Deep surrounding rock large deformation rapid prediction method and system based on PCA algorithm

Abstract

The invention relates to the technical field of deep surrounding rock large deformation prediction, and particularly discloses a deep surrounding rock large deformation rapid prediction method and system based on a PCA algorithm, and the method comprises the steps: selecting a prediction index from a plurality of indexes affecting surrounding rock deformation; obtaining the measured value of each prediction index and the corresponding maximum relative deformation according to the prediction indexes, and taking the measured value and the corresponding maximum relative deformation as a data set; extracting a principal component vector from the measured value of each prediction index through a principal component analysis algorithm, and constructing a deformation prediction model in combination with a corresponding principal component coefficient; performing least square regression modeling on the deformation prediction model by using the data set, determining the principal component coefficient, and obtaining a trained deformation prediction model; and performing deformation prediction on a to-be-predicted item according to the trained deformation prediction model. According to the method, historical data of a large deformation project can be fully utilized, multiple indexes are comprehensively considered, the weight of each index does not depend on artificial subjective judgment, and the accuracy of large deformation prediction is greatly improved.

Description

Technical Field

[0001] This invention relates to the field of deep surrounding rock large deformation prediction technology, and in particular to a rapid prediction method and system for deep surrounding rock large deformation based on PCA algorithm. Background Technology

[0002] During the construction of deep underground spaces, the initial stress field of the surrounding rock is extremely high, and adverse geological conditions are frequently encountered, leading to frequent large deformation disasters of the surrounding rock after excavation. Currently, effective prediction of large deformation of the surrounding rock is still lacking in engineering practice. Excavation is often started hastily without clearly identifying high-risk areas for large deformation, and corresponding pre-control and support measures are insufficient, resulting in casualties, project delays, and significantly increased project costs. Therefore, there is an urgent need to propose a rapid and accurate method for predicting large deformation of the surrounding rock, providing a reference for the differentiated design of support and construction schemes for different sections and parts of deep underground engineering, thereby reducing the probability of large deformation of the surrounding rock.

[0003] Currently, some related technologies exist for the identification, prediction, and classification of large deformations in surrounding rock in underground engineering. CN109506614A discloses a method for determining large deformations in layered surrounding rock. This method first determines whether the tunnel is located in layered surrounding rock based on the exposed surrounding rock at the tunnel face and the deformation characteristics of the surrounding rock after excavation. Secondly, it assesses the content of weak interlayers and relatively hard rock layers in the exposed surrounding rock at the tunnel face. Finally, it determines whether large deformations will occur in the subsequent layered surrounding rock and the magnitude of such deformations based on the content of weak interlayers in the excavated section and the initial support deformation monitoring data. This method only considers the influence of weak interlayers. CN108871262A discloses a method for identifying large deformations in compression-type surrounding rock in deep-buried caverns. This method obtains the maximum principal stress value σ0max through in-situ in-situ stress testing and obtains the rock strength σ through uniaxial compression tests on rock samples. c The method calculates the rock strength stress ratio (SSR) under the self-weight stress field and tectonic stress field according to the stress environment of the underground cavern using the corresponding formula. If SSR > 1, it is determined that large deformation of the surrounding rock due to compression will occur. However, this method can only determine whether it will occur and only considers the strength stress ratio. CN111412885A discloses a method for predicting large deformation of surrounding rock due to compression in deep tunnels. Through in-situ in-situ stress testing and rock mechanics experiments, the strength stress ratio and deformation modulus ratio of the stratum are obtained, and then the surrounding rock deformation prediction formula is applied. The method predicts large deformations of surrounding rock, with coefficients determined by fitting the results of ergodic parameter combination numerical simulations. However, this method only considers the strength-stress ratio and deformation modulus ratio, and the numerical simulations of sample points involve many assumptions, generally resulting in significant errors compared to actual deformation results. CN117648737A discloses a rapid identification method and device for large deformations in soft rock. The identification indicators include first, second, and third deformation data, all of which are discrete categorical data. The probability and level of large deformation are predicted by comprehensively considering the three indicators. In this prediction method, only the first deformation data is automatically classified using a support vector machine (SVM) model; the latter two indicators require manual qualitative determination. CN115470553A discloses a method for predicting and evaluating large deformations in thin-layered, weak, and fractured rock masses. First, it establishes an evaluation index system consisting of eight indicators: tunnel depth, rock stratum dip angle, angle between the rock stratum strike and the tunnel axis, rock strength, rock weathering degree, average thickness of the rock strata at the tunnel face, rock mass structure type, and groundwater development characteristics. Second, expert scoring determines the weight of each indicator. Finally, a fuzzy evaluation method is used to calculate the predicted level of large deformation. However, the indicator weights in this method are determined by expert scoring, which introduces significant errors, and the influence mechanisms of various factors on the large deformation results cannot be accurately reflected.

[0004] The current methods for predicting large deformations are mainly lacking in the following aspects: (1) They focus more on the correlation between a single influencing factor (such as the attitude of rock strata, strength-stress ratio, etc.) and the large deformation of the surrounding rock, and then qualitatively determine whether the large deformation has occurred; (2) For some large deformation prediction methods that comprehensively consider multiple influencing factors, the weight of each indicator still depends on the qualitative determination by expert scoring, and the influence mechanism of each factor on the final large deformation result cannot be accurately reflected; (3) When calculating the final prediction result with multiple indicators, it is still a subjective classification judgment by humans, and there is a lack of clear and stable prediction formulas; (4) It is impossible to conduct error testing, and the level of quantitative prediction is low. This invention provides a method that can significantly improve the level of large deformation prediction based on solving the above-mentioned shortcomings. Summary of the Invention

[0005] To overcome the problems of frequent rock deformation and lack of effective prediction in existing deep rock deformation prediction methods, this invention provides a method and system for rapid prediction of large deformations in deep rock based on the PCA algorithm.

[0006] In a first aspect, the present invention provides a method for rapid prediction of large deformations in deep surrounding rock based on the PCA algorithm, the method comprising: Predictive indicators are selected from multiple indicators affecting the deformation of surrounding rock using the analytic hierarchy process (AHP). The measured values ​​of each prediction index and the corresponding maximum relative deformation in historical deep-earth engineering with large deformation of surrounding rock are obtained based on the prediction index, and used as a dataset. Principal component vectors are extracted from the measured values ​​of each prediction index using the principal component analysis algorithm, and then combined with the corresponding principal component coefficients to construct a deformation prediction model. The deformation prediction model is modeled using least squares regression using the dataset to determine the principal component coefficients, thus obtaining a trained deformation prediction model. Deformation prediction is performed on the project to be predicted based on the trained deformation prediction model.

[0007] According to a specific implementation method, the above prediction method extracts principal component vectors from the measured values ​​of each prediction index using a principal component analysis algorithm, specifically including: Scoring is performed based on the measured values ​​of each prediction indicator according to the corresponding preset data range, and then the data is processed to be dimensionless. The processed prediction indicators are standardized, and the number of principal components is determined based on the standardized prediction indicators. Then, principal component transformation is performed to obtain the principal component vector.

[0008] According to a specific implementation method, the deformation prediction model in the above prediction method is specifically as follows:

[0009] in, This is the predicted value of the deformation. ~ Principal component coefficients, PC1~PC n Principal component vectors.

[0010] According to a specific implementation method, the above prediction method, which involves predicting the deformation of the project to be evaluated based on the deformation prediction model, specifically includes: The relative deformation prediction value of the project to be evaluated is obtained based on the deformation prediction model. The corresponding large deformation level is obtained based on the predicted value of the relative deformation.

[0011] According to a specific implementation, in the above prediction method, obtaining the corresponding large deformation level based on the predicted value of the relative deformation amount specifically includes: If the predicted relative deformation is less than 1.5%, the large deformation level is Level I. Based on the predicted relative deformation of 1.5% to 2.5%, the large deformation level is Level II; Based on the predicted relative deformation of 2.5% to 4%, the large deformation level is Level III; Based on the predicted relative deformation of 4% to 6%, the large deformation level is Level IV; Based on the predicted relative deformation of 6% to 10%, the large deformation level is Class V; If the predicted relative deformation is ≥10%, the large deformation level is VI.

[0012] According to one specific implementation, in the above prediction method, the relative deformation refers to the ratio of the maximum deformation on one side or the maximum settlement of the arch to the horizontal excavation span of the tunnel.

[0013] According to a specific implementation method, in the above prediction method, prediction indicators are selected from multiple indicators affecting the deformation of the surrounding rock using the analytic hierarchy process (AHP), specifically including: Multiple indicators affecting surrounding rock deformation were scored by experts, and a judgment matrix was constructed. A consistency check is performed on the judgment matrix, and the influence weight of each indicator is calculated based on the fact that the consistency check passes. Indicators with an influence weight higher than a preset threshold are selected as prediction indicators.

[0014] According to a specific implementation method, the above prediction method includes multiple indicators affecting the deformation of the surrounding rock, such as burial depth, maximum principal stress of the surrounding rock, strength-stress ratio of the surrounding rock, orientation of the surrounding rock at the tunnel face, groundwater conditions, tunnel span, gravity density, elastic modulus, Poisson's ratio, and internal friction angle.

[0015] Secondly, the present invention provides a rapid prediction system for large deformations in deep surrounding rock based on the PCA algorithm, the system including a memory and a processor; The memory is used to store computer programs; the processor is used to call and execute the computer programs, so that the system performs the rapid prediction method for large deformations in deep surrounding rock based on the PCA algorithm described above. Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention provides a novel method for predicting large deformations, which can make full use of historical data of large deformation projects, comprehensively consider multiple indicators and the weight of each indicator does not depend on human subjective judgment, forming a clear quantitative prediction formula. At the same time, it supports error self-verification, and the error is concentrated within 10%, which greatly improves the accuracy of large deformation prediction. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a rapid prediction method for large deformations in deep surrounding rock based on the PCA algorithm, provided in an embodiment of the present invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to specific embodiments. However, this should not be construed as limiting the scope of the present invention to the following embodiments; all technologies implemented based on the content of the present invention fall within the scope of the present invention.

[0018] Please refer to Figure 1 This document illustrates a flowchart of a rapid prediction method for large deformations in deep surrounding rock based on the PCA algorithm, provided by an embodiment of the present invention. The method includes: Step 1: Select prediction indicators from multiple indicators affecting the deformation of the surrounding rock using the analytic hierarchy process (AHP).

[0019] Specifically, based on engineering experience and related research, potential influencing factors for large deformation of surrounding rock may include burial depth, maximum principal stress of surrounding rock, strength-stress ratio of surrounding rock, occurrence of surrounding rock, groundwater conditions, tunnel span, gravity density, elastic modulus, Poisson's ratio, and internal friction angle. To eliminate factors with extremely minor influence and simplify the operation of the prediction model proposed in this invention, this step uses the analytic hierarchy process (AHP) to initially determine the influence weight of each indicator.

[0020] First, experts score multiple indicators affecting surrounding rock deformation and construct a judgment matrix. Then, the judgment matrix undergoes a consistency check, and based on the successful consistency check, the influence weight of each indicator is calculated. Finally, indicators with influence weights higher than a preset threshold are selected as prediction indicators.

[0021] Step 2: Based on the predicted indicators, obtain the measured values ​​of each predicted indicator and the corresponding maximum relative deformation in historical deep-earth engineering with large deformation of surrounding rock, as a dataset.

[0022] This step uses the prediction indicators selected in step 1 as the basis for further predictions. Measured values ​​of the prediction indicators and corresponding maximum relative deformation data can be collected from relevant engineering data of tunnels where large deformation of the surrounding rock has occurred.

[0023] Step 3: Extract principal component vectors from the measured values ​​of each prediction index using the principal component analysis algorithm, and construct a deformation prediction model by combining the corresponding principal component coefficients.

[0024] Specifically, Principal Component Analysis (PCA) is a statistical method that transforms original variables into uncorrelated new variables through orthogonal transformations. Its core is to find the linear combination that maximizes variance. Using PCA requires data dimensionality reduction and principal component extraction. Then, least squares regression modeling is performed on the principal components and the maximum relative deformation to determine the weight values ​​of each principal component, thus forming a prediction model for large deformations in deep surrounding rock.

[0025] In one possible implementation, this step specifically includes: Scoring is performed based on the measured values ​​of each prediction indicator according to the corresponding preset data range, and then the data is processed to be dimensionless. The processed prediction indicators are standardized, and the number of principal components is determined based on the standardized prediction indicators. Then, principal component transformation is performed to obtain the principal component vector.

[0026] Step 4: Using the principal component vector as the independent variable and the corresponding maximum deformation in the dataset as the dependent variable, perform least squares regression modeling to determine the principal component coefficients and obtain the trained deformation prediction model.

[0027] Specifically, the deformation prediction model is as follows:

[0028] in, This is the predicted value of the deformation. ~ Principal component coefficients, PC1~PC n The principal component vectors are also the general formula for multiple linear regression in mathematics.

[0029] Step 5: Perform deformation prediction on the project to be predicted based on the trained deformation prediction model.

[0030] Specifically, this step includes: The relative deformation prediction value of the project to be evaluated is obtained based on the deformation prediction model. The corresponding large deformation level is obtained based on the predicted value of the relative deformation.

[0031] In one possible implementation, the corresponding large deformation level is obtained based on the predicted value of the relative deformation, specifically including: If the predicted relative deformation is <1.5%, the large deformation level is Level I; if the predicted relative deformation is 1.5% to 2.5%, the large deformation level is Level II; if the predicted relative deformation is 2.5% to 4%, the large deformation level is Level III; if the predicted relative deformation is 4% to 6%, the large deformation level is Level IV; if the predicted relative deformation is 6% to 10%, the large deformation level is Level V; and if the predicted relative deformation is ≥10%, the large deformation level is Level VI.

[0032] The embodiments of the present invention will be described and explained in detail below with reference to specific implementation methods.

[0033] As described in step 1 above, the above indicators are constructed into a judgment matrix scored by experts, as shown in Table 1.

[0034] Table 1. Schematic diagram of the judgment matrix

[0035] After consistency testing, the consistency ratio CR of the above judgment matrix is ​​0.0705, which is less than 0.1. Therefore, the matrix can be considered to have satisfactory consistency and passes the consistency test.

[0036] Using the arithmetic mean method, the influence weights of the above 10 indicators are obtained as follows: (0.1626, 0.1787, 0.3031, 0.1166, 0.0910, 0.0155, 0.0190, 0.0251, 0.0519, 0.0361). Using the geometric mean method, the weights are calculated as follows: (0.1675, 0.1908, 0.2993, 0.1199, 0.0873, 0.0140, 0.0173, 0.0227, 0.0472, 0.0335). Using the eigenvalue method, the weights are obtained as follows: (0.1649, 0.1829, 0.3173, 0.1145, 0.0869, 0.0146, 0.0173, 0.0223, 0.0466, 0.0323). The influence weights obtained from the three methods show that the influence weights of five factors—burial depth, maximum principal stress of the surrounding rock, strength-stress ratio of the surrounding rock, orientation of the surrounding rock at the tunnel face, and groundwater conditions—are significantly higher than the other five. Furthermore, the weights of factors such as tunnel span, gravity density, elastic modulus, Poisson's ratio, and internal friction angle are concentrated below 0.04, indicating that these factors have a negligible effect on large deformations. Therefore, in this embodiment, the subsequent steps use burial depth, maximum principal stress of the surrounding rock, strength-stress ratio of the surrounding rock, orientation of the surrounding rock at the tunnel face, and groundwater conditions as prediction indicators.

[0037] Furthermore, as described in step 2 above, the relevant engineering data of 31 tunnels that experienced large deformation disasters in the surrounding rock, such as the Zhegushan Tunnel on National Highway 317 and the Guanjiao Tunnel on Xige Second Line, were used as the training dataset to collect the above five prediction indicators and the maximum relative deformation data for each engineering case.

[0038] Further, as described in step 3 above, the predicted indicators are first processed to be dimensionless. Specifically, based on geological survey data, design drawing data, and construction measurement data, the specific values ​​of the predicted indicators are determined, and scores are assigned according to rules to eliminate the influence of different dimensions of each predicted indicator. This includes: (1) Indicator 1 ( I 1): Tunnel burial depth H Based on the cross-sectional and longitudinal sections of the design drawings, determine the structural burial depth of each section and part of the project, and assign scores according to the table below.

[0039] Table 2. Schematic diagram of tunnel burial depth scoring

[0040] (2) Indicator 2 ( I 2): Maximum principal stress

[0041] Based on the specific conditions of the construction site, methods such as hydraulic fracturing and sensor measurement were used to measure the initial geostress field in situ and obtain the maximum principal stress value. The following table will be used to assign scores.

[0042] Table 3. Schematic diagram of maximum principal stress assignment

[0043] (3) Indicator 3 ( I 3): Strength-stress ratio of surrounding rock

[0044] Rock cores were drilled at the construction site, with minimal disturbance during sampling and wax sealing for moisture retention. A uniaxial compression test was then conducted to determine the uniaxial compressive strength of the rock. The strength-stress ratio of the surrounding rock was obtained. And assign scores according to the table below.

[0045] Table 4. Schematic diagram of stress ratio assignment for surrounding rock strength

[0046] (4) Indicator 4 ( I 4): Occurrence of the surrounding rock at the working face The attitude of the surrounding rock at the working face needs to be clearly defined, including parameters such as the degree of rock fragmentation, joint spacing, joint strike, and joint dip angle. The degree of rock fragmentation can be classified into five levels: "intact, relatively intact, relatively fragmented, fragmented, and extremely fragmented," according to the "Engineering Rock Mass Classification Standard GB / T50218-2014." The main joint spacing, joint strike, and joint dip angle are determined by measurement using tools such as a compass and theodolite, and assigned scores according to the table below.

[0047] Table 5. Schematic diagram of the attitude assignment of the surrounding rock at the tunnel face.

[0048] (5) Indicator 5 ( I 5): Groundwater conditions Based on the groundwater type and head disclosed in the survey data, and combined with the groundwater seepage volume and water pressure value after on-site drilling and excavation, the groundwater conditions are evaluated with reference to the "Engineering Rock Mass Classification Standard GB / T50218-2014" and scored according to the following table.

[0049] Table 6. Schematic diagram of groundwater condition scoring

[0050] Furthermore, the scores of each prediction index are added to the training database, and then each prediction index is standardized using Z-score. The standardized scores form the prediction index vector for the project to be predicted. I 1 ，I 2 ，I 3 ，I 4 ，I 5), and convert it into a 5×1 matrix. IND Following the framework of principal component analysis algorithms in mathematics, the following formula is used to perform principal component transformation: PC = X·IND , in, PC The principal component matrix of the project to be predicted has a shape of 5×1. X This is the transformation coefficient matrix, with a shape of 5×5.

[0051] [ ] = · [ ] In one possible implementation, the transformation coefficient matrix X The determination process includes: Based on the predictive index matrix IND history The matrix is ​​then normalized column-wise using z-scores to obtain... IND history-norm Then calculate the covariance matrix. COV = ( IND history-norm ) T ·( IND history-norm Find the covariance matrix. COV eigenvalues ​​( , , , ) and eigenvector matrix V .

[0052] Calculate the cumulative contribution rate of eigenvalues ,when When the percentage of principal components is greater than or equal to a certain threshold (generally 85%~90%), or when adjacent eigenvalues ​​differ by an order of magnitude after sorting from largest to smallest, then the number of principal components is determined to be n, and the eigenvector matrix is ​​taken. V The first n rows are used as the transformation coefficient matrix X .

[0053] Further, as described in step 4 above, the principal component vector is used as the independent variable, and the corresponding maximum deformation in the dataset is used as the dependent variable. Least squares regression modeling is then performed to determine the principal component coefficients, resulting in the trained deformation prediction model. Specifically, the calculation formula is as follows:

[0054] in, This is the predicted value of the deformation. ~ Principal component coefficients, PC1~PC n The principal component vector. In this embodiment, n is 5.

[0055] Furthermore, principal component coefficients ~ The specific method determined by least squares regression modeling using the training dataset is as follows: The prediction index matrix of the training dataset IND history Transform into the corresponding principal component matrix PC history The calculation formula is: PC history = X ·( IND history ) T .

[0056] exist PC history Add a new row of vectors consisting of the constant 1 at the top, and transpose it to form... PC history-train The matrix has a shape of 31×6.

[0057] The maximum relative deformation of 31 historical engineering cases in the training dataset is shown in Table 6 below. The deformation data are used to construct a deformation matrix. S train The matrix has a shape of 31×1; according to the general formula for matrix calculation in multiple linear regression in mathematics: PC history-train · = S train The formula for calculating the principal component coefficient matrix can be obtained through simple evolution of matrix operations. = ( PCT history-train · PC history-train ) -1 · PC T history-train · S train ,matrix From top to bottom, they are respectively ~ .

[0058] Table 6. Schematic diagram of historical data on maximum relative deformation

[0059] As described in step 5 above, the deformation prediction model is used to predict the deformation of the project to be predicted, and the predicted value of the relative deformation of the project is obtained. Classify the large deformation level of the surrounding rock and carry out corresponding support design and adjust construction measures.

[0060] Furthermore, relative deformation It refers to the ratio of the maximum deformation on one side or the maximum settlement of the arch Δu to the horizontal excavation span d of the tunnel.

[0061] Another specific implementation method is provided below.

[0062] Taking a railway tunnel as an example, the tunnel span is 10.6m, and the actual burial depth at the construction site reaches 715m; the maximum principal stress of the surrounding rock was measured to be 27.16MPa by water pressure fracturing method; the uniaxial compressive strength of the surrounding rock was measured to be 0.54MPa by drilling rock cores and conducting uniaxial compression tests; the construction excavation showed that the surrounding rock was extremely broken, mostly in a loose state; the seepage water at the site was in the form of rain, with local streams flowing out, and the water head reached 0.38MPa.

[0063] First, the five prediction indicators for the tunnel were assigned scores of (3, 2, 30, 4, 2). After Z-score standardization, the prediction indicator matrix for the tunnel was formed. IND = [-0.808, 0.855, 2.229, 1.922, 0.527].

[0064] The relevant engineering data of 32 tunnels that experienced large deformation disasters in the surrounding rock, such as the Zhegushan Tunnel on National Highway 317 and the Guanjiao Tunnel on Xige Second Line, were then converted into the prediction index scores of each engineering case to form a model training dataset, as shown in the table above.

[0065] Extract data from the table to form a 31×5 predictive index matrix. IND historyThe matrix is ​​then normalized column-wise using z-scores to obtain... IND history-norm ; Then calculate the covariance matrix. COV = ( IND history-norm ) T ·( IND history-norm ), covariance matrix COV as follows:

[0066] Find the covariance matrix COV eigenvalues ​​( , , , The values ​​are (1.5864, 0.6014, 1.1318, 1.0333, 0.8138). eigenvector matrix V is:

[0067] Calculate the cumulative contribution rate of eigenvalues Since the eigenvalues ​​are not more than an order of magnitude apart, and =87%, therefore, the total number of principal components is selected as 5, and the eigenvector matrix is ​​taken. V The first 5 rows are used as the transformation coefficient matrix X The matrix is ​​shown below.

[0068]

[0069] Obtain the principal component matrix of the tunnel. PC for: PC = [0.4321, -0.3281, -0.7434, -3.0565, 0.3694] Further calculate the principal component coefficients ~ The prediction index matrix of the training dataset IND history Transform into the corresponding principal component matrix PC history The calculation formula is: PC history = X ·( IND history ) T ; exist PC history Add a new row of vectors consisting of the constant 1 at the top, and transpose it to form...PC history-train From the matrix, we can see that the shape of the matrix is ​​31×6; The maximum relative deformation of 31 historical engineering cases in the training dataset is shown in the table below. The deformation data are used to construct a deformation matrix. S train The matrix has a shape of 31×1; Then the principal component coefficient matrix can be obtained. = ( PC T history-train · PC history-train ) -1 · PC T history-train · S train The principal component coefficients are obtained. ~ The values ​​are (0.1059, -0.0049, 0.0113, -0.0077, -0.0408, 0.0141).

[0070] Furthermore, the predicted relative deformation of the tunnel is calculated: =0.0313. The large deformation level is "general large deformation".

[0071] In this embodiment, the final measured relative deformation under conventional support conditions is 0.033. At the same time, the prediction model proposed in this invention is self-tested using the training dataset. It can be seen that the error is within 10%, and the accuracy is at a relatively good level.

[0072] Based on the above technical solution, the present invention provides a novel large deformation prediction method that can make full use of historical data of large deformation projects, comprehensively consider multiple indicators and the weight of each indicator does not depend on human subjective judgment, forming a clear quantitative prediction formula, while supporting error self-verification, with the error concentrated within 10%, and the accuracy of large deformation prediction is greatly improved.

[0073] On the other hand, the present invention also provides a rapid prediction system for large deformation of deep surrounding rock based on PCA algorithm, including a memory and a processor, wherein the memory is used to store a computer program; the processor is used to call and execute the computer program so that the system executes the rapid prediction method for large deformation of deep surrounding rock based on PCA algorithm described above.

[0074] In embodiments of the present invention, the processor can be an integrated circuit chip with signal processing capabilities. The processor can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0075] The various methods, steps, and logic diagrams disclosed in the embodiments of this invention can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this invention can be directly implemented by a hardware decoding processor, or implemented by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The processor reads information from the storage medium and, in conjunction with its hardware, completes the steps of the above methods.

[0076] The storage medium can be memory, such as volatile memory or non-volatile memory, or may include both volatile and non-volatile memory.

[0077] Among them, non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0078] Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), sync link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM).

[0079] The storage media described in the embodiments of the present invention are intended to include, but are not limited to, these and any other suitable types of memory.

[0080] It should be understood that the system disclosed in the embodiments of the present invention can be implemented in other ways. For example, the division of modules is merely a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the communication connection between modules can be through some interfaces, indirect coupling or communication connections between servers or units, and can be electrical or other forms.

[0081] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each module can exist physically separately, or two or more modules can be integrated into one processing unit. The integrated unit described above can be implemented in hardware or as a software functional unit.

[0082] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0083] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for rapid prediction of large deformations in deep surrounding rock based on PCA algorithm, characterized in that, The method includes: Predictive indicators are selected from multiple indicators affecting the deformation of surrounding rock using the analytic hierarchy process (AHP). The measured values ​​of each prediction index and the corresponding maximum relative deformation in historical deep-earth engineering with large deformation of surrounding rock are obtained based on the prediction index, and used as a dataset. Principal component analysis is used to extract principal component vectors from the measured values ​​of each prediction index and combine them with the corresponding principal component coefficients to construct a deformation prediction model. Using the principal component vector as the independent variable and the maximum deformation in the dataset as the dependent variable, least squares regression modeling is performed to determine the principal component coefficients and obtain the trained deformation prediction model. Deformation prediction is performed on the project to be predicted based on the trained deformation prediction model.

2. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 1, characterized in that, Principal component analysis (PCA) is used to extract principal component vectors from the measured values ​​of each prediction indicator. Specifically, this includes: Scoring is performed based on the measured values ​​of each prediction indicator according to the corresponding preset data range, and then the data is processed to be dimensionless. The processed prediction indicators are standardized, and the number of principal components is determined based on the standardized prediction indicators. Then, principal component transformation is performed to obtain the principal component vector.

3. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 2, characterized in that, The deformation prediction model is specifically as follows: in, This is the predicted value of the deformation. ~ Principal component coefficients, PC1~PC n Principal component vectors.

4. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 1, characterized in that, Based on the deformation prediction model, deformation prediction is performed on the project to be evaluated, specifically including: The relative deformation prediction value of the project to be evaluated is obtained based on the deformation prediction model. The corresponding large deformation level is obtained based on the predicted value of the relative deformation.

5. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 4, characterized in that, Based on the predicted relative deformation value, the corresponding large deformation level is obtained, specifically including: If the predicted relative deformation is less than 1.5%, the large deformation level is Level I. Based on the predicted relative deformation of 1.5% to 2.5%, the large deformation level is Level II; Based on the predicted relative deformation of 2.5% to 4%, the large deformation level is Level III; Based on the predicted relative deformation of 4% to 6%, the large deformation level is Level IV; Based on the predicted relative deformation of 6% to 10%, the large deformation level is Class V; If the predicted relative deformation is ≥10%, the large deformation level is VI.

6. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 5, characterized in that, The relative deformation refers to the ratio of the maximum deformation on one side or the maximum settlement of the arch to the horizontal excavation span of the tunnel.

7. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 1, characterized in that, The analytic hierarchy process (AHP) is used to select predictive indicators from multiple factors affecting surrounding rock deformation, specifically including: Multiple indicators affecting surrounding rock deformation were scored by experts, and a judgment matrix was constructed. A consistency check is performed on the judgment matrix, and the influence weight of each indicator is calculated based on the fact that the consistency check passes. Indicators with an influence weight higher than a preset threshold are selected as prediction indicators.

8. The method for rapid prediction of large deformation in deep surrounding rock based on PCA algorithm according to claim 7, characterized in that, Several factors that affect the deformation of the surrounding rock include burial depth, maximum principal stress of the surrounding rock, strength-stress ratio of the surrounding rock, orientation of the surrounding rock at the tunnel face, groundwater conditions, tunnel span, gravity density, elastic modulus, Poisson's ratio, and internal friction angle.

9. A rapid prediction system for large deformations in deep surrounding rock based on the PCA algorithm, characterized in that, The system includes a memory and a processor; The memory is used to store computer programs; the processor is used to call and execute the computer programs so that the system executes the method for rapid prediction of large deformations in deep surrounding rock based on the PCA algorithm according to any one of claims 1 to 8.

Images