Extremely soft phyllite tunnel surrounding rock parameter inversion method and system

Abstract

The invention discloses an extremely soft phyllite tunnel surrounding rock parameter inversion method and system, and relates to the technical field of tunnel engineering. Comprising the following steps: acquiring basic mechanical parameters, partial pressure parameters and damage parameters of a rock mass; determining to-be-inverted parameters of the rock mass and setting a value range, and performing sensitivity analysis on the to-be-inverted parameters; carrying out a mixed horizontal orthogonal experiment on the to-be-inverted parameters based on a sensitivity analysis result, and carrying out numerical simulation on the rock mass; constructing a prediction model by utilizing a support vector machine (SVM), taking a numerical simulation result as input, and training the model by adopting a random division cross validation method; and predicting the optimal solution of the to-be-inverted parameters of the rock mass by using the trained prediction model. Based on sensitivity analysis of a single to-be-inverted parameter of the HJC model, parameters which greatly influence response variables are identified, so that the proportion of the key factors in a mixed level orthogonal experiment is increased by focusing on the key factors, and the efficiency and accuracy of model training are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel engineering, and particularly relates to a method and system for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels. Background Art

[0002] The engineering characteristics of extremely soft phyllite tunnels are complex and variable, and their materials exhibit complex mechanical behaviors under different environments, which puts higher requirements on tunnel construction and safety assessment. The HJC (Holmquist-Johnson-Cook) model has been widely used in the simulation analysis of rock tunnels such as extremely soft phyllite due to its excellent ability to describe the dynamic mechanical behaviors of engineering materials under large strain, high hydrostatic pressure, and high strain rate conditions. This model can accurately simulate the material failure conditions of phyllite tunnels under changing geological conditions and construction processes, as well as the dynamic responses of tunnels when encountering loads such as earthquakes or explosions during their service life. The HJC model has high accuracy in simulating the responses of tunnel surrounding rocks and can effectively predict the performance of extremely soft phyllite tunnels under different load conditions. Especially during the tunnel excavation and lining construction processes, the mechanical behaviors of materials and the crack propagation patterns are the key points of its research.

[0003] However, although the HJC model has significant advantages in simulating extremely soft phyllite tunnels, the determination of material constitutive parameters remains a major problem in numerical simulation research, which is directly related to the reliability of analysis results. The HJC model contains 21 undetermined parameters, many of which are extremely sensitive in extremely soft phyllite and are difficult or costly to measure experimentally. Currently, most calibration methods use empirical formula methods, but due to the complexity and diversity of the properties of natural rock masses in actual tunnel engineering, this method is limited in its applicable conditions, resulting in obvious differences between the calibrated parameters and the actual situation.

[0004] When conducting numerical simulations based on the HJC constitutive model, if one pursues simulation results that highly match the engineering reality, it is necessary to continuously adjust the parameters to improve adaptability, which undoubtedly increases the computational cost and reduces the computational efficiency. Summary of the Invention

[0005] The purpose of the present invention is to propose a method and system for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels, which perform parameter inverse analysis based on a machine learning method combining orthogonal experiments and support vector machines, effectively improving the accuracy of parameter inverse analysis and shortening the calculation time.

[0006] According to the first aspect of the embodiments of the present disclosure, a method for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels is provided, including the following steps:

[0007] Obtain the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0008] Determine the parameters to be inverted for the rock mass and set the value ranges, and conduct a sensitivity analysis on the parameters to be inverted;

[0009] Based on the sensitivity analysis results, conduct a mixed-level orthogonal experiment on the parameters to be inverted, and perform numerical simulation on the rock mass accordingly;

[0010] Use the support vector machine (SVM) to construct a prediction model, take the numerical simulation results as input, and train the model using the randomly divided cross-validation method;

[0011] Use the trained prediction model to predict the optimal solution of the parameters to be inverted for the rock mass.

[0012] In one embodiment, obtain the basic mechanical parameters of the rock mass through uniaxial compression experiments, tensile and shear experiments, triaxial confining pressure experiments, and Hugoniot experiments. The basic mechanical parameters of the rock mass include the rock mass density ρ, uniaxial compressive strength f c , shear strength G, tensile strength T, and compaction pressure P l , the minimum plastic strain EFMIN at material fracture, and pressure parameters K 1 , K 2 , K 3 .

[0013] In one embodiment, to further reduce the cost of indoor experiments for inverting the sensitive parameters of the HJC constitutive model, the Hugoniot experiments for calibrating the pressure parameters K 1 , K 2 , K 3 can be omitted. The pressure parameters K 1 , K 2 , K 3 are non-sensitive parameters and can be obtained by fitting according to the following formula:

[0014]

[0015] where is the corrected volumetric strain, and C and S are empirical constants;

[0016] The elastic limit pressure P c and elastic limit strain μ c of the rock mass, and the compaction strain μ l are obtained from the following formula:

[0017]

[0018] where K is the bulk modulus of the rock mass, ρ g is the compaction density of the rock mass; K and ρ g are also obtained from the above experiments;

[0019] For the reference strain rate EPSO, failure type F s , damage constant D 2 , the maximum value S that the normalized equivalent stress can reach max , these 4 insensitive parameters adopt default values.

[0020] In one embodiment, for the four parameters A, B, C, and N among the remaining relatively sensitive 5 parameters, if obtained through curve fitting of multiple triaxial compression experiments and Hopkinson bar (SHPB) experiment results, there are problems such as high experimental costs and large errors. D 1 If determined by continuous debugging, there are also problems of excessive errors. Therefore, the normalized cohesive strength A, the normalized pressure hardening factor B, the strain rate coefficient C, the pressure hardening index N, and the failure damage degree D 1 are set as parameters to be inverted, and by referring to relevant literature and industry norms, the value ranges of each parameter to be inverted are set.

[0021] Perform sensitivity analysis on each parameter to be inverted in the HJC model, that is, when considering the sensitivity of one parameter, other parameters remain unchanged, and use the dimensionless numerical simulation results (the ratio obtained when the inversion parameter takes the reference value) as the objective function; each parameter to be inverted is respectively changed by on the basis of the reference value to obtain the influence of the change rate of the parameter to be inverted on the simulation result; the reference value is the median of the value range corresponding to the parameter to be inverted.

[0022] In one embodiment, determine the proportion of different factors in the orthogonal experiment according to the objective functions of the 5 parameters to be inverted, that is, design a mixed-level orthogonal experiment because the number of levels of each factor (parameter to be inverted) is different; use the coincidence degree of the numerical simulation results with the rock mass compression stress-strain curve, the rock mass tensile curve, and the rock mass shear curve obtained from the experiment as a measure of whether the inversion parameter is the optimal solution;

[0023] Use the mean relative error MRE to describe the coincidence degree of the numerical simulation results with the rock mass compression stress-strain curve, the rock mass tensile curve, and the rock mass shear curve as follows:

[0024]

[0025] In the formula, Y i , y i represent the stress values obtained from the numerical simulation and the stress values obtained from the experiment corresponding to the case of strain ε i ; n is the number of different strain points taken on the curve;

[0026] The average value of the mean relative error (MRE) of three stress-strain curves is used as the objective function for evaluating the overall performance of the inversion parameters, serving as the criterion to measure whether the inversion parameters are the optimal solution; that is, the closer the objective function is to 0, the better the performance of the inversion parameters and the closer they are to the optimal solution.

[0027] In one embodiment, constructing a prediction model using a support vector machine (SVM) includes:

[0028] Data preprocessing: Obtain the objective function of the parameters to be inverted, convert the numerical simulation results into a data set recognizable by the model, divide this data set into a test set and a training set, and perform feature scaling simultaneously.

[0029] Model construction: Select appropriate kernel functions and parameters according to experience and establish the Lagrangian formula:

[0030]

[0031] where \(0\leq\beta\) i \(\leq C, i = 1, 2,....., N\), and \(N\) represents the number of data in the training set;

[0032] Training the model: Input the training data set \(T=(X\) 1 ,\(\xi\) 1 ),(X\) 2 ,\(\xi\) 2 ),...,(X\) N ,\(\xi\) N ) into the above Lagrangian formula, where \(X\) i represents the data set of 5 parameters to be inverted, \(\xi\) i \(\in\{-1, 1\}, i = 1, 2, 3,..., N\);

[0033] Obtain the optimal solution

[0034] Select a positive component of \(\beta\) * and calculate: Calculate:

[0035]

[0036] Obtain the prediction model

[0037] In one embodiment, during the construction and training of the prediction model, the parameters are continuously adjusted. Specifically, grid search is used to find the best hyperparameters in the hyperparameter space, and these hyperparameters include the regularization coefficient, error tolerance, kernel function coefficient, and kernel function.

[0038] According to the second aspect of the embodiments of the present disclosure, a system for inverting surrounding rock parameters of an extremely soft phyllite tunnel is provided, including:

[0039] A parameter acquisition module that acquires the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0040] A sensitivity analysis module that determines the parameters to be inverted of the rock mass, sets the value range, and conducts sensitivity analysis on the parameters to be inverted;

[0041] A numerical simulation module that conducts a mixed-level orthogonal experiment on the parameters to be inverted based on the sensitivity analysis results, and numerically simulates the rock mass accordingly;

[0042] A model establishment module that constructs a prediction model using the support vector machine (SVM), takes the numerical simulation results as input, and trains the model using the randomly divided cross-validation method;

[0043] A prediction module that uses the trained prediction model to predict the optimal solution of the parameters to be inverted of the rock mass.

[0044] According to the third aspect of the embodiments of the present disclosure, an electronic device is provided, including a memory, a processor, and a computer program running on the memory. When the processor executes the program, it implements the method for inverting the surrounding rock parameters of an extremely soft phyllite tunnel as described above.

[0045] According to the fourth aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, it implements the method for inverting the surrounding rock parameters of an extremely soft phyllite tunnel as described above.

[0046] The above technical solutions adopted by the present invention, compared with the prior art, have the following advantages: (1) By analyzing the sensitivity of individual parameters to be inverted in the HJC model, the present invention identifies the parameters that have a greater impact on the response variable. Focusing on these key factors, it increases their proportion in the mixed-level orthogonal experiment, thus significantly improving the efficiency and accuracy of model training.

[0047] (2) The present invention uses the average value of the mean relative error (MER) of the curves of the numerical simulation results and the results of 3 groups of indoor test results as the objective function for evaluating the overall performance of the inversion parameters. This method can make the most of the existing experimental results, thereby improving the accuracy of the prediction model for the sensitive parameters to be inverted.

[0048] (3) The present invention uses grid search to find the optimal hyperparameters for establishing the prediction model of the sensitive parameters to be inverted based on the support vector machine (SVM). At the same time, using the shuffled division cross-validation method, on the basis of improving the comprehensiveness and accuracy of the prediction model, it effectively avoids the overfitting phenomenon that may occur on a fixed data set. It can predict the optimal solution of the sensitive parameters to be inverted, which is beneficial to improving the accuracy of numerical simulation.

[0049] (4) The present invention only needs to be based on four simple experiments, namely, uniaxial compression experiment, tensile experiment, shear experiment and triaxial confining pressure experiment, to complete the inversion of the sensitive parameters of the HJC rock mass constitutive model, without the need to conduct Hugoniot and Hopkinson bar experiments. This greatly reduces the workload and experimental cost of calibrating sensitive parameters. Compared with the existing empirical method and formula method, the present invention has stronger universality and flexibility. Description of the Drawings

[0050] The specification drawings forming a part of the present application are used to provide a further understanding of the present application. The schematic embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation to the present application.

[0051] Figure 1 It is a flowchart of a method for inverting the parameters of the surrounding rock of an extremely soft phyllite tunnel;

[0052] Figure 2 It is a flowchart for constructing a prediction model according to an embodiment of the present invention;

[0053] Figure 3 It is a sensitivity analysis diagram of the HJC parameters to be inverted. Detailed Embodiments

[0054] The present disclosure will be further described below in conjunction with the drawings and embodiments.

[0055] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0056] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. 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 "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0057] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of methods and systems according to various embodiments of the present disclosure. It should be noted that each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, program segment, or part of code may include one or more executable instructions for implementing the logical functions specified in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. It should also be noted that each block in the flowchart and / or block diagram, as well as the combinations of blocks in the flowchart and / or block diagram, may be implemented using a dedicated hardware-based system that executes the specified functions or operations, or may be implemented using a combination of dedicated hardware and computer instructions.

[0058] Embodiment 1:

[0059] Please refer to Figure 1 , this embodiment provides a method for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels, including the following steps:

[0060] Step 1: Obtain the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0061] In a specific embodiment, take a cross-section of the surrounding rock of the Longfeng Tunnel at DK447+297.5 from Ankang to Chongqing section of the Xi'an-Chongqing High-Speed Railway as an example. According to the on-site geological report, the surrounding rock type is extremely soft phyllite. Further, through uniaxial compression tests, tensile and shear tests, triaxial confining pressure tests, and Hugoniot tests, the basic mechanical parameters of the rock mass are obtained, including the rock mass density ρ, uniaxial compressive strength f c , shear strength G, tensile strength T, and compaction pressure P l , the minimum plastic strain EFMIN at material fracture, and the pressure parameter K 1 , K 2 , K 3 . The elastic limit pressure P c , elastic limit strain μ c , compaction strain μ l can be calculated by the following formula:

[0062]

[0063] Among them, K is the bulk modulus of the rock mass, ρ g is the compaction density of the rock mass. K, ρ g can also be obtained from the above four groups of experiments. For the reference strain rate EPSO, failure type Fs , damage constant D 2 , the maximum value S that the normalized equivalent stress can reach max , these 4 insensitive parameters adopt the system default values. The known parameters of the rock mass are shown in Table 1 below

[0064] Table 1 Known parameters of the rock mass

[0065]

[0066]

[0067] Step 2: Determine the parameters to be inverted for the rock mass and set the value ranges, and conduct sensitivity analysis on the parameters to be inverted;

[0068] Specifically, among the remaining relatively sensitive 5 parameters in this example, four parameters, namely the normalized cohesion strength A, the normalized pressure hardening factor B, the strain rate coefficient C, and the pressure hardening index N, need to be obtained through curve fitting of the results of multiple triaxial compression experiments and Hopkinson bar (SHPB) experiments, which have problems such as high experimental costs and large errors. The failure damage degree D 1 also has the problem of excessive error. Therefore, these 5 parameters are set as the parameters to be inverted. Referring to relevant literature, the value ranges of the HJC parameters to be inverted for sandstone in the example are shown in Table 2

[0069] Table 2 Value range table of parameters to be inverted

[0070]

[0071] For the uniaxial compression numerical model, taking sandstone as the research object and taking the peak stress σ of the sandstone after compression max as the sensitivity evaluation index. The sensitivity evaluation index σ max is mainly affected by 5 parameters {A, B, C, N, D 1}}, that is, σ max = f(A, B, C, N, D 1 ). Take the median of the value range as the reference value of the parameter {0.3, 2.0, 0.01, 0.825, 0.02}, and select four points of ±10% and ±20% in the change range on the reference parameters for numerical simulation analysis

[0072] The dimensionless numerical simulation results (the ratio obtained when the inversion parameters take the reference value) are used as the objective function as shown in the following formula

[0073]

[0074] In the formula, δ is the absolute value of the stress peak change rate, which is the sensitivity analysis standard, and σ max is the stress peak after a certain sensitive parameter changes The peak stress under the reference parameters. The influence of the change rate of each parameter of the HJC model on the dimensionless numerical simulation results of the sandstone in this example was calculated, as Figure 3 shown.

[0075] Step 3: Based on the sensitivity analysis results, conduct a mixed-level orthogonal experiment on the parameters to be inverted, and numerically simulate the rock mass accordingly;

[0076] Specifically, according to the objective function of the 5 parameters to be inverted, determine the proportions of different factors A: B: C: N: D in the orthogonal experiment 1 to be 10:46:10:31:8. Therefore, the number of rows m of the mixed-level orthogonal experiment can be determined by the following formula:

[0077]

[0078] In the formula, k n , t n represent that k n factors have t n levels. Substituting the number of levels of each factor in this example, the number of rows of the mixed-level orthogonal experiment is 99. The mixed-level orthogonal table is shown in Table 3 below:

[0079] Table 3 Mixed-level orthogonal table of parameters to be inverted

[0080]

[0081] The ellipsis in the table represents the combination of cases between serial numbers 4 - 96. Since the combination results are too many, they are represented by ellipsis in this example. Import the various combinations of parameters to be inverted obtained in the above table and the created groups into LS-DYNA for numerical simulation to obtain the compressive stress-strain curve, tensile curve, and shear curve of the rock mass under various combination cases. The average relative error MRE is used to measure the coincidence degree of the numerical simulation results and the indoor test results curve as follows:

[0082]

[0083] In the formula, Y i , y i represent the stress values obtained from numerical simulation and the stress values obtained from experiments corresponding to the strain ε i . n is the number of different strain points taken on the curve. The average value of the average relative error MRE of the three stress-strain curves is used as the objective function for evaluating the overall performance of the inverted parameter values. The objective function (this value is the standard for measuring whether the parameters to be inverted of the rock mass are the optimal solution) for various combination cases obtained through numerical simulation in this example is shown in Table 4.

[0084] Table 4 Objective functions with inversion parameters for each group

[0085]

[0086] Step 4: Use the support vector machine (SVM) to construct a prediction model. Take the numerical simulation results as input and train the model using the randomly divided cross-validation method.

[0087] Specifically, first perform data preprocessing. Check the numerical simulation results, identify whether there are missing values and outliers in the dataset, and convert the processed data into a format recognizable by the model. Then, divide the data into a training set and a test set. In this example, use the data with serial numbers 1 - 80 as the training set and the data with serial numbers 81 - 99 as the test set. After grouping, standardize the data and convert it into a standard distribution form with a mean of 0 and a variance of 1. During the construction and training of the model, it is necessary to continuously adjust the parameters. Use the grid search method to find the optimal parameters in the hyperparameter space. These hyperparameters include: kernel function, kernel function coefficient, regularization coefficient, and error tolerance. During the adjustment process, find the hyperparameters suitable for the SVM regression algorithm by traversing all candidate parameters and comparing the fitting degree of the model. The optimal hyperparameters in this example are to use the RBF kernel function and set the error tolerance to 0.015. In addition, the best kernel function coefficient (γ) of the SVM regression model is determined to be 0.32 and the regularization coefficient (C) is 5144 through the cross-validation method. Under this hyperparameter setting, the R-squared value of the model on the test set is stable above 94.2%, and the mean squared error (MSE) is 0.11, indicating that the model has a high fitting degree and no overfitting phenomenon, so it can be considered that the model performance is good.

[0088] Step 5: Use the trained prediction model to predict the optimal solution of the parameters to be inverted for the rock mass.

[0089] It should be noted that the R-squared value (R 2 ) represents the proportion of the variance explained by the model, ranging from 0 to 1. The closer it is to 1, the stronger the explanatory ability of the model; while the mean squared error (MSE) represents the average squared difference between the predicted value and the actual value. The smaller the MSE, the more accurate the model prediction.

[0090] In this example, the prediction results are shown in Table 5. When the parameters to be inverted take the optimal solution, the corresponding objective function is only 1.68%.

[0091] Table 5 Optimal solutions of the parameters to be inverted in this example

[0092]

[0093] Example 2

[0094] This embodiment provides a system for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels, including:

[0095] A parameter acquisition module that acquires the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0096] A sensitivity analysis module that determines the parameters to be inversely analyzed for the rock mass, sets the value range, and conducts sensitivity analysis on the parameters to be inversely analyzed;

[0097] A numerical simulation module that conducts a mixed-level orthogonal experiment on the parameters to be inversely analyzed based on the sensitivity analysis results, and conducts numerical simulation on the rock mass accordingly;

[0098] A model establishment module that constructs a prediction model using the support vector machine (SVM), takes the numerical simulation results as input, and trains the model using the randomly divided cross-validation method;

[0099] A prediction module that predicts the optimal solution of the parameters to be inversely analyzed for the rock mass using the trained prediction model.

[0100] Embodiment Three:

[0101] An electronic device includes a memory, a processor, and a computer program running on the memory. When the processor executes the program, it implements the above-mentioned method for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels, including:

[0102] Acquire the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0103] Determine the parameters to be inversely analyzed for the rock mass, set the value range, and conduct sensitivity analysis on the parameters to be inversely analyzed;

[0104] Conduct a mixed-level orthogonal experiment on the parameters to be inversely analyzed based on the sensitivity analysis results, and conduct numerical simulation on the rock mass accordingly;

[0105] Construct a prediction model using the support vector machine (SVM), take the numerical simulation results as input, and train the model using the randomly divided cross-validation method;

[0106] Predict the optimal solution of the parameters to be inversely analyzed for the rock mass using the trained prediction model.

[0107] Embodiment Four:

[0108] A computer-readable storage medium stores a computer program. When the program is executed by a processor, it implements the above-mentioned method for inverse analysis of surrounding rock parameters of extremely soft phyllite tunnels, including:

[0109] Acquire the basic mechanical parameters, partial pressure parameters, and damage parameters of the rock mass;

[0110] Determine the parameters to be inverted for the rock mass and set the value ranges, and conduct a sensitivity analysis on the parameters to be inverted;

[0111] Based on the results of the sensitivity analysis, conduct a mixed-level orthogonal experiment on the parameters to be inverted, and perform numerical simulation on the rock mass accordingly;

[0112] Use the support vector machine (SVM) to construct a prediction model, take the numerical simulation results as the input, and train the model using the randomly divided cross-validation method;

[0113] Use the trained prediction model to predict the optimal solution of the parameters to be inverted for the rock mass.

[0114] Those skilled in the art should understand that the above-mentioned modules or steps of the present disclosure can be implemented by a general-purpose computer device. Optionally, they can be implemented by program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple of them can be fabricated into a single integrated circuit module. The present disclosure is not limited to any specific combination of hardware and software.

[0115] The above are only the preferred embodiments of the present application and are not used to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0116] Although the specific implementation manners of the present disclosure have been described above in conjunction with the accompanying drawings, it is not a limitation to the protection scope of the present disclosure. Those skilled in the art should understand that, based on the technical solutions of the present disclosure, various modifications or deformations that can be made without creative efforts by those skilled in the art are still within the protection scope of the present disclosure.

Claims

1. A method for inversion of surrounding rock parameters of extremely soft phyllite tunnels, characterized in that: The following steps are involved: Obtain basic mechanical parameters, partial pressure parameters and damage parameters of rock mass; Determine the parameters of the rock mass to be inverted and set the value range, and perform sensitivity analysis on the parameters to be inverted; Based on the sensitivity analysis results, mixed level orthogonal experiments are carried out on the inversion parameters, and the rock mass is numerically simulated based on them. The prediction model was constructed using support vector machine (SVM), the numerical simulation results were used as input, and the model was trained using the random partitioning cross-validation method. The trained prediction model is used to predict the optimal solution of the parameters to be inverted in the rock mass.

2. According to claim 1, a method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel is characterized in that: The basic mechanical parameters of the rock mass are obtained through uniaxial compression test, tensile test, shear test, triaxial confining pressure test and Hugoniot test. The basic mechanical parameters of the rock mass include rock mass density ρ, uniaxial compressive strength f c , shear strength G, tensile strength T and compaction pressure P l , the minimum plastic strain EFMIN when the material breaks, and pressure parameters K1, K2, and K3.

3. According to claim 1, a method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel is characterized in that: The pressure parameters K1, K2, and K3 are non-sensitive parameters and are obtained by fitting according to the following formula: In the formula, is the corrected volumetric strain, C and S are empirical constants; Rock elastic limit pressure P c , elastic limit strain μ c , compaction strain μ l , obtained by the following formula: Where K is the bulk modulus of rock mass, ρ g is the compacted density of the rock mass; K, ρ g Obtained by experiment; For the reference strain rate EPSO, failure type F s , damage constant D2, the maximum value S that the normalized equivalent stress can reach max , these four insensitive parameters use default values.

4. According to claim 1, a method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel is characterized in that: The parameters to be inverted include normalized cohesive strength A, normalized pressure hardening factor B, strain rate coefficient C, pressure hardening index N, and failure damage degree D1; A sensitivity analysis is performed on each parameter to be inverted, that is, when considering the sensitivity of a parameter, other parameters remain unchanged, and the dimensionless numerical simulation results are used as the objective function; each parameter to be inverted is analyzed based on the benchmark value. The influence of the change rate of the parameter to be inverted on the simulation results is obtained by the change of the parameter to be inverted; the benchmark value is the median of the value range of the corresponding parameter to be inverted.

5. According to claim 4, a method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel is characterized in that: The proportion of different factors in the orthogonal experiment is determined according to the objective function of the parameters to be inverted, and the degree of coincidence between the numerical simulation results and the rock mass compression stress-strain curve, rock mass tensile curve, and rock mass shear curve obtained from the experiment is used as a criterion for measuring whether the inversion parameters are the optimal solution; The mean relative error (MRE) is used to describe the degree of coincidence between the numerical simulation results and the rock mass compression stress-strain curve, rock mass tensile curve, and rock mass shear curve as follows: Where Y i ,y i Indicates the strain ε i In the case of , the stress values ​​obtained by numerical simulation and the stress values ​​obtained by experiment correspond to ; n is the number of different strain points on the curve; The mean relative error (MRE) of the three stress-strain curves is used as the objective function for the overall performance evaluation of the inversion parameters to measure whether the inversion parameters are the optimal solution. That is, the closer the objective function is to 0, the better the performance of the inversion parameters is and the closer it is to the optimal solution.

6. The method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel according to claim 1, characterized in that: Building a prediction model using support vector machine (SVM) includes: Data preprocessing: Obtain the objective function of the parameters to be inverted, convert the numerical simulation results into a data set that can be used for model recognition, and divide the data set into a test set and a training set, and perform feature scaling at the same time; Model construction: Select kernel functions and parameters based on experience, and establish the Lagrangian formula: Where 0≤β i ≤C,i=1,2,.....,N, where N represents the number of data in the training set; Training model: The training data set T = (X1, ξ1), (X2, ξ2), ..., (X N ,ξ N ) Input the above Lagrangian formula, where X i represents the data set of 5 parameters to be inverted, ξ i ∈{-1,1},i=1,2,3,...,N; Get the optimal solution Select Beta * A positive component of calculate: Get the prediction model 7. The method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel according to claim 6, characterized in that: Parameters are continuously adjusted during the prediction model construction and training process by using grid search to find the best hyperparameters in the hyperparameter space. These hyperparameters include regularization coefficient, error tolerance, kernel function coefficient, and kernel function.

8. An extremely soft phyllite tunnel surrounding rock parameter inversion system, characterized in that: include: Parameter acquisition module, which obtains basic mechanical parameters, partial pressure parameters and damage parameters of rock mass; Sensitivity analysis module, which determines the parameters to be inverted in the rock mass and sets the value range, and performs sensitivity analysis on the parameters to be inverted; Numerical simulation module, which conducts mixed-level orthogonal experiments on inversion parameters based on sensitivity analysis results, and uses this to perform numerical simulation of rock mass; The model building module uses support vector machine (SVM) to build a prediction model, takes the numerical simulation results as input, and uses the random partitioning cross-validation method to train the model; The prediction module uses the trained prediction model to predict the optimal solution of the parameters to be inverted of the rock mass.

9. An electronic device comprising a memory, a processor and a computer program stored and running on the memory, characterized in that: When the processor executes the program, the method for inversion of surrounding rock parameters of an extremely soft phyllite tunnel is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for inversion of parameters of surrounding rock of an extremely soft phyllite tunnel is implemented.

Images