Underground chamber stability assessment method based on deep learning and uncertainty fusion

Abstract

The present invention provides an underground chamber stability assessment method based on deep learning and fusion uncertainty, which relates to the technical field of underground chamber stability assessment. The specific steps include: by constructing an instability assessment model, using historically collected geostress parameters and rock mechanics parameters as a training set, and taking the stability coefficient of the underground chamber as a label, the stability of the underground chamber can be analyzed more accurately through the training of the deep learning model; in addition, a functional relationship of the influence of rock parameters on the stability of the underground chamber is constructed, and multiple groups of current rock parameters are input into the function to obtain multiple groups of current second influence coefficients, thereby improving the comprehensiveness and accuracy of the assessment; finally, through data processing and correction, the uncertainty factors are comprehensively considered to improve the stability and reliability of the assessment results. The present invention can overcome the technical defects of the traditional method, such as the difficulty in determining constants, the incomplete parameter processing, and the unstable assessment results.

Description

Technical Field

[0001] The present invention relates to the technical field of stability assessment of underground chambers, and specifically to a stability assessment method for underground chambers based on deep learning and fusion of uncertainty. Background Art

[0002] The collapse of underground chamber roof is one of the main forms of underground engineering accidents. Many tunnel (chamber) collapse accidents have occurred during the construction of underground chamber infrastructure such as tunnels and subways in my country. Chamber collapses mostly occur in tunnels under construction, so it is necessary to carry out intelligent early warning before the tunnel is excavated to reduce the risk of safety accidents and prevent them before they happen.

[0003] In the prior art, cavern collapse risk analysis mainly relies on the HoekBrown criterion and the molar envelope. However, the difficulty in determining the HoekBrown constants (such as A and B) limits their widespread application. To solve this problem, many studies have turned to the use of the simpler and more readily available molar envelope, but its accuracy and applicability are still limited under different geological conditions; the processing of rock mechanics parameters and geostress parameters is relatively simple, and it is difficult to fully consider the impact of rock parameters on the stability of underground caverns. Often, only simplified assumptions or average values ​​can be used, resulting in a lack of comprehensiveness in the evaluation results; the processing and correction of data is relatively simple, and it is difficult to fully consider the differences and uncertainty factors between multiple sets of data, and the stability and reliability of the evaluation results are limited.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method for evaluating the stability of underground chambers based on deep learning and fusion of uncertainty to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method for evaluating the stability of underground chambers based on deep learning and fusion uncertainty, the specific steps include:

[0008] S1. Collect the size parameters of the underground chamber and multiple related parameters affecting the stability of the underground chamber. The size parameters of the underground chamber include the length, width and height of the chamber. The related parameters affecting the stability of the underground chamber include ground stress parameters and rock mechanics parameters. The ground stress parameters include horizontal stress, vertical stress and ground stress gradient. The rock mechanics parameters include compressive strength, tensile strength, shear strength, elastic modulus and rock density.

[0009] S2. Process the length, width and height of the underground chamber to obtain the volume of the underground chamber, use finite element analysis to input the volume and geometric shape of the chamber, analyze the stress field distribution under different loading conditions, and estimate the stability coefficient of the underground chamber based on the analysis results;

[0010] S3. Use a deep learning network to build an instability assessment model for underground chambers, use the historically collected horizontal stress, vertical stress and ground stress gradient as a training set, use the first influence coefficient of the historical underground chamber as a label, input it into the instability assessment model for training, input multiple sets of currently collected horizontal stress, vertical stress and ground stress gradient into the trained instability assessment model, and predict multiple sets of current first influence coefficients;

[0011] S4. construct a function between the historical compressive strength, tensile strength, shear strength, elastic modulus, density of the rock and the second influence coefficient of the underground chamber, input multiple sets of current rock compressive strength, tensile strength, shear strength, elastic modulus, density into the function, and obtain multiple sets of current second influence coefficients;

[0012] S5. Perform data processing on multiple groups of current first influence coefficients and multiple groups of current second influence coefficients, respectively obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient, perform data processing on the stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient, and generate a corrected stability coefficient of the underground chamber;

[0013] S6. Compare the calibrated stability coefficient of the underground chamber with the coefficient threshold, and determine the stability level of the underground chamber based on the comparison result.

[0014] Furthermore, the instability assessment model is a convolutional neural network model, which is composed of a deep neural network based on a multilayer perceptron. The deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer. The first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function.

[0015] Furthermore, the functional relationship between the compressive strength, tensile strength, shear strength, elastic modulus, density of the rock history and the second influence coefficient of the underground chamber history is constructed. The specific process is as follows:

[0016] Collect sample data including rock historical compressive strength, tensile strength, shear strength, elastic modulus, density and historical underground chamber second influence coefficient;

[0017] Assuming that the linear relationship is satisfied, set the linear model:

[0018] EXS=a×KQ+b×LQ+c×JQ+d×TM+e×MD+f

[0019] Among them, EXS is the second influence coefficient of the historical underground chamber, KQ is the historical compressive strength of the rock, LQ is the historical tensile strength of the rock, JQ is the historical shear strength of the rock, TM is the historical elastic modulus of the rock, MD is the historical density of the rock, a is the coefficient of the compressive strength of the rock, b is the coefficient of the tensile strength of the rock, c is the coefficient of the shear strength of the rock, d is the coefficient of the elastic modulus of the rock, e is the coefficient of the density of the rock, and f is the intercept of the linear model.

[0020] Furthermore, based on the current compressive strength, tensile strength, shear strength, elastic modulus, and density of the given rock, the current second influence coefficient is predicted according to the following formula:

[0021] Using the least squares method, the collected data are applied to the linear model, the data are fitted, and the coefficients a, b, c, d and e, as well as the intercept f of the optimal linear model are determined;

[0022] Based on the obtained linear model and coefficients a, b, c, d and e, and intercept f, a mapping function between the functional relationship between the rock history's compressive strength, tensile strength, shear strength, elastic modulus, density and the second influence coefficient of the underground chamber history is constructed, and the above linear model is regarded as a mapping function;

[0023] The mapping function is used for prediction and analysis. According to the current compressive strength, tensile strength, shear strength, elastic modulus, and density of the given rock, the current second influence coefficient is predicted based on the following formula:

[0024] EXS dq =a×KQ dq +b×LQ dq +c×JQ dq +d×TM dq +e×MD dq +f

[0025] Among them, EXS dq is the current second influence coefficient, KQ dq is the current compressive strength of the rock, LQ dq is the current tensile strength of rock, JQ dq is the current shear strength of rock, TM dq is the current elastic modulus of rock, MD dq is the current density of the rock.

[0026] Furthermore, the process of obtaining multiple sets of current second influence coefficients is as follows:

[0027] After obtaining the current second influence coefficient, multiple sets of rock current compressive strength, tensile strength, shear strength, elastic modulus, and density are input into the above mapping function to obtain a set A of multiple sets of current second influence coefficients, that is, is the second influence coefficient of the current i-th group, i is the index of the second influence coefficient, the value range of i is 1, 2, …, m, and m is the number of second influence coefficients.

[0028] Furthermore, multiple groups of current first influence coefficients and multiple groups of current second influence coefficients are processed to obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient respectively, and the stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient are processed to generate a corrected stability coefficient of the underground chamber, according to the following formula:

[0029]

[0030] in, is the mean value of the current first influence coefficient, is the first influence coefficient of the current j-th group, j is the index of the current first influence coefficient, the value range of j is 1, 2, ..., n, n is the number of the current first influence coefficient, is the mean value of the current second influence coefficient, F is the stability coefficient of the underground chamber, XS jz The stability coefficient of the underground chamber after correction, α is the weight factor of the mean value of the current first influence coefficient, β is the weight factor of the mean value of the current second influence coefficient, α>β, α+β=1.

[0031] Furthermore, the stability coefficient of the underground chamber after correction is compared with the coefficient threshold, and the stability level of the underground chamber is determined according to the comparison result. The specific process is as follows:

[0032] When the stability factor of the underground chamber after correction is XS jz >Coefficient threshold YZ xS , then the stability level of the underground chamber is judged to be good;

[0033] When the stability factor of the underground chamber after correction is XS jz ≤ coefficient threshold YZ XS , the stability level of the underground chamber is judged to be poor.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] The present invention uses deep learning technology to construct an instability assessment model for underground chambers, uses historically collected geostress parameters and rock mechanics parameters as training sets, and uses the stability coefficient of underground chambers as labels. Through the training of the deep learning model, the stability of underground chambers can be analyzed more accurately, overcoming the problem that constants are difficult to determine in traditional methods.

[0036] In addition, by constructing a function between rock mechanical parameters and underground chamber stability coefficients, multiple sets of current rock parameters are input into the function to obtain multiple sets of current second influence coefficients. This step can more comprehensively consider the impact of rock parameters on underground chamber stability, thereby improving the comprehensiveness and accuracy of the assessment.

[0037] By performing data processing on multiple groups of current first influence coefficients and multiple groups of current second influence coefficients, the average of the current first influence coefficient and the current second influence coefficient are obtained respectively, and the stability coefficient and the average of the underground chamber are processed to generate a corrected stability coefficient of the underground chamber. This step can better comprehensively consider uncertainty factors and improve the stability and reliability of the evaluation results. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION

[0039] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0040] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0041] Example:

[0042] See also Figure 1 , the present invention provides a technical solution:

[0043] A method for evaluating the stability of underground chambers based on deep learning and fusion uncertainty, the specific steps include:

[0044] S1. Collect the size parameters of the underground chamber and multiple related parameters affecting the stability of the underground chamber. The size parameters of the underground chamber include the length, width and height of the chamber. The related parameters affecting the stability of the underground chamber include ground stress parameters and rock mechanics parameters. The ground stress parameters include horizontal stress, vertical stress and ground stress gradient. The rock mechanics parameters include compressive strength, tensile strength, shear strength, elastic modulus and rock density.

[0045] S2. Process the length, width and height of the underground chamber to obtain the volume of the underground chamber, use finite element analysis to input the volume and geometric shape of the chamber, analyze the stress field distribution under different loading conditions, and estimate the stability coefficient of the underground chamber based on the analysis results;

[0046] S3. Use a deep learning network to build an instability assessment model for underground chambers, use the historically collected horizontal stress, vertical stress and ground stress gradient as a training set, and use the first influence coefficient of the historical underground chamber as a label. The first influence coefficient of the historical underground chamber can be obtained through expert group testing and input into the instability assessment model for training. Input multiple sets of currently collected horizontal stress, vertical stress and ground stress gradient into the trained instability assessment model, respectively, and predict multiple sets of current first influence coefficients;

[0047] S4. construct a function between the historical compressive strength, tensile strength, shear strength, elastic modulus, density of the rock and the second influence coefficient of the underground chamber, the second influence coefficient of the historical underground chamber can be obtained through expert group testing, and multiple sets of current rock compressive strength, tensile strength, shear strength, elastic modulus, and density are input into the function to obtain multiple sets of current second influence coefficients;

[0048] S5. Perform data processing on multiple groups of current first influence coefficients and multiple groups of current second influence coefficients, respectively obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient, perform data processing on the stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient, and generate a corrected stability coefficient of the underground chamber;

[0049] S6. Compare the calibrated stability coefficient of the underground chamber with the coefficient threshold, and determine the stability level of the underground chamber based on the comparison result.

[0050] On the basis of the above embodiments, the present invention studies the rectangular chamber, and therefore uses a laser rangefinder for non-contact measurement. The laser beam is emitted and reflected back to the instrument, and the distance, i.e., the length, width and height of the underground chamber, is determined by calculating the time difference.

[0051] The methods and equipment for collecting horizontal stress, vertical stress and geostress gradient are as follows:

[0052] Install strain gauges on rock or concrete structures and deduce horizontal stress by measuring strain changes;

[0053] The pressure sensor is installed in the borehole to directly measure the vertical stress at that depth;

[0054] Multiple pressure sensors are set up in a borehole to measure the stress at different depths layer by layer, and then the stress gradient is calculated.

[0055] The methods and equipment for collecting the compressive strength, tensile strength, shear strength, elastic modulus and density of rocks are as follows:

[0056] Using a compression testing machine, axial pressure is applied to the rock sample until the sample fails, and its maximum bearing capacity is recorded to calculate the compressive strength;

[0057] Using a tensile testing machine, tensile force is applied to the rock sample until the sample fails, and the maximum tensile stress is recorded to calculate the tensile strength;

[0058] Use a shear tester to apply shear force on the rock sample and record the maximum shear stress at failure;

[0059] Using an ultrasonic tester, by applying a small amplitude dynamic load, the rock response is measured to calculate the elastic modulus;

[0060] Density is calculated directly by measuring the mass and volume of the rock using a graduated cylinder.

[0061] On the basis of the above embodiment, the length, width and height of the underground chamber are processed to obtain the volume of the underground chamber according to the following formula:

[0062] V=L×W×H

[0063] Among them, V is the volume of the underground chamber, L is the length of the underground chamber, W is the width of the underground chamber, and H is the height of the underground chamber.

[0064] On the basis of the above embodiment, the volume and geometric shape of the underground chamber are input by using the finite element method, the stress field distribution under different loading conditions is analyzed, and the stability coefficient of the underground chamber is estimated according to the analysis results. The specific process is as follows:

[0065] Input the volume and geometric shape of the underground chamber. The geometric shape of the underground chamber is a rectangle. Convert the geometric shape of the underground chamber into a finite element model. Divide the chamber into many small units through meshing, and establish the connection relationship between nodes and units.

[0066] In finite element analysis, material models are used to describe the mechanical properties of rocks. Common material models include elastic models. Elastic models assume that rocks are completely elastic. For elastic models, linear elastic models are used for modeling, in which the relationship between stress and strain follows Hooke's law. Based on the stress-strain relationship in Hooke's law, the normal stress and shear stress at the material nodes are calculated.

[0067] Set loading conditions, such as the geostress applied to the chamber and the thermal stress caused by temperature changes. The geostress can be simulated by applying predefined pressure loads on the top and side walls of the model. The thermal stress can be simulated by applying predefined temperature fields or thermal loads. The total normal stress on the node is simulated by applying predefined geostress and thermal stress. The stress is transferred to the node through the static equilibrium equation, and the total normal stress on the node is calculated.

[0068] Apply boundary conditions and select fixed node displacements on the bottom and side walls of the chamber, indicating that the boundary between the chamber and the surrounding soil is fixed;

[0069] Apply boundary conditions and loading conditions to the established finite element model, solve the displacement field of the model, and calculate the strain components of the nodes through the displacement field. The strain components include normal strain and shear strain.

[0070] According to the displacement field of the node, the normal strain and shear strain of the node and the normal stress and shear stress at the node are calculated, and the stability coefficient of the underground chamber is calculated according to the following formula:

[0071] σ' n =E×ε n

[0072] τ'=G×ε τ

[0073]

[0074] Among them, σ' n is the normal stress at the node, E is the elastic modulus of the material, ε n is the normal strain at the node, τ' is the shear stress at the node, G is the shear modulus of the material, ε τ is the shear strain at the node, F is the stability coefficient of the underground chamber, σ n is the total normal stress on the node.

[0075] Based on the above embodiment, the instability assessment model is a convolutional neural network model, which is composed of a deep neural network based on a multilayer perceptron. The deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer. The first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function.

[0076] In this embodiment, the input features of the deep neural network of the multilayer perceptron include: three features: current multiple sets of horizontal stress, vertical stress and ground stress gradient.

[0077] The structure of a deep neural network with a multilayer perceptron is:

[0078] Input layer**: receives input of 3 features;

[0079] The first hidden layer has 64 neurons and uses ReLU as the activation function.

[0080] The second hidden layer has 128 neurons and also uses the ReLU activation function.

[0081] The third hidden layer has 64 neurons and uses the ReLU activation function.

[0082] Output layer: has a single neuron and the current multiple groups of first influence coefficients.

[0083] On the basis of the above embodiment, a functional relationship between the compressive strength, tensile strength, shear strength, elastic modulus, density of the rock history and the second influence coefficient of the underground chamber history is constructed. The specific process is as follows:

[0084] Collect sample data including rock historical compressive strength, tensile strength, shear strength, elastic modulus, density and historical underground chamber second influence coefficient;

[0085] Assuming that the linear relationship is satisfied, set the linear model:

[0086] EXS=a×KQ+b×LQ+c×JQ+d×TM+e×MD+f

[0087] Among them, EXS is the second influence coefficient of the historical underground chamber, KQ is the historical compressive strength of the rock, LQ is the historical tensile strength of the rock, JQ is the historical shear strength of the rock, TM is the historical elastic modulus of the rock, MD is the historical density of the rock, a is the coefficient of the compressive strength of the rock, b is the coefficient of the tensile strength of the rock, c is the coefficient of the shear strength of the rock, d is the coefficient of the elastic modulus of the rock, e is the coefficient of the density of the rock, and f is the intercept of the linear model;

[0088] Using the least squares method, the collected data are applied to the linear model, the data are fitted, and the coefficients a, b, c, d and e, as well as the intercept f of the optimal linear model are determined;

[0089] Based on the obtained linear model and coefficients a, b, c, d and e, and intercept f, a mapping function between the functional relationship between the rock history's compressive strength, tensile strength, shear strength, elastic modulus, density and the second influence coefficient of the underground chamber history is constructed, and the above linear model is regarded as a mapping function;

[0090] The mapping function is used for prediction and analysis. According to the current compressive strength, tensile strength, shear strength, elastic modulus, and density of the given rock, the current second influence coefficient is predicted based on the following formula:

[0091] EXS dq =a×KQ dq +b×LQ dq +c×JQ dq +d×TM dq +e×MD dq +f

[0092] Among them, EXS dq is the current second influence coefficient, KQ dq is the current compressive strength of the rock, LQ dq is the current tensile strength of rock, JQ dq is the current shear strength of rock, TM dq is the current elastic modulus of rock, MD dq is the current density of the rock;

[0093] After obtaining the current second influence coefficient, multiple sets of rock current compressive strength, tensile strength, shear strength, elastic modulus, and density are input into the above mapping function to obtain a set A of multiple sets of current second influence coefficients, that is, is the second influence coefficient of the current i-th group, i is the index of the second influence coefficient, the value range of i is 1, 2, …, m, and m is the number of second influence coefficients.

[0094] On the basis of the above embodiment, multiple groups of current first influence coefficients and multiple groups of current second influence coefficients are processed to obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient respectively, and the stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient are processed to generate the corrected stability coefficient of the underground chamber, according to the following formula:

[0095]

[0096] in, is the mean value of the current first influence coefficient, is the first influence coefficient of the current j-th group, j is the index of the current first influence coefficient, the value range of j is 1, 2, ..., n, n is the number of the current first influence coefficient, is the mean value of the current second influence coefficient, XS jz The stability coefficient of the underground chamber after correction, α is the weight factor of the mean value of the current first influence coefficient, and β is the weight factor of the mean value of the current second influence coefficient;

[0097] The reasons why the geostress parameters have a more significant impact on the stability coefficient of underground chambers than the rock mechanics parameters are mainly as follows:

[0098] Geostress is the external force exerted on the underground rock mass, which directly acts on the rock mass around the underground chamber. The magnitude and direction of geostress directly affect the internal stress distribution and stress concentration of the rock mass. If the geostress is large, the stability of the rock mass will be greatly threatened;

[0099] Larger ground stress will lead to stress concentration in the rock mass, especially in the rock mass around the underground chamber. Stress concentration will lead to local reduction in the strength of the rock mass and increase the risk of rock mass damage. Rock mechanics parameters mainly describe the mechanical properties of rock materials, and have relatively little effect on stress distribution and stress concentration.

[0100] The main destructive mechanism of the influence of geostress on the stability of underground chambers is the fracture and slip of rock formations. Larger geostress will increase the risk of fracture and slip of rock formations, thus affecting the stability of the chambers. Rock mechanics parameters mainly describe the strength and deformation properties of rock materials, and have relatively little effect on fracture and slip of rock formations.

[0101] In summary, the influence of geostress parameters on the stability of underground chambers is more obvious because geostress directly acts on the rock mass around the chamber, causing stress concentration and triggering failure mechanisms. Rock mechanics parameters mainly describe the mechanical properties of rock materials and have relatively little influence on the distribution and concentration of stress. Therefore, in the stability analysis of underground chambers, the influence of geostress parameters is usually considered to be more important, so α>β is set, and α+β=1 is set when there is no influence of other parameters.

[0102] On the basis of the above embodiment, the stability coefficient of the underground chamber after correction is compared with the coefficient threshold, and the stability level of the underground chamber is determined according to the comparison result. The specific process is as follows:

[0103] When the stability factor of the underground chamber after correction is XS jz >Coefficient threshold YZ XS , then the stability level of the underground chamber is judged to be good;

[0104] When the stability factor of the underground chamber after correction is XS jz ≤ coefficient threshold YZ XS , the stability level of the underground chamber is judged to be poor.

[0105] The specific values ​​of α and β in the formula are generally determined by technical personnel in this field according to actual conditions. The essence of this formula is a comprehensive analysis based on weighted summation. Technical personnel in this field collect multiple groups of sample data and set corresponding weight factors for each group of sample data. The set weight factors and the collected sample data are substituted into the formula. Through repeated experiments and parameter adjustments, the accuracy of the model output and the rationality of the results are observed, the weight factors are gradually adjusted, and the performance and effect of the model under different parameter settings are compared to find the optimal combination of coefficients. The calculated weight factors are screened and averaged to obtain the values ​​of α and β.

[0106] In addition, the size of the weight factor is a specific value obtained by quantifying each parameter. In order to facilitate subsequent comparison, the size of the weight factor depends on the amount of sample data and the technical personnel in this field preliminarily set the corresponding weight factor for each set of sample data. It is not unique as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0107] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0108] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0109] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0110] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for evaluating underground chamber stability based on deep learning and fusion uncertainty, characterized in that: The specific steps include: S1. Collect the size parameters of the underground chamber and multiple related parameters affecting the stability of the underground chamber. The size parameters of the underground chamber include the length, width and height of the chamber. The related parameters affecting the stability of the underground chamber include ground stress parameters and rock mechanics parameters. The ground stress parameters include horizontal stress, vertical stress and ground stress gradient. The rock mechanics parameters include compressive strength, tensile strength, shear strength, elastic modulus and rock density. S2. Process the length, width and height of the underground chamber to obtain the volume of the underground chamber, use finite element analysis to input the volume and geometric shape of the chamber, analyze the stress field distribution under different loading conditions, and estimate the stability coefficient of the underground chamber based on the analysis results; S3. Use a deep learning network to build an instability assessment model for underground chambers, use the historically collected horizontal stress, vertical stress and ground stress gradient as a training set, use the first influence coefficient of the historical underground chamber as a label, input it into the instability assessment model for training, input multiple sets of currently collected horizontal stress, vertical stress and ground stress gradient into the trained instability assessment model, and predict multiple sets of current first influence coefficients; S4. construct a function between the historical compressive strength, tensile strength, shear strength, elastic modulus, density of the rock and the second influence coefficient of the underground chamber, input multiple sets of current rock compressive strength, tensile strength, shear strength, elastic modulus, density into the function, and obtain multiple sets of current second influence coefficients; S5. Perform data processing on multiple groups of current first influence coefficients and multiple groups of current second influence coefficients, respectively obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient, perform data processing on the stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient, and generate a corrected stability coefficient of the underground chamber; S6. Compare the corrected stability coefficient of the underground chamber with the coefficient threshold, and determine the stability level of the underground chamber according to the comparison result; Collect sample data including rock historical compressive strength, tensile strength, shear strength, elastic modulus, density and historical underground chamber second influence coefficient; Assuming that the linear relationship is satisfied, set the linear model: EXS=a×KQ+b×LQ+c×JQ+d×TM+e×MD+f Among them, EXS is the second influence coefficient of the historical underground chamber, KQ is the historical compressive strength of the rock, LQ is the historical tensile strength of the rock, JQ is the historical shear strength of the rock, TM is the historical elastic modulus of the rock, MD is the historical density of the rock, a is the coefficient of the compressive strength of the rock, b is the coefficient of the tensile strength of the rock, c is the coefficient of the shear strength of the rock, d is the coefficient of the elastic modulus of the rock, e is the coefficient of the density of the rock, and f is the intercept of the linear model.

2. The underground chamber stability assessment method based on deep learning and fusion uncertainty according to claim 1 is characterized in that: The instability assessment model is a convolutional neural network model, which is composed of a deep neural network based on a multilayer perceptron. The deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer and an output layer. The first hidden layer, the second hidden layer and the third hidden layer each have at least two neurons, and each uses ReLU as an activation function.

3. The underground chamber stability assessment method based on deep learning and fusion uncertainty according to claim 1 is characterized in that: Based on the current compressive strength, tensile strength, shear strength, elastic modulus, and density of the given rock, the current second influence coefficient is predicted according to the following formula: Using the least squares method, the collected data are applied to the linear model, the data are fitted, and the coefficients a, b, c, d and e, as well as the intercept f of the optimal linear model are determined; Based on the obtained linear model and coefficients a, b, c, d and e, and intercept f, a mapping function between the functional relationship between the rock history's compressive strength, tensile strength, shear strength, elastic modulus, density and the second influence coefficient of the underground chamber history is constructed, and the above linear model is regarded as a mapping function; The mapping function is used for prediction and analysis. According to the current compressive strength, tensile strength, shear strength, elastic modulus, and density of the given rock, the current second influence coefficient is predicted based on the following formula: EXS dq =a×KQ dq +b×LQ dq +c×JQ dq +d×TM dq +e×MD dq +f Among them, EXS dq is the current second influence coefficient, KQ dq is the current compressive strength of the rock, LQ dq is the current tensile strength of rock, JQ dq is the current shear strength of the rock, TM dq is the current elastic modulus of rock, MD dq is the current density of the rock.

4. The underground chamber stability assessment method based on deep learning and fusion uncertainty according to claim 3 is characterized in that: The process of obtaining multiple sets of current second influence coefficients is as follows: After obtaining the current second influence coefficient, multiple sets of rock current compressive strength, tensile strength, shear strength, elastic modulus, and density are input into the above mapping function to obtain a set A of multiple sets of current second influence coefficients, that is, is the second influence coefficient of the current i-th group, i is the index of the second influence coefficient, the value range of i is 1, 2, …, m, and m is the number of second influence coefficients.

5. The underground chamber stability assessment method based on deep learning and fusion uncertainty according to claim 4 is characterized in that: Data processing is performed on multiple groups of current first influence coefficients and multiple groups of current second influence coefficients to obtain the mean of the current first influence coefficient and the mean of the current second influence coefficient respectively. The stability coefficient of the underground chamber and the mean of the current first influence coefficient and the mean of the current second influence coefficient are processed to generate the corrected stability coefficient of the underground chamber, according to the following formula: in, is the mean value of the current first influence coefficient, is the first influence coefficient of the current j-th group, j is the index of the current first influence coefficient, the value range of j is 1, 2, ..., n, n is the number of the current first influence coefficient, is the mean value of the current second influence coefficient, F is the stability coefficient of the underground chamber, XS jz The stability coefficient of the underground chamber after correction, α is the weight factor of the mean value of the current first influence coefficient, β is the weight factor of the mean value of the current second influence coefficient, α>β, α+β=1.

6. The underground chamber stability assessment method based on deep learning and fusion uncertainty according to claim 5 is characterized in that: The stability coefficient of the calibrated underground chamber is compared with the coefficient threshold, and the stability level of the underground chamber is determined according to the comparison result. The specific process is as follows: When the stability factor of the underground chamber after correction is XS jz >Coefficient threshold YZ XS , then the stability level of the underground chamber is judged to be good; When the stability factor of the underground chamber after correction is XS jz ≤ coefficient threshold YZ XS , the stability level of the underground chamber is judged to be poor.

Images