Method and system for acquiring REV numerical simulation of jointed rock mass based on image recognition

Abstract

The present disclosure relates to the technical field of rock mass numerical simulation, and proposes a method and system for obtaining the REV numerical simulation of jointed rock mass based on image recognition, which can automatically obtain the occurrence information of the jointed rock mass and ensure the accuracy of the jointed rock mass modeling. The image of the jointed rock mass is identified by a convolutional neural network, which simplifies the acquisition of structural surface information, improves the accuracy and efficiency of structural surface acquisition, and establishes a multi-scale synthetic rock mass model based on the structural surface information obtained by image recognition, accurately characterizing the properties of the jointed rock mass. The model size-mechanical property parameter change curve is drawn, and the size of the characterization unit body of the jointed rock mass is determined by using the difference ratio, which improves the determination of the characterization unit body and realizes the quantification and standardization of the characterization unit body acquisition method. It provides a basis for simulating non-continuous media using continuous medium mechanics calculation methods.

Description

Technical Field

[0001] The present disclosure relates to the technical field related to rock mass numerical simulation, and in particular, to a method and system for acquiring REV numerical simulation of jointed rock mass based on image recognition. Background Art

[0002] The statements in this section merely provide background information related to the present disclosure and do not necessarily constitute prior art.

[0003] Tunnels are located in a rock mass environment, which is quite different from the environment of ground buildings. Therefore, it is very important to explore the mechanical properties and deformation mechanism of rock mass. Rock mass is composed of discontinuities and complete rock blocks, and the mechanical properties of rock mass are also determined by these two. Among them, discontinuities include joints, cracks, bedding and faults. Due to the ever-changing discontinuities, the properties of rock mass are also extremely complex. Jointed rock mass belongs to non-continuous medium, and its mechanical properties change with the increase of rock mass size. This is the size effect of jointed rock mass. When the volume of rock mass increases to a critical value, the mechanical properties of jointed rock mass tend to be stable. This value is the characterization unit volume (REV) size of jointed rock mass. The mechanical properties of jointed rock mass larger than the characterization unit volume size remain basically unchanged. At this time, the jointed rock mass can be regarded as a continuous medium, and its corresponding mechanical parameters can be used in the calculation of continuous medium mechanics. Therefore, determining the characterization unit volume of jointed rock mass is the basis for simulating jointed rock mass using continuous medium mechanics calculation methods (finite element method, finite difference method, etc.), which has important research significance.

[0004] The inventors found in their research that the current methods for determining characterization units mostly use numerical simulation methods. This method is not limited by the rock mass environment and can use computers to simulate the mechanical test process of engineering-scale rock masses, which has certain research advantages. Before simulation, an accurate jointed rock mass model needs to be established, but obtaining the joint distribution in the real rock mass is a major problem. Currently, rock mass images are mostly captured by line surveying and laser scanning methods, and jointed rock masses are manually identified and occurrence information is determined, which is subjective and consumes a lot of cost and time. Summary of the invention

[0005] In order to solve the above problems, the present disclosure proposes a method and system for obtaining the REV of jointed rock mass based on image recognition, which can automatically obtain the occurrence information of the jointed rock mass and ensure the accuracy of the jointed rock mass modeling, thereby laying the foundation for obtaining the REV of the jointed rock mass.

[0006] In order to achieve the above objectives, the present disclosure adopts the following technical solutions:

[0007] One or more embodiments provide a method for obtaining REV numerical simulation of jointed rock mass based on image recognition, comprising the following steps:

[0008] The acquired jointed rock mass image to be identified is used to identify the image features of the jointed rock mass through a trained convolutional neural network to obtain the occurrence parameters of the joints;

[0009] The discrete element method is used to establish a multi-scale jointed rock model based on the occurrence parameters, and the mechanical property parameters of the multi-scale jointed rock model are obtained by simulating the indoor test process.

[0010] According to the obtained mechanical property parameters, the model size-mechanical property parameter curve is drawn to obtain the REV size of the jointed rock mass and the corresponding mechanical parameters.

[0011] One or more embodiments provide a jointed rock mass REV numerical simulation acquisition system based on image recognition, an image acquisition device and a processor;

[0012] The image acquisition device is used to acquire images of jointed rock mass;

[0013] The processor is configured to execute the steps of the above method.

[0014] An electronic device comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps described in the above method are completed.

[0015] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the steps described in the above method are completed.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] In the present disclosure, the image of the jointed rock mass is identified by a convolutional neural network, the work of obtaining structural surface information is simplified, the accuracy and efficiency of obtaining structural surface information are improved, and a multi-scale synthetic rock mass model is established based on the structural surface information obtained by image recognition, and the properties of the jointed rock mass are accurately characterized. The model size-mechanical property parameter change curve is drawn, and the size of the characterization unit body of the jointed rock mass is determined by using the difference ratio, which improves the determination of the characterization unit body and realizes the quantification and standardization of the characterization unit body acquisition method. It provides a basis for simulating non-continuous media using continuous medium mechanics calculation methods.

[0018] The advantages of the present disclosure and the advantages of additional aspects will be described in detail in the following specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The accompanying drawings constituting a part of the present disclosure are used to provide a further understanding of the present disclosure. The exemplary embodiments of the present disclosure and the description thereof are used to explain the present disclosure but do not constitute a limitation of the present disclosure.

[0020] Figure 1 is a block diagram of a jointed rock mass REV numerical simulation acquisition system according to Embodiment 2 of the present disclosure;

[0021] Figure 2 is a schematic diagram of the structure of the synthetic rock model of Example 1 of the present disclosure;

[0022] Figure 3 is a method flow chart of a method for obtaining REV numerical simulation of jointed rock mass according to Example 1 of the present disclosure;

[0023] Figure 4 is an engineering scale fracture discrete network model diagram of Example 1 of the present disclosure;

[0024] Figure 5 is a multi-scale discrete fracture network model diagram of Example 1 of the present disclosure;

[0025] Figure 6 is a multi-scale synthetic rock model diagram of Example 1 of the present disclosure;

[0026] Figure 7 is a schematic diagram of mechanical test simulation of Example 1 of the present disclosure;

[0027] Figure 8 is a schematic diagram of a method for obtaining a characterization unit body according to Embodiment 1 of the present disclosure;

[0028] Fig. 9 It is a schematic diagram of the finite element simulation of the tunnel excavation process of Example 1 of the present disclosure. DETAILED DESCRIPTION

[0029] The present disclosure is further described below in conjunction with the accompanying drawings and embodiments.

[0030] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present disclosure belongs.

[0031] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present disclosure. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof. It should be noted that, in the absence of conflict, the various embodiments in the present disclosure and the features in the embodiments can be combined with each other. The embodiments will be described in detail below in conjunction with the accompanying drawings.

[0032] Example 1

[0033] In the technical solutions disclosed in one or more embodiments, Figure 3-Figure 9 As shown, a method for obtaining REV numerical simulation of jointed rock mass based on image recognition includes the following steps:

[0034] Step 1: The acquired jointed rock mass image to be identified is used to identify the image features of the jointed rock mass through a trained convolutional neural network to obtain the occurrence parameters of the joints;

[0035] Step 2: Using the discrete element method, a multi-scale jointed rock model is established based on the occurrence parameters, and the indoor test process is simulated to test and obtain the mechanical property parameters of the multi-scale jointed rock model;

[0036] Step 3: Based on the obtained mechanical property parameters, draw the model size-mechanical property parameter curve to obtain the REV size of the jointed rock mass and the corresponding mechanical parameters.

[0037] In this embodiment, a convolutional neural network is used to identify the image of the jointed rock mass, simplify the work of obtaining structural surface information, improve the accuracy and efficiency of structural surface acquisition, and establish a multi-scale synthetic rock model based on the structural surface information obtained by image recognition, accurately characterizing the properties of the jointed rock mass. The model size-mechanical property parameter change curve is drawn, and the size of the characterization unit body of the jointed rock mass is determined by using the difference ratio, which improves the determination of the characterization unit body and realizes the quantification and standardization of the characterization unit body acquisition method. It provides a basis for simulating non-continuous media (such as jointed rock mass, etc.) using continuous medium mechanics calculation methods (such as finite element method, finite difference method, etc.).

[0038] In step 1, the jointed rock mass image is acquired by using an image sampling device.

[0039] Optionally, the image sampling device may include a lighting device, a laser pointer and a camera device; the camera device may be a camera or a digital camera.

[0040] The lighting device can be placed near the tunnel face to illuminate the jointed rock surface to be collected and provide a good lighting environment. The jointed rock surface is illuminated with a laser pointer, and the laser point is used to calibrate the collection area for the digital camera, such as a 10×10m rectangular area. The jointed rock image is collected with a digital camera and the image file is saved.

[0041] Step 1 also includes a step of preprocessing the acquired jointed rock mass image, and the preprocessing includes image rotation, grayscale processing, etc.

[0042] Optionally, the image angles can be rotated to make all jointed rock mass images have the same angle. The pixels of all images can be unified by adjusting the image size.

[0043] Convert the RGB image file to grayscale. The calculation formula can be:

[0044] Gray = (299 + 587 + 114 + 500) / 1000

[0045] Where Gray is grayscale, R is red, G is green, and B is blue.

[0046] In step 1, the image features of the jointed rock mass are identified by the trained convolutional neural network, including:

[0047] Step 11: Perform convolution, pooling, and connection operations on the preprocessed image to filter out irrelevant information in the image and extract joint image features;

[0048] Step 12, nonlinearly combining the obtained joint image features to obtain integrated high-order information;

[0049] Step 13: assign classification labels according to the high-order information of the jointed rock mass image to obtain the occurrence information of the structural surface of the jointed rock mass;

[0050] The information on the occurrence of the structural surface of the jointed rock mass may include information such as the dip, inclination, trace length and spacing of the jointed rock mass.

[0051] Optionally, the structure of the convolutional neural network may include an input layer, a convolution layer, a pooling layer, a fully connected layer, and an output layer. The input layer is used to input the preprocessed jointed rock mass image; the convolution layer is used to extract the joint image features; the pooling layer is used to select features and filter irrelevant information; the fully connected layer nonlinearly combines the image features obtained by the convolution layer and the pooling layer to obtain integrated high-order information; the output layer assigns classification labels to the high-order information obtained by the fully connected layer, and transmits the label information to the output;

[0052] This embodiment uses a convolutional neural network to identify the image of the jointed rock mass and obtain the occurrence parameters of the structural surface, providing basic information for jointed rock mass modeling. At the same time, it makes up for the subjectivity and inaccuracy of the line survey method and three-dimensional laser scanning, simplifies the work of obtaining structural surface information, and improves the accuracy and efficiency of structural surface acquisition.

[0053] Furthermore, it also includes the process of training the convolutional neural network, as follows:

[0054] Step S11, acquiring a jointed rock mass image, and performing preprocessing to obtain a jointed rock mass image set;

[0055] The preprocessing steps are the same as those in the recognition stage and will not be described in detail here.

[0056] Step S12, construct a feature data set in which jointed rock mass images correspond one to one with the occurrence labels according to the structural surface occurrence labels of the images in the image set; optionally, 70% of the feature data set can be used as a training set and the remaining 30% of the feature data set can be used as a test set.

[0057] Among them, the occurrence mark labels may include: dip, inclination, trace length and spacing.

[0058] Step S13, using the preprocessed jointed rock mass image as an input sample and the structural surface occurrence information as an output sample, inputting them into a convolutional neural network for training, obtaining a convolutional neural network model, and establishing a nonlinear mathematical relationship between the jointed rock mass image and the structural surface occurrence (dip, inclination, trace length and spacing);

[0059] Step S14: Use the test set to test the accuracy of the convolutional neural network model, using the regression quantization index R 2 The model accuracy is determined. If the model accuracy does not meet the requirements, the training step continues. If the model accuracy meets the requirements, the model parameters are determined to obtain the trained convolutional neural network.

[0060] Specifically, the method for judging the accuracy of the model can set the judgment threshold to 0.9; if R 2 If R is greater than the threshold, the convolutional neural network model is considered accurate, the model is saved and new jointed rock images are continuously identified; 2 If it is less than the threshold, continue to adjust the parameters and optimize the model algorithm;

[0061] Among them, R 2 is the percentage of the regression sum of squares in the total sum of squares. The larger the value, the better the image recognition effect, the more accurate the model, and the more significant the regression effect. The calculation formula is:

[0062]

[0063] In the formula, y i is the actual value of the i-th test sample, is the i-th sample value output by the model, is the average of the actual values ​​of the test samples.

[0064] In step 2, optionally, the indoor test may include a uniaxial compression test, a triaxial compression test, etc.

[0065] Among them, the mechanical property parameters may include uniaxial compressive strength, elastic modulus, Poisson's ratio, tensile strength, cohesion and internal friction angle.

[0066] In step 2, a multi-scale jointed rock mass model is established based on the occurrence parameters, including the following steps:

[0067] Step 21, constructing a probability distribution model for the occurrence parameters of the structural surface through statistical analysis, and determining the occurrence distribution characteristic value;

[0068] Specifically, the probability distribution model may adopt normal distribution, negative exponential distribution, lognormal distribution, etc. The distribution characteristic values ​​may include: mean and standard deviation.

[0069] Step 22: Establish an engineering-scale fracture discrete network model based on the distribution characteristics of the structural surface, such as Figure 4 As shown in the figure, it is used to characterize the actual joint distribution, and a series of multi-scale discrete fracture network models with the same proportion and increasing size are selected from the model center point of the engineering scale fracture discrete network model, such as Figure 5 shown.

[0070] The synthetic rock mass model includes a discrete fracture network model and a cohesive particle model, such as Figure 2 As shown, a discrete fracture network model is established based on the structural surface probability distribution model and distribution eigenvalues.

[0071] Step 23: The mechanical property parameters of the jointed rock mass obtained by the indoor test are calibrated to obtain the microscopic parameters of the bonding particle model, and the bonding particle model is assigned to the multi-scale fracture network model to obtain a multi-scale synthetic rock mass model, such as Figure 6 shown.

[0072] Mechanical property parameters of jointed rock mass include: uniaxial compressive strength, Poisson's ratio, elastic modulus, tensile strength, cohesion and internal friction angle;

[0073] Specifically, the microscopic parameters of the bonded particle model include particle radius, bond stiffness ratio, bond tensile strength, bond shear strength, and effective modulus.

[0074] After constructing the multi-scale synthetic rock mass model, Figure 7 As shown, mechanical test simulation was carried out to obtain the mechanical property parameters (uniaxial compressive strength, elastic modulus, Poisson's ratio, tensile strength, cohesion and internal friction angle) of the multi-scale synthetic rock model.

[0075] Conduct mechanical test simulations, which may include uniaxial compression, Brazilian splitting, and triaxial compression test simulations.

[0076] Optionally, the mechanical property parameters of the multi-scale synthetic rock model may include: uniaxial compressive strength, elastic modulus, Poisson's ratio, tensile strength, cohesion and internal friction angle.

[0077] In this embodiment, a multi-scale synthetic rock model is established based on the structural surface information obtained by image recognition, and the properties of the jointed rock mass are accurately characterized. The mechanical property parameters of the multi-scale synthetic rock model are obtained through experimental simulation, and the model size-mechanical property parameter change curve is drawn. The size of the characterization unit of the jointed rock mass is determined by using the difference ratio, which improves the determination of the characterization unit and realizes the quantification and standardization of the characterization unit acquisition method.

[0078] Furthermore, in order to intuitively compare the changes in rock mass mechanical properties parameters, such as Figure 8 As shown, a model size-mechanical property parameter curve is drawn. Specifically, a curve is drawn with the mechanical property parameters of the multi-scale synthetic rock model as the vertical axis and the size of the synthetic rock model as the horizontal axis to obtain the curve change trend.

[0079] A method for obtaining the REV size and corresponding mechanical parameters of a jointed rock mass according to a size-mechanical property parameter curve is specifically as follows: setting a difference ratio threshold, which can be set to 5%-15%, preferably, 10%; calculating the difference ratio of each synthetic rock mass model in the multi-scale synthetic rock mass model, and the synthetic rock mass model with a difference ratio less than the threshold (10%) is a characterization unit body, and the corresponding mechanical property parameter is also the REV parameter, thereby determining the characterization unit body size and the corresponding mechanical property parameter;

[0080] Among them, the difference ratio d i Reflects the degree of change in the mechanical property parameters of adjacent size models, d i The smaller the parameter change, the more stable the mechanical properties of the jointed rock mass. i The calculation formula is:

[0081]

[0082] In the formula, p i is the property parameter of the i-th-level size model, p i-1 is the property parameter of the i-1th level size model.

[0083] In this embodiment, the REV size and corresponding mechanical parameters of the jointed rock mass are obtained by the above method. Further, the step of verifying the accuracy of the REV size and corresponding mechanical parameters of the jointed rock mass is also included: Fig. 9 As shown in the figure, the tunnel excavation process is simulated by finite element software, and the stratum model in the finite element software adopts the mechanical property parameters corresponding to the characterization unit body. The accuracy of the obtained characterization unit body is verified by comparing the displacement curves obtained by finite element simulation and tunnel construction monitoring measurement. When the curves are consistent, REV is considered accurate and the simulation ends. When the curves are inconsistent, REV is considered inaccurate, and the parameters of the synthetic rock model need to be adjusted, and the numerical simulation process can be repeated.

[0084] In this embodiment, the tunnel excavation process is simulated based on finite element method, and the obtained characterization unit volume and its corresponding mechanical property parameters are assigned to the stratum model, and the accuracy of the characterization unit is verified by comparing the results of finite element simulation and tunnel construction monitoring measurement. This improves the characterization unit acquisition method, improves the accuracy of REV, improves the characterization unit verification work, and the obtained tunnel section joint rock mass REV information provides a basis for simulating joint rock mass using continuous medium mechanics calculation method.

[0085] Example 2

[0086] This embodiment provides a jointed rock mass REV numerical simulation acquisition system based on image recognition, such as Figure 1 As shown, it includes: an image acquisition device and a processor, and the processor is configured to execute the steps of the method described in Example 1.

[0087] In some embodiments, the image acquisition device is used to collect and acquire the jointed rock mass image at the tunnel site, and specifically, may include a digital camera, a lighting device, and a laser pointer.

[0088] Digital camera: used to take images of jointed rock mass and store image files;

[0089] Lighting device: used to assist in collecting images. Since there is insufficient light in the tunnel, the lighting device can provide a good lighting environment for the digital camera.

[0090] Laser pointer: used to assist in image acquisition. Laser points are used to calibrate the acquisition area for the digital camera to facilitate image acquisition.

[0091] The processor includes an image recognition unit, a numerical simulation unit and a characterization unit acquisition unit.

[0092] In some embodiments, the image recognition unit is used to identify the acquired jointed rock mass image, and is configured to: identify the image features of the jointed rock mass through a trained convolutional neural network to obtain the occurrence parameters of the joints;

[0093] The image recognition unit includes an image preprocessing module, a convolutional neural network module and a structural surface information acquisition module;

[0094] Furthermore, the image preprocessing module is used to standardize the collected images and grayscale the collected RGB images to facilitate subsequent image processing. At the same time, the image angle is rotated to make the angles of all jointed rock mass images the same. By adjusting the image size, the pixels of all images are unified to ensure the consistency of the image set.

[0095] Furthermore, the convolutional neural network module is used to learn and identify the structural surface occurrence information of the jointed rock mass image. The module consists of an input layer, a convolution layer, a pooling layer, a fully connected layer, and an output layer. The input layer is used to input the preprocessed jointed rock mass image; the convolution layer is used to extract the joint image features; the pooling layer is used to select features and filter irrelevant information; the fully connected layer nonlinearly combines the image features obtained by the convolution layer and the pooling layer to obtain integrated high-order information; the output layer assigns classification labels to the high-order information obtained by the fully connected layer, and passes the label information to the structural surface information acquisition module;

[0096] Furthermore, the structural surface information acquisition module is used to obtain structural surface occurrence parameters obtained through image recognition, including dip, inclination, trace length and spacing; classify the label information of the output layer to obtain the structural surface occurrence information on the jointed rock mass image.

[0097] In some embodiments, the numerical simulation unit is used to establish jointed rock models of different volumes and test their mechanical parameters, and is configured to: use a discrete element method to establish a multi-scale jointed rock model based on occurrence parameters, simulate the indoor test process, and test to obtain the mechanical property parameters of the multi-scale jointed rock model.

[0098] The numerical simulation unit includes structural surface statistical analysis module, jointed rock mass modeling module and mechanical test simulation module;

[0099] Furthermore, the structural surface statistical analysis module is used to determine the probability distribution model (normal distribution, negative exponential distribution, lognormal distribution, etc.) and distribution characteristic values ​​(mean and standard deviation) that the structural surface occurrence obeys. The statistical analysis results of the structural surface are obtained by drawing a structural surface information-probability density bar graph and fitting the curve;

[0100] Furthermore, the jointed rock mass modeling module is used to establish a multi-scale synthetic rock mass model of engineering scale that characterizes the jointed rock mass.

[0101] Specifically, an engineering-scale synthetic rock model is used to characterize the jointed rock mass. The synthetic rock model consists of a discrete fracture network model and a cohesive particle model, such as Figure 2 As shown. A discrete fracture network model is established based on the structural surface probability distribution model and distribution eigenvalues. The mechanical property parameters of the jointed rock mass (uniaxial compressive strength, Poisson's ratio, elastic modulus, tensile strength, cohesion and internal friction angle) obtained through indoor test tests are calibrated to obtain the microscopic parameters of the synthetic rock mass model (particle radius, bonding stiffness ratio, bonding tensile strength, bonding shear strength and effective modulus, etc.). A series of models with consistent proportions and increasing volumes are selected from the center of the established engineering-scale synthetic rock mass model, that is, a multi-scale synthetic rock mass model;

[0102] Furthermore, the mechanical test simulation module is configured to perform mechanical test simulation (uniaxial compression, Brazilian splitting and triaxial compression test simulation) to obtain mechanical property parameters of the multi-scale synthetic rock model.

[0103] In some embodiments, the characterization unit body acquisition unit is used to obtain the characterization unit body size of the jointed rock mass and its corresponding mechanical property parameters, and is configured to draw a model size-mechanical property parameter curve according to the obtained mechanical property parameters to obtain the REV size and corresponding mechanical parameters of the jointed rock mass;

[0104] The characterization unit acquisition unit includes a mechanical property curve drawing module, a quantitative index calculation module and a characterization unit verification module;

[0105] Furthermore, the mechanical property curve drawing module is configured to draw a change curve with the mechanical property parameters of the multi-scale synthetic rock model as the vertical axis and the model size as the horizontal axis to obtain the curve change trend.

[0106] Furthermore, the quantitative index calculation module is configured to determine the size of the characterization unit body. Specifically, the difference ratio of the multi-scale synthetic rock model is calculated, and the absolute value of the difference ratio is less than 10% as the quantitative index for determining the characterization unit body, and the size of the characterization unit body and the corresponding mechanical property parameters are determined;

[0107] Further, the characterization unit verification module is configured to verify the accuracy of the obtained characterization unit. The tunnel excavation process is simulated by finite element software, wherein the stratum model adopts the mechanical property parameters corresponding to the characterization unit. The accuracy of the obtained characterization unit is verified by comparing the displacement curves obtained by the finite element simulation and the tunnel construction monitoring measurement.

[0108] Example 3

[0109] This embodiment provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are run by the processor, the steps described in the method of Embodiment 1 are completed.

[0110] Example 4

[0111] This embodiment provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps described in the method of Embodiment 1 are completed.

[0112] The above description is only a preferred embodiment of the present disclosure and is not intended to limit the present disclosure. For those skilled in the art, the present disclosure may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present disclosure shall be included in the protection scope of the present disclosure.

[0113] Although the above describes the specific implementation methods of the present disclosure in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present disclosure. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present disclosure, various modifications or variations that can be made by those skilled in the art without creative work are still within the scope of protection of the present disclosure.

Claims

1. A method for obtaining REV numerical simulation of jointed rock mass based on image recognition. It is characterized in that The steps include: The acquired jointed rock mass image to be identified is used to identify the image features of the jointed rock mass through a trained convolutional neural network to obtain the occurrence parameters of the joints; The discrete element method is used to establish a multi-scale jointed rock model based on the occurrence parameters, and the mechanical property parameters of the multi-scale jointed rock model are obtained by simulating the indoor test process. According to the obtained mechanical property parameters, the model size-mechanical property parameter curve is drawn to obtain the REV size and corresponding mechanical parameters of the jointed rock mass; It also includes the process of training the convolutional neural network, as follows: Acquire jointed rock mass images, and perform preprocessing to obtain jointed rock mass image sets; According to the structural surface occurrence mark labels of the images in the image set, a feature data set is constructed in which the jointed rock mass images correspond to the occurrence labels one by one; The preprocessed jointed rock mass image is used as the input sample, and the structural surface occurrence information is used as the output sample, which are input into the convolutional neural network for training; The test set is used to test the accuracy of the convolutional neural network model, using regression quantitative indicators Determine the model accuracy. If the model accuracy does not meet the requirements, continue to execute the training step; The model accuracy meets the requirements, the model parameters are determined, and the trained convolutional neural network is obtained.

2. The method for obtaining REV numerical simulation of jointed rock mass based on image recognition according to claim 1, Features: The image features of jointed rock mass are identified through the trained convolutional neural network, including: The preprocessed image is convolved, pooled, and connected to filter out irrelevant information in the image and extract joint image features; The obtained joint image features are nonlinearly combined to obtain integrated high-order information; Classification labels are assigned according to the high-order information of the jointed rock mass image to obtain the occurrence information of the structural surface of the jointed rock mass.

3. The method for obtaining REV numerical simulation of jointed rock mass based on image recognition according to claim 1, It is characterized in that The multi-scale jointed rock mass model is established based on the occurrence parameters, including the following steps: Through statistical analysis, a probability distribution model of the structural surface attitude parameters is constructed, and the characteristic value of the attitude distribution is determined; An engineering scale fracture discrete network model is established according to the distribution characteristics of the structural surface, and a series of multi-scale discrete fracture network models with the same proportion and increasing size are selected from the center point of the engineering scale fracture discrete network model. The mechanical property parameters of jointed rock mass obtained through indoor test are used to calibrate the mesoscopic parameters of the bonding particle model, and the bonding particle model is assigned to the multi-scale fracture network model to obtain a multi-scale synthetic rock model.

4. The method for obtaining REV numerical simulation of jointed rock mass based on image recognition according to claim 1, It is characterized in that A model size-mechanical property parameter curve is drawn. Specifically, a curve is drawn with the mechanical property parameter of the multi-scale synthetic rock model as the vertical axis and the size of the synthetic rock model as the horizontal axis to obtain a curve change trend.

5. The method for obtaining REV numerical simulation of jointed rock mass based on image recognition according to claim 1, Features: The method for obtaining the REV size and corresponding mechanical parameters of the jointed rock mass according to the size-mechanical property parameter curve is as follows: Set the difference ratio threshold; The difference ratio of each synthetic rock model in the multi-scale synthetic rock model is calculated. The synthetic rock model with a difference ratio less than the threshold is the characterization unit body, and the corresponding mechanical property parameters are also the parameters of REV.

6. Jointed rock mass REV numerical simulation acquisition system based on image recognition, Features: Image acquisition device and processor; The image acquisition device is used to acquire images of jointed rock mass; The processor is configured to execute the steps of the method according to any one of claims 1 to 5.

7. The jointed rock mass REV numerical simulation acquisition system based on image recognition as claimed in claim 6, Features: The image acquisition device comprises a digital camera, a lighting device and a laser pointer; Alternatively, the processor includes an image recognition unit, a numerical simulation unit, and a characterization unit volume acquisition unit; An image recognition unit is configured to use the acquired image of the jointed rock mass to be identified to identify the image features of the jointed rock mass through a trained convolutional neural network to obtain the occurrence parameters of the joints; The numerical simulation unit is configured to establish a multi-scale jointed rock model based on the occurrence parameters by using a discrete element method, simulate the indoor test process, and test and obtain the mechanical property parameters of the multi-scale jointed rock model; The characterization unit acquisition unit is configured to draw a model size-mechanical property parameter curve according to the obtained mechanical property parameters, and obtain the REV size and corresponding mechanical parameters of the jointed rock mass.

8. An electronic device, It is characterized in that The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps described in any one of the methods of claims 1 to 5 are completed.

9. A computer-readable storage medium, It is characterized in that Used to store computer instructions, which, when executed by a processor, complete the steps described in any one of claims 1 to 5.

Images