Deep rock mass mechanical parameter inversion method based on multi-source investigation information

Through multi-source survey information and deep learning model, combined with Hawke-Brown criterion, efficient and accurate inversion of deep rock mechanic parameters is achieved, and the problem of difficulty in obtaining deep rock parameters in the existing technology is solved, and the problem of the difficulty in accurately obtaining deep rock parameters is improved. Accuracy and safety of tunnel construction.

CN118916962AActive Publication Date: 2024-11-08CHINA STATE RAILWAY GRP CO LTD +2

Patent Information

Application Number
CN202410961829.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-18
Publication Date
2024-11-08
Estimated Expiration
2044-07-18

AI Technical Summary

Technical Problem

It is difficult for the existing technology to accurately obtain the in-situ engineering mechanical parameters of deep rock mass, affecting the accuracy and safety of tunnel construction.

Method used

Using the deep rock mechanics parameter inversion method based on multi-source exploration information, by obtaining rock mechanics, structural and engineering parameters, combining Hawke-Brown criterion and deep learning neural network model, a sample database is constructed, and the inversion model is trained to achieve the goal of the Efficient and accurate inversion of mechanical parameters of deep rock.

Benefits of technology

This method can rely on on-site data to improve the accuracy and diversity of parameters, reduce indoor tests, and provide higher accuracy to guide the subsequent construction of the tunnel, solving the problem that deep rock mass parameters are difficult to accurately obtain.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916962A_ABST
    Figure CN118916962A_ABST
Patent Text Reader

Abstract

The invention relates to a deep rock mass mechanical parameter inversion method based on multi-source investigation information. The deep rock mass mechanical parameter inversion method comprises the steps that rock mechanical parameters, rock mass structure parameters and rock mass engineering parameters are acquired; calculating rock mass mechanical parameters according to the rock mechanical parameters, the rock mass structure parameters and the rock mass engineering parameters based on a Hook-Brown criterion; taking the rock mechanical parameters, the rock mass structure parameters and the rock mass engineering parameters as input information, taking the corresponding rock mass mechanical parameters as output information, constructing a sample database, and training a deep learning neural network model by using the sample database to obtain a deep rock mass mechanical parameter inversion model; and performing inversion by using the deep rock mass mechanical parameter inversion model to obtain deep rock mass engineering mechanical parameters. By means of the deep rock mass mechanical parameter inversion method, the rock mass mechanical parameters can be rapidly and accurately obtained, and high practicability is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of tunnel engineering, and in particular to a deep rock mass mechanical parameter inversion method based on multi-source exploration information. Background Art

[0002] my country has become the country in the world with the largest scale of highway tunnel construction, the fastest construction efficiency, the largest number of operating tunnels, and the most complex structural forms. It has made milestone developments in tunnel structure design, lining disease monitoring, and key maintenance technologies.

[0003] With the proposal and application of the New Austrian Tunneling Method, my country's tunnel engineering construction technology has been greatly developed, and the scale of tunnel construction has become larger and larger. The large amount of field monitoring data generated has provided researchers with another way to determine rock mass parameters, namely the inversion analysis method.

[0004] However, the problems faced by rock mechanics, such as "limited data discreteness", "complexity of failure mechanism" and "ambiguity of influencing factors", have seriously affected the development of numerical analysis methods. The deep rock mass is a geological body that has existed in a high-stress environment for a long time. Its past history will inevitably affect the current and future behavior of the rock mass. Accurately obtaining the in-situ engineering mechanics parameters of the deep rock mass has important guiding significance for the subsequent construction of the tunnel site. At present, the methods for accurately obtaining the in-situ engineering mechanics parameters of the deep rock mass are still not mature enough. How to quickly and accurately obtain the in-situ mechanical parameters of the deep rock mass, so as to more accurately guide the subsequent excavation and support construction of the tunnel, has become a problem that needs to be solved. Summary of the invention

[0005] In order to at least partially solve the problem that it is difficult to accurately obtain the in-situ mechanical parameters of deep rock mass, the present invention provides a deep rock mass mechanical parameter inversion method based on multi-source exploration information, which can be expected to achieve efficient and accurate acquisition of deep rock mass mechanical parameters.

[0006] The present invention provides a deep rock mass mechanical parameter inversion method based on multi-source exploration information, comprising the following steps:

[0007] S1. Obtain rock mechanics parameters, wherein the rock mechanics parameters include rock uniaxial compressive strength, rock cohesion and rock internal friction angle;

[0008] S2. Obtain rock mass structural parameters, wherein the rock mass structural parameters include a rock mass integrity coefficient;

[0009] S3, obtaining rock mass engineering parameters, wherein the rock mass engineering parameters include tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient;

[0010] S4, using a calculation model of the Hawke-Brown criterion to calculate corresponding rock mass mechanical parameters according to the rock mechanical parameters, rock mass structural parameters and rock mass engineering parameters, wherein the rock mass mechanical parameters include uniaxial compressive strength of the rock mass, elastic modulus of the rock mass, cohesion of the rock mass and internal friction angle of the rock mass;

[0011] S5. Using rock mechanics parameters, rock structure parameters and rock engineering parameters as input information and corresponding rock mechanics parameters as output information, construct a sample database and establish a deep learning neural network model, and use the sample database to train the model to obtain a deep rock mechanics parameter inversion model;

[0012] S6. Using the deep rock mass mechanical parameter inversion model to invert the deep rock mass mechanical parameters.

[0013] As a preferred method, in step S4, the rock mass elastic modulus E, rock mass cohesion C, rock mass internal friction angle φ, rock mass uniaxial compressive strength σ are calculated by the following calculation model: c ;

[0014]

[0015]

[0016] In the above formulas, D is the explosion disturbance coefficient, σ UCS is the uniaxial compressive strength of rock, K4 is the correction factor for groundwater influence, K v is the rock mass integrity coefficient, γ is the rock mass density; H is the tunnel depth; m i is the rock mechanics parameter; SCF is the strength correction factor, SCF = 22.25K v -10K4-74.725; p is the intermediate variable for calculation, p=SCF+0.27σ UCS ; GSI is the geomechanical strength index.

[0017] As a preferred example, the rock mechanics parameter m i Calculate using the following formula:

[0018]

[0019] Where: UCS is the uniaxial compressive strength of rock, c1 is the cohesion of rock, σ is the confining pressure, θ is the internal friction angle of rock, when f(σ,m i ) takes the minimum value, the obtained m i It is the rock mechanics parameter.

[0020] Preferably, the geomechanical strength index GSI is quantified by the following formula:

[0021] GSI=0.089(100+3σ UCS +250K V -100K4+0.5θ-4c1)+10.375

[0022] Where: UCS is the uniaxial compressive strength of rock, c1 is the cohesion of rock, θ is the internal friction angle of rock, K V is the rock mass integrity coefficient, and K4 is the groundwater influence correction coefficient.

[0023] Preferably, in step S1, for hard rock, the rock mechanical parameters are obtained based on measurement while drilling; for soft rock, the rock mechanical parameters are obtained through dynamic compression shear test.

[0024] Preferably, when obtaining the rock mechanical parameters based on downhole measurement, the borehole is divided into units along the downhole direction, and the rock mechanical parameters of the corresponding unit are obtained based on the downhole test parameters of 0.5 m within the range every 20 cm of drilling, and the cycle is repeated to obtain rock mechanical parameter data at different drilling depths.

[0025] Preferably, in step S2, an image of the rock mass in the hole is obtained based on the in-hole television, and the image is extracted and analyzed to obtain rock mass structural surface parameters including structural surface crack density, and the rock mass integrity coefficient is calculated based on the structural surface crack density;

[0026] For every 20cm of drilling, an extraction and analysis is performed based on the digital image of 0.5m within the range to obtain a rock integrity coefficient to ensure strict correspondence between the rock integrity coefficient and the rock mechanics parameters.

[0027] Preferably, in step S5, the deep learning neural network model adopts a fully connected layer neural network model, which consists of 3 fully connected layers, a ReLU activation function is used after each fully connected layer, and the network contains a skip connection;

[0028] The model input characteristics include at least the uniaxial compressive strength of rock, rock cohesion, rock internal friction angle, rock integrity coefficient, tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient. The number of model output parameters is 4, namely: uniaxial compressive strength of rock, rock elastic modulus, rock internal friction angle and rock cohesion. The loss function is composed of 4 parameters, and the coefficients are 0.2, 0.2, 0.5 and 0.1 respectively.

[0029] Preferably, the method further includes a step of verifying the accuracy of the deep rock mass mechanical parameter inversion model:

[0030] For specific work points, actual displacement monitoring is carried out to record the three-dimensional displacement deformation data of the arch crown, spandrel and side wall to obtain the actual monitoring displacement curve; using a large number of three-dimensional numerical calculations, the displacement deformation of the corresponding monitoring points under different rock mass parameters is recorded to obtain the numerical calculation displacement curve;

[0031] The bee colony algorithm is used to search for the numerically calculated displacement curve that is closest to the actual monitored displacement curve to obtain the corresponding rock mass mechanical parameters; and the parameters are compared with the calculated parameters of the deep rock mass mechanical parameter inversion model obtained based on step S5. If they are incorrect, the displacement curve inversion parameters are added to the sample database and the inversion model is trained again.

[0032] Compared with the related prior art, the beneficial effects of the present invention are at least embodied in:

[0033] All parameters required for the inversion method of the present invention can be measured based on on-site construction, all data are from the site, there are more samples of structural surfaces and rock parameters, the accuracy is higher, and there is no need to carry out a large number of indoor tests.

[0034] In addition, in some implementations, the deformation measured on site is used as the reference value for accuracy verification, and with the support of a large amount of on-site data, it has very high accuracy, solving the problem of the difficulty in accurately obtaining deep in-situ engineering mechanics parameters. It plays an important guiding role in the later construction of the tunnel. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A schematic diagram of a deep rock mass mechanical parameter inversion method according to an embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the unit division in a hole while drilling according to an embodiment of the present invention;

[0037] Figure 3 This is an example data diagram of rock mechanics parameters calculated and obtained in an embodiment of the present invention;

[0038] Figure 4 It is a schematic diagram of inversion parameters of the deep rock mass mechanical parameter inversion model according to an embodiment of the present invention. DETAILED DESCRIPTION

[0039] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] See also Figure 1As shown, a method for inverting deep rock mass mechanical parameters based on multi-source exploration information in an embodiment includes the following steps:

[0041] S1. Obtain rock mechanics parameters, wherein the rock mechanics parameters include rock uniaxial compressive strength, rock cohesion and rock internal friction angle;

[0042] S2. Obtain rock mass structural parameters, wherein the rock mass structural parameters include a rock mass integrity coefficient;

[0043] S3, obtaining rock mass engineering parameters, wherein the rock mass engineering parameters include tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient;

[0044] S4, using a calculation model of the Hawke-Brown criterion to calculate corresponding rock mass mechanical parameters according to the rock mechanical parameters, rock mass structural parameters and rock mass engineering parameters, wherein the rock mass mechanical parameters include uniaxial compressive strength of the rock mass, elastic modulus of the rock mass, cohesion of the rock mass and internal friction angle of the rock mass;

[0045] S5. Using rock mechanics parameters, rock structure parameters and rock engineering parameters as input information and corresponding rock mechanics parameters as output information, construct a sample database and establish a deep learning neural network model, and use the sample database to train the model to obtain a deep rock mechanics parameter inversion model;

[0046] S6. Using the deep rock mass mechanical parameter inversion model to invert the deep rock mass mechanical parameters.

[0047] It can be understood that the deep rock mechanical parameters inversion method based on multi-source exploration information provided in this embodiment can effectively invert the deep rock mechanical parameters based on multi-source data such as in-situ mechanical parameters and three-dimensional rock structure more accurately and efficiently.

[0048] In some embodiments, in step S4, the rock mass elastic modulus E, rock mass cohesion C, rock mass internal friction angle φ and rock mass uniaxial compressive strength σ are calculated by the following calculation model: c ;

[0049]

[0050] In the above formulas, D is the explosion disturbance coefficient, σ UCS is the uniaxial compressive strength of rock, K4 is the correction factor for groundwater influence, K v is the rock mass integrity coefficient, γ is the rock mass density; H is the tunnel depth; m i is the rock mechanics parameter; SCF is the strength correction factor, SCF = 22.25K v -10K4-74.725; p is the intermediate variable for calculation, p=SCF+0.27σ UCS; GSI is the geomechanical strength index.

[0051] It can be understood that the calculation model of the improved Hawke-Brown criterion is adopted in this embodiment, which makes the calculation of the mechanical parameters of the rock mass more accurate and reasonable, and can more accurately reflect the mechanical parameters of the deep rock mass.

[0052] In some embodiments, the rock mechanical parameter m in the Hawke-Brown criterion i Calculate using the following formula:

[0053]

[0054] Where: UCS is the uniaxial compressive strength of rock, C1 is the cohesion of rock, σ is the confining pressure, θ is the internal friction angle of rock, when f(σ,m i ) takes the minimum value, the obtained m i It is the rock mechanics parameter.

[0055] It is understandable that the rock mechanics parameter m in the Hoek-Brown criterion is i It is a parameter used to describe the hardness of intact rock. This parameter reflects the hardness of intact rock. In traditional applications, it is used as an empirical parameter. Usually, the corresponding value can be obtained according to the value table of "Technical Specifications for Slope Engineering of Non-coal Open-pit Mines" (GB 51016-2014). However, the value of the empirical formula based on the value table is inevitably inaccurate. In this embodiment, the rock mechanics parameter m i The value can be calculated and determined based on field test data with higher accuracy.

[0056] In some embodiments, the geomechanical strength index GSI is quantified by the following formula:

[0057] GSI=0.089(100+3σ UCS +250K V -100K4+0.5θ-4c1)+10.375

[0058] Where: UCS is the uniaxial compressive strength of rock, c1 is the cohesion of rock, θ is the internal friction angle of rock, K V is the rock mass integrity coefficient, and K4 is the groundwater influence correction coefficient.

[0059] In some embodiments, in step S1, for hard rock, the rock mechanical parameters can be obtained based on the measurement while drilling; for soft rock, the rock mechanical parameters are obtained through dynamic compression and shear test. In this embodiment, different test methods are adopted to obtain rock mechanical parameters for different types of rock mass properties, so that the obtained rock mechanical parameters are more accurate and reasonable. It can be understood that the measurement while drilling technology uses various parameters that are monitored and recorded in real time during the drilling process. These parameters include drilling speed, torque, thrust, etc. These data can be obtained through sensors installed on the drill bit or drill rod. By analyzing these drilling parameters, the structural characteristics and mechanical properties of the rock mass can be identified. Specifically, in this embodiment, the rock mechanical parameters can be obtained by analyzing the pressure, torque, speed and vibration parameters of the drill rod during the drilling process. Similarly, obtaining rock mechanical parameters through dynamic compression and shear test is a method that can be adopted by technicians in this field, but it is not limited to this and will not be repeated here.

[0060] As a preferred embodiment, see Figure 2 As shown, when obtaining the rock mechanical parameters based on the measurement while drilling, the borehole is divided into units along the direction of the drilling. Every 20cm of drilling, the rock mechanical parameters of the corresponding unit are calculated based on the parameters of the drilling test within the range of 0.5m, and the cycle is repeated to obtain the rock mechanical parameter data at different drilling depths. More specifically, in order to obtain more effective sample data within a reasonable range, when dividing the borehole into units, the unit size of the drilling calculation model can be set to 1m×1m×0.7m, with a step size of 0.2m, so that 100 units can be obtained at a depth of 20m in the borehole. Figure 3 The rock mechanical parameters calculated by the drilling test are listed as examples.

[0061] It is worth noting that based on the solution of this embodiment, the rock mechanical parameters that change with the depth of the borehole can be obtained by relying on the on-site drilling test parameters: rock uniaxial compressive strength, cohesion (internal cohesion) and internal friction angle. In this embodiment, the target rock is directly tested in its actual state. Compared with laboratory testing after sampling, the test data is more real and reliable, and the data obtained are rock mechanical parameters that change with the depth of the borehole, which are more representative and diverse, and conform to the state in the real scene.

[0062] In some embodiments, in step S2, an image of the rock mass in the hole is obtained based on the in-hole television, and the image is extracted and analyzed to obtain rock mass structural surface parameters including structural surface crack density, and the rock mass integrity coefficient is calculated based on the structural surface crack density. It should be noted that the rock mass structural surface parameters can be determined based on the in-hole television combined with stereological theory; specifically, a 360° image in the borehole can be obtained based on the in-hole television technology, and the rock mass structural surface parameters can be determined using an automatic structural surface recognition method based on stereological theory and deep learning image processing. The rock mass structural surface parameters that can be identified include the average inclination of rock mass structural cracks, the average inclination angle, the number of cracks, the rock mass structural crack density, and the rock mass integrity coefficient. The specific principle process will not be repeated here. Of course, other measurement methods can also be used in this embodiment to determine the rock mass integrity coefficient.

[0063] Furthermore, the boreholes are divided into units to obtain more structural surface parameter data. That is, every time 20 cm is drilled, an extraction and analysis is performed based on the digital image of 0.5 m within the range to obtain a rock integrity coefficient, so as to achieve a strict correspondence between the rock integrity coefficient and the rock mechanics parameters.

[0064] In some implementations, different values ​​of the blasting disturbance coefficient D may be selected according to different blasting methods actually used in tunnel excavation. For specific values, refer to Table 1:

[0065] Table 1

[0066] Excavation method TMB Controlled blasting Manual, mechanical Low quality blasting Hard Rock 0 0.1 0.2 0.8 Soft Rock 0 0.2 0.5 1

[0067] In some embodiments, the groundwater impact correction coefficient K4 can be calculated based on the groundwater development degree of the slope and the rock mass integrity coefficient K V The value range of K4 is 0-1. The better the rock mass integrity and the worse the groundwater development, the smaller the K4 value. The specific value of K4 can be found in Table 2:

[0068] Table 2

[0069]

[0070]

[0071] In some embodiments, in step S5, the deep learning neural network model adopts a fully connected layer neural network model, which is composed of 3 fully connected layers, each fully connected layer is followed by a ReLU activation function, and the network contains a jump connection; the model input features (input information) include at least the uniaxial compressive strength of rock, rock cohesion, rock internal friction angle, rock mass integrity coefficient K V, tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient. The number of model output parameters is 4, namely: uniaxial compressive strength of rock mass, elastic modulus of rock mass, internal friction angle of rock mass and rock mass cohesion. The loss function is composed of 4 parameters, and the coefficients are 0.2, 0.2, 0.5 and 0.1 respectively.

[0072] See also Figure 4 As shown, in some embodiments, the input information of the inversion model mainly includes rock parameters, engineering parameters and rock mass structure parameters, wherein the rock parameters include rock uniaxial strength (rock uniaxial compressive strength), rock cohesion, and rock internal friction angle; the rock mass structure parameters include rock mass integrity coefficient K V ,The engineering parameters include tunnel burial depth, blasting intensity (blasting disturbance coefficient) and groundwater status (groundwater influence correction coefficient).,The output information of the inversion model includes rock mass parameters, specifically: uniaxial compressive strength of rock mass, rock mass elastic modulus, rock mass internal friction angle and rock mass cohesion.

[0073] In some embodiments, the method further includes a step of verifying the accuracy of the deep rock mass mechanical parameter inversion model:

[0074] Carry out actual displacement monitoring for specific work points, record the three-dimensional displacement deformation data of the arch crown, spandrel and side wall, and obtain the actual monitoring displacement curve; use a large number of three-dimensional numerical calculations (for example, use 3DEC program to carry out a large number of three-dimensional numerical calculations) to record the displacement deformation of the corresponding monitoring points under different rock mass parameters, and obtain the numerical calculation displacement curve;

[0075] The bee colony algorithm is used to search for the numerically calculated displacement curve that is closest to the actual monitored displacement curve to obtain the corresponding rock mass mechanical parameters; and the parameters are compared with the calculated parameters of the deep rock mass mechanical parameter inversion model obtained based on step S5. If they are incorrect, the displacement curve inversion parameters are added to the sample database and the inversion model is trained again to further improve the inversion accuracy and effectiveness of the inversion model; otherwise, if they are correct, it further proves that the inversion model can more accurately and effectively invert the real deep rock mass mechanical parameters.

[0076] In the description of the embodiments of the present invention, specific features, structures, materials or characteristics may be combined in a suitable manner in any one or more embodiments or examples.

[0077] References to "one embodiment" or "some embodiments" etc. described in the specification of this application mean that one or more embodiments of the present application include specific features, structures or characteristics described in conjunction with the embodiment. Therefore, the statements "in one embodiment", "in some embodiments", "in other embodiments", etc. that appear in different places in this specification do not necessarily refer to the same embodiment, but mean one or more but not all embodiments, unless otherwise specifically emphasized in other ways. The terms "including", "comprising", "having" and their variations all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0078] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A deep rock mass mechanical parameter inversion method based on multi-source exploration information, characterized in that: The following steps are involved: S1. Obtain rock mechanics parameters, wherein the rock mechanics parameters include rock uniaxial compressive strength, rock cohesion and rock internal friction angle; S2. Obtain rock mass structural parameters, wherein the rock mass structural parameters include a rock mass integrity coefficient; S3, obtaining rock mass engineering parameters, wherein the rock mass engineering parameters include tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient; S4, using a calculation model of the Hawke-Brown criterion to calculate corresponding rock mass mechanical parameters according to the rock mechanical parameters, rock mass structural parameters and rock mass engineering parameters, wherein the rock mass mechanical parameters include uniaxial compressive strength of the rock mass, elastic modulus of the rock mass, cohesion of the rock mass and internal friction angle of the rock mass; S5. Using rock mechanics parameters, rock structure parameters and rock engineering parameters as input information and corresponding rock mechanics parameters as output information, construct a sample database and establish a deep learning neural network model, and use the sample database to train the model to obtain a deep rock mechanics parameter inversion model; S6. Using the deep rock mass mechanical parameter inversion model to invert the deep rock mass mechanical parameters.

2. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 1 is characterized by: In step S4, the rock mass elastic modulus E, rock mass cohesion C, rock mass internal friction angle φ and rock mass uniaxial compressive strength σ are calculated by the following calculation model: c ; In the above formulas, D is the explosion disturbance coefficient, σ UCS is the uniaxial compressive strength of rock, K4 is the correction factor for groundwater influence, K v is the rock mass integrity coefficient, γ is the rock mass density; H is the tunnel depth; m i is the rock mechanics parameter; SCF is the strength correction factor, SCF = 22.25K v -10K4-74.725; p is the intermediate variable for calculation, p=SCF+0.27σ UCS ; GSI is the geomechanical strength index.

3. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 2 is characterized by: Rock mechanical parameters m i Calculate using the following formula: Where: UCS is the uniaxial compressive strength of rock, C1 is the cohesion of rock, σ is the confining pressure, θ is the internal friction angle of rock, when f(σ,m i ) takes the minimum value, the obtained m i It is the rock mechanics parameter.

4. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 2 is characterized in that: The geomechanical strength index GSI is quantified by the following formula: GSI=0.089(100+3σ UCS +250K V -100K4+0.5θ-4c1)+10.375 Where: UCS is the uniaxial compressive strength of rock, c1 is the cohesion of rock, θ is the internal friction angle of rock, K V is the rock mass integrity coefficient, and K4 is the groundwater influence correction coefficient.

5. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 2 is characterized by: In the step S1, for hard rock, the rock mechanical parameters are obtained based on measurement while drilling; for soft rock, the rock mechanical parameters are obtained through dynamic compression shear test.

6. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 5 is characterized by: When obtaining the rock mechanical parameters based on the measurement while drilling, the borehole is divided into units along the drilling direction. Every time 20 cm is drilled, the rock mechanical parameters of the corresponding unit are obtained based on the drilling test parameters of 0.5 m within this range. The cycle is repeated to obtain rock mechanical parameter data at different drilling depths.

7. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 6 is characterized by: In the step S2, an image of the rock mass in the hole is obtained based on the in-hole television, and the image is extracted and analyzed to obtain rock mass structural surface parameters including structural surface crack density, and the rock mass integrity coefficient is calculated based on the structural surface crack density; For every 20cm of drilling, an extraction and analysis is performed based on the digital image of 0.5m within the range to obtain a rock integrity coefficient to ensure strict correspondence between the rock integrity coefficient and the rock mechanics parameters.

8. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 1 is characterized by: In step S5, the deep learning neural network model adopts a fully connected layer neural network model, which consists of 3 fully connected layers, a ReLU activation function is used after each fully connected layer, and a skip connection is included in the network; The model input characteristics include at least the uniaxial compressive strength of rock, rock cohesion, rock internal friction angle, rock integrity coefficient, tunnel burial depth, blasting disturbance coefficient and groundwater influence correction coefficient. The number of model output parameters is 4, namely: uniaxial compressive strength of rock, rock elastic modulus, rock internal friction angle and rock cohesion. The loss function is composed of 4 parameters, and the coefficients are 0.2, 0.2, 0.5 and 0.1 respectively.

9. The deep rock mass mechanical parameter inversion method based on multi-source exploration information according to claim 1 is characterized by: The method also includes the following steps to verify the accuracy of the inversion model of the deep rock mass mechanical parameters: Carry out actual displacement monitoring for specific work points, record the three-dimensional displacement deformation data of the vault, spandrel and side wall, and obtain the actual monitoring displacement curve; Using a large number of three-dimensional numerical calculations, the displacement and deformation of the corresponding monitoring points under different rock mass parameters are recorded to obtain the numerically calculated displacement curves; The bee colony algorithm is used to search for the numerically calculated displacement curve that is closest to the actual monitored displacement curve to obtain the corresponding rock mass mechanical parameters; and the parameters are compared with the calculated parameters of the deep rock mass mechanical parameter inversion model obtained based on step S5. If they are incorrect, the displacement curve inversion parameters are added to the sample database and the inversion model is trained again.

Citation Information

Patent Citations

  • Method for determining jointed rock mechanical parameters

    CN103175730A

  • Rock mass mechanical parameter back analysis method

    CN111414658A

  • Real-time inversion method for creep parameters of surrounding rock mass

    CN111666671A

  • Automatic inversion method for joint development tunnel surrounding rock mechanical parameters

    CN114169238A

  • Deep learning-based tunnel mechanical property rapid inversion method

    CN114611193A

Cited By

  • Surrounding rock mechanical parameter inversion method

    CN119962039A

  • Geotechnical engineering investigation safety monitoring system and method based on artificial intelligence

    CN120509248A

  • Underground engineering surrounding rock surface and inner structure identification method

    CN121640443A