends extending to the left and right in a hyperbolic shape, forming an inverted bowl ( Figure 1 a); A single crack perpendicular to the detection surface produces a radar image that appears as a high and narrow hyperbola with a weak field strength ( Figure 1 b); A single crack inclined to the detection surface produces a radar image as an inclined line segment, with a stronger field strength in the middle and a gradually weakening field strength at both ends ( Figure 1 c).

[0067] In this embodiment, the criterion for identifying a pair of cracks is that a pair of parallel cracks parallel to the detection surface appear as two bowl-shaped flat-top hyperbolas distributed up and down in the image ( Figure 2 a); A pair of parallel cracks perpendicular to the detection surface appear as two high and narrow hyperbolas arranged left and right in the image ( Figure 2 b); A pair of parallel cracks inclined to the detection surface appear as two inclined line segments arranged vertically in the image ( Figure 2 c); A pair of orthogonally distributed cracks, whose radar image shows a bowl-shaped hyperbola superimposed with a high and narrow hyperbola ( Figure 2 d); a V-shaped crack opening upward, whose radar image appears as an "X"-shaped intersection with an upward convex hyperbola below ( Figure 2 e); A V-shaped crack with a downward opening, whose radar image appears as a straight line with two arms and an inverted "V" curve at the top ( Figure 2 f).

[0068] In this embodiment, the ground penetrating radar B-scan image is input into a CNN-based image recognition model to identify the cracks in the B-scan image as a specific one of the preset crack distribution forms.

[0069] S3. Based on the distribution form of the cracks, determine the geometric parameters required to characterize this type of cracks, and invert the geometric parameter values ​​of this type of cracks by combining the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image, where image feature I includes the number of acquisition channels corresponding to the cracks in the image; image feature II includes the time difference between the peaks in the image.

[0070] In this embodiment, based on the distribution form of the cracks, the geometric parameters required to characterize the cracks are determined by comparing with a preset distribution form and geometric parameter correspondence table.

[0071] For two cracks parallel to the detection surface, assuming that the width of the crack is constant and very small, six parameters are required to determine their position distribution and geometric parameters: (1) the horizontal coordinate x1 of the left end point of the first crack; (2) the vertical coordinate y1 of the left end point of the first crack; (3) the length l1 of the first crack; (4) the horizontal coordinate x2 of the left end point of the second crack; (5) the distance Δd between the two cracks; (6) the length l2 of the second crack, see Figure 3 .

[0072] For two mutually orthogonal cracks, assuming that the width of the crack is constant and very small, six parameters are required to determine their position distribution and geometric parameters: (1) the horizontal coordinate x1 of the left end point of the horizontal crack; (2) the vertical coordinate y1 of the left end point of the horizontal crack; (3) the length l1 of the horizontal crack; (4) the horizontal coordinate x2 of the top of the vertical crack; (5) the vertical coordinate y2 of the top of the vertical crack; (6) the length l2 of the vertical crack, see Figure 4 .

[0073] For two V-shaped cracks, assuming that the width of the cracks is constant and very small, six parameters are required to determine their position distribution and geometric parameters: (1) the horizontal coordinate x1 of the end point A, (2) the horizontal coordinate x2 of the end point B, (3) the horizontal coordinate x3 of the vertex C, (4) the vertical coordinate y1 of the end point A, (5) the vertical coordinate y2 of the end point B, (6) the vertical coordinate y3 of the vertex C, see Figure 5 .

[0074] In this embodiment, the image features of the ground penetrating radar are used to invert the geometric and positional parameters of the rock cracks. The ground penetrating radar transmits broadband short-pulse electromagnetic waves into the ground through the transmitting antenna. When the electromagnetic waves propagate in the underground medium, they will produce reflection, transmission and refraction when they encounter interfaces with different electrical properties. There are differences in the electrical properties of rock cracks and the surrounding intact rocks, such as dielectric constant, conductivity, etc. When electromagnetic waves encounter cracks, obvious reflection signals will occur. These signals are received and recorded by the receiving antenna to form radar images. These images show the two-dimensional or three-dimensional distribution of the internal structure of the rock mass.

[0075] Assume that the underground target is a point mass, which is located at dm underground, and the ground penetrating radar is located at (x0,0) ( Figure 6 ), during the detection process:

[0076] Distance between underground particle and antenna:

[0077]

[0078] The time it takes for a radar wave to travel from the transmitting antenna to the receiving antenna:

[0079]

[0080] Combining the two equations, we get:

[0081]

[0082] It can be seen from formula (3) that the B-scan image formed by underground particle A is a hyperbola, opening downward, with the vertex The specific working principle of ground penetrating radar is shown in the figure. For rock cracks, it can be regarded as a combination of multiple particles arranged in different ways, so the B-scan image it produces is the deformation and superposition of multiple hyperbolas. The vertices, inflection points and other characteristic points in the image theoretically correspond to the end points of the cracks.

[0083] The electromagnetic wave propagates underground according to the two-way travel time formula:

[0084]

[0085] Where c is the propagation speed of electromagnetic waves in air (3×10 8 m / s), ε r is the relative dielectric constant of the medium, h is the distance traveled by the electromagnetic wave (m), and t is the time taken by the electromagnetic wave (ns). Therefore, the burial depth of the crack can be inverted based on the time in the GPR image.

[0086] The inversion calculation formula of the parallel crack parameters in this embodiment is:

[0087]

[0088] l1=Δn1×Δl (7)

[0089] l2=Δn2×Δl (8)

[0090] x1=n1×Δl+x i (9)

[0091] x2=n2×Δl+x i (10)

[0092] Among them, t1 is the time difference between the ground direct wave and the top of the first bowl-shaped curve, Δt is the time difference between the tops of the two bowl-shaped curves, and n 1、 n2 is the number of acquisition channels corresponding to the left ends of the two bowl-shaped hyperbolic straight segments in the image, Δl is the channel spacing (m), x i is the initial position of the transmitting antenna (m).

[0093] The parameter inversion calculation formula of the orthogonal crack in this embodiment is:

[0094]

[0095] l1=Δn1×Δl (13)

[0096] x1=n1×Δl+x i (14)

[0097] x2=n2×Δl+x i (15)

[0098] l2=Δn2×Δl (16)

[0099] Where t1 is the time difference between the ground direct wave and the top of the bowl-shaped curve in the image, t2 is the time difference between the ground direct wave and the top of the high and narrow hyperbola in the image, Δn1 is the number of acquisition channels spanned by the straight section of the bowl-shaped hyperbola in the image, Δl is the channel spacing (m), n1 is the number of acquisition channels corresponding to the left end of the straight section of the bowl-shaped hyperbola in the image, and n2 is the number of acquisition channels corresponding to the top of the high and narrow hyperbola in the image; x i is the initial position of the transmitting antenna (m), and Δn2 represents the number of acquisition channels spanned by the horizontal segment of the bowl-shaped hyperbola in the B-scan image in the other direction perpendicular to the original detection plane.

[0100] The V-shaped crack parameter inversion method in this embodiment is:

[0101] x1=n1×Δl+x i (17)

[0102] x2=n2×Δl+x i (18)

[0103] x3=n3×Δl+x i (19)

[0104]

[0105] Among them, n1, n2, n3 are the number of acquisition channels corresponding to the three arc turning points in the image, Δl is the channel spacing (m), x i is the initial position of the transmitting antenna (m), t1, t2, and t3 are the time differences between the three bending points in the image and the direct wave from the ground.

[0106] Figure 7 is a pair of horizontal crack case models, Figure 8 yes Figure 7 Corresponding to the ground penetrating radar image, according to the inversion method proposed in this embodiment, the calculated parameters and model parameters are sorted into Table 1. Figure 8 (a) is a B-scan image formed by merging 68 single-channel waves generated by the radar moving along the long side of the model. Figure 8 (b) is the A-scan image obtained by simulation when the transmitting antenna is located at (0.7, 0). It can be found that the calculation results are in good agreement with the original data set in the model, indicating that the proposed parameter inversion method can effectively invert the geometric and position parameters of a pair of parallel horizontal cracks.

[0107] Table 1 Parallel crack parameter calculation table

[0108]

[0109] Fig. 9 is a pair of orthogonal crack case models, Fig. 9 (a) represents the original detection direction, Fig. 9 (b) indicates the vertical detection direction. It should be noted that since the coordinate system changes after the detection direction changes, the coordinates of each endpoint also change, but this is not the goal we need to reposition.

[0110] Fig.10 yes Fig. 9 The corresponding GPR image, Fig.10 (a) is the B-scan image obtained by merging 68 single-channel waves generated when the GPR moves along the long side of the model. Fig.10 (b) is the A-scan image generated when the transmitting antenna is located at (0.7, 0). Fig.10(c) is the B-scan image in another direction perpendicular to the original detection plane. The calculated parameters and model parameters are summarized in Table 2. It can be found that the calculated results are in good agreement with the original data set in the model, indicating that the proposed parameter inversion method can effectively invert all geometric and position parameters of a pair of orthogonal cracks.

[0111] Table 2 Orthogonal crack parameter calculation table

[0112]

[0113]

[0114] Fig.11 This is a pair of V-shaped crack case models. Fig.12 Yes, it is a ground penetrating radar. Fig.11 The B-scan image is a combination of 68 single-channel waves generated by the long-side movement of the model shown in the figure. The parameters calculated according to the inversion method proposed above and the model parameters are summarized in Table 3. It can be found that the calculation errors are no more than 10%, and the calculation results are in good agreement with the original data set in the model, indicating that the proposed parameter inversion method can effectively invert the six geometric and position parameters of a pair of V-shaped cracks.

[0115] Table 3 V-shaped crack parameter calculation table

[0116]

[0117] Embodiment 2: This embodiment is a device for inverting internal fracture parameters of rock mass based on ground penetrating radar images, including: a radar image acquisition module, a distribution form recognition module and a geometric parameter inversion module, etc.

[0118] In this example, the radar image acquisition module is used to acquire the ground penetrating radar image of the rock mass; the distribution form recognition module is used to identify the cracks on the image as one of the preset crack distribution forms based on the ground penetrating radar B-scan image through a CNN-based image recognition model; the geometric parameter inversion module is used to determine the geometric parameters required to characterize this type of crack based on the distribution form of the crack, and invert the geometric parameter values ​​of the corresponding type of the crack by combining the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image.

[0119] In this embodiment, the image feature I includes the number of acquisition channels corresponding to the cracks in the image; the image feature II includes the time difference between the peaks in the image.

[0120] Embodiment 3: This embodiment is a storage medium on which a computer program that can be executed by a processor is stored. When the computer program is executed, the steps of the method in Embodiment 1 are implemented.

[0121] Embodiment 4: A data processing device comprises a memory and a processor, wherein the memory stores a computer program executable by the processor, and the steps of the method in Embodiment 1 are implemented when the computer program is executed.

Claims

1. A method for inverting internal fracture parameters of rock mass based on ground penetrating radar images, characterized in that: include: Acquire ground penetrating radar images of rock masses; Based on the ground penetrating radar B-scan image, the cracks on the image are identified as one of the preset crack distribution forms; Based on the distribution form of the cracks, the geometric parameters required to characterize this type of cracks are determined, and the geometric parameter values ​​of the corresponding type of cracks are inverted by combining the image features Ⅰ of the ground penetrating radar B-scan image and the image features Ⅱ of the ground penetrating radar A-scan image; The image feature I includes the number of acquisition channels corresponding to the cracks in the image; the image feature II includes the time difference between the peaks in the image.

2. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 1, characterized in that: The method of identifying the cracks on the image as one of the preset crack distribution forms based on the ground penetrating radar B-scan image includes: The ground penetrating radar B-scan image is input into the CNN-based image recognition model to identify the crack distribution pattern corresponding to the B-scan image.

3. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 1, characterized in that: The method of determining the geometric parameters required to characterize the cracks based on the distribution form of the cracks includes: Based on the distribution form of the cracks, the geometric parameters required to characterize the cracks are determined by comparing the preset distribution form and geometric parameter correspondence table.

4. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 1, characterized in that: The preset crack distribution forms include: a single crack parallel to the detection surface, a single crack perpendicular to the detection surface, and a single crack inclined to the detection surface; Also, a pair of parallel cracks parallel to the detection surface, a pair of parallel cracks perpendicular to the detection surface, a pair of parallel cracks inclined to the detection surface, a pair of orthogonal cracks, a V-shaped crack opening upward, and a V-shaped crack opening downward.

5. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 4, characterized in that: The method of determining the geometric parameters required to characterize the cracks based on the distribution form of the cracks, and inverting the geometric parameter values ​​corresponding to the cracks in combination with the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image includes: For a pair of parallel cracks parallel to the detection surface, the corresponding geometric parameters include: the horizontal coordinate x1 of the left end point of the first crack, the vertical coordinate y1 of the left end point of the first crack, the length l1 of the first crack, the horizontal coordinate x2 of the left end point of the second crack, the distance Δd between the two cracks, and the length l2 of the second crack; l1=Δn1×Δl l2=Δn2×Δl x1=n1×Δl+x i x2=n2×Δl+x i Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the medium, t1 is the time difference between the ground direct wave and the top of the first bowl-shaped curve, Δt is the time difference between the tops of the two bowl-shaped curves, n1 and n2 are the acquisition channel numbers corresponding to the left ends of the two bowl-shaped hyperbolic straight segments in the image, Δl is the channel spacing, and x i is the initial position of the transmitting antenna.

6. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 4, characterized in that: The method of determining the geometric parameters required to characterize the cracks based on the distribution form of the cracks, and inverting the geometric parameter values ​​corresponding to the cracks in combination with the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image includes: For a pair of orthogonally distributed cracks, the corresponding geometric parameters include: the horizontal coordinate x1 of the left end point of the horizontal crack, the vertical coordinate y1 of the left end point of the horizontal crack, the length l1 of the horizontal crack, the horizontal coordinate x2 of the top of the vertical crack, the vertical coordinate y2 of the top of the vertical crack, and the length l2 of the vertical crack; l1=Δn1×Δl x1=n1×Δl+x i x2=n2×Δl+x i l2=Δn2×Δl Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the medium, t1 is the time difference between the ground direct wave and the top of the bowl-shaped curve in the image, t2 is the time difference between the ground direct wave and the top of the high and narrow hyperbola in the image, Δn1 is the number of acquisition channels spanned by the straight section of the bowl-shaped hyperbola in the image, Δl is the channel spacing, n1 is the number of acquisition channels corresponding to the left end of the straight section of the bowl-shaped hyperbola in the image, and n2 is the number of acquisition channels corresponding to the top of the high and narrow hyperbola in the image; x i is the initial position of the transmitting antenna, and Δn2 represents the number of acquisition channels spanned by the horizontal segment of the bowl-shaped hyperbola in the B-scan image in the other direction perpendicular to the original detection plane.

7. The method for inverting internal fracture parameters of rock mass based on ground penetrating radar images according to claim 4, characterized in that: The method of determining the geometric parameters required to characterize the cracks based on the distribution form of the cracks, and inverting the geometric parameter values ​​corresponding to the cracks in combination with the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image includes: For a V-shaped crack opening upward or a V-shaped crack opening downward, the corresponding geometric parameter types include: the abscissa x1 of the endpoint A, the abscissa x2 of the endpoint B, the abscissa x3 of the vertex C, the ordinate y1 of the endpoint A, the ordinate y2 of the endpoint B, and the ordinate y3 of the vertex C; x1=n1×Δl+x i x2=n2×Δl+x i x3=n3×Δl+x i Where c is the propagation speed of electromagnetic waves in the air, ε r is the relative dielectric constant of the medium, n1, n2, n3 are the number of acquisition channels corresponding to the three arc-shaped turning points in the image, Δl is the channel spacing, x i is the initial position of the transmitting antenna, and t1, t2, and t3 are the time differences between the three bending points in the image and the direct wave from the ground.

8. A device for inverting internal fracture parameters of rock mass based on ground penetrating radar images, characterized in that: include: A radar image acquisition module, used to acquire ground penetrating radar images of rock masses; A distribution form recognition module, used for identifying the corresponding crack as one of the preset crack distribution forms based on the ground penetrating radar B-scan image; The geometric parameter inversion module is used to determine the geometric parameters required to characterize this type of cracks based on the distribution form of the cracks, and invert the corresponding geometric parameter values ​​of this type of cracks by combining the image feature I of the ground penetrating radar B-scan image and the image feature II of the ground penetrating radar A-scan image; The image feature I includes the number of acquisition channels corresponding to the cracks in the image; the image feature II includes the time difference between the peaks in the image.

9. A storage medium having stored thereon a computer program executable by a processor, characterized in that: When the computer program is executed, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A data processing device comprising a memory and a processor, wherein the memory stores a computer program executable by the processor, wherein: When the computer program is executed, the steps of the method according to any one of claims 1 to 7 are implemented.

Images