Weak surrounding rock tunnel face advance pre-reinforcement construction method suitable for large-section excavation

Abstract

The invention discloses a weak surrounding rock tunnel face advanced pre-reinforcement construction method suitable for large-section excavation, and belongs to the field of tunnel engineering. The method comprises the steps that a sample database containing historical engineering case basic indexes and result indexes is established; distributing weights for the basic index characteristics to form weight vectors; based on the weighted sample matrix and a Gaussian kernel function, calculating a similarity vector of the new engineering sample and the historical sample; according to the similarity vector and a construction excavation method matrix in the result index, calculating an excavation method probability vector of the new project and determining a recommended excavation method; and constructing a condition matrix in combination with the excavation method matrix and the pre-reinforcement measure matrix, calculating a pre-reinforcement measure activation probability vector under the recommended excavation method, and determining a recommended pre-reinforcement measure according to a preset threshold value. According to the method, data driving and similarity matching are utilized, and standardized and intelligent decision making of the pre-reinforcement scheme and the excavation method is achieved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of tunnel engineering, and particularly relates to a soft surrounding rock tunnel face advance pre-reinforcement construction method suitable for large-section excavation. BACKGROUND

[0002] In the construction of soft surrounding rock tunnels, the New Austrian Tunnelling Method (NATM) is still the most commonly used design and construction theoretical system in China. The basic idea is "subdivision excavation, short footage, and quick closure", which reduces the area of one-time excavation through CD, CRD, double-side-wall pilot tunnel and other subdivision methods, and uses sprayed concrete, anchor rods, steel arches and surrounding rock to form a load ring, so as to control the deformation of surrounding rock. At the same time, the specification system also provides several options for advance pre-reinforcement, such as advance anchor rods, advance pipe roof, advance small catheter grouting, face closure, core soil reservation, etc., as auxiliary means for face stability and stratum settlement control. Because the technical threshold of the New Austrian Tunnelling Method is relatively low, it is highly operable, and a large amount of engineering experience has been accumulated in China, so it will continue to dominate the construction of soft surrounding rock tunnels for a long time in the future.

[0003] However, as soft surrounding rock tunnels develop in the direction of "large section, high risk, and high efficiency", the traditional technology has obvious defects: first, the current "Highway Tunnel Design Specification" and "Highway Tunnel Construction Technology Specification" lack unified definition of the classification and parameters of advance pre-reinforcement technology, resulting in a huge difference in support types and parameters given by designers under similar geological and span conditions; second, the experience-dependent decision-making method is highly subjective and difficult to quantitatively evaluate the interaction between surrounding rock and support, which is prone to over-reinforcement or under-reinforcement; third, the subdivision excavation method has frequent process conversion, narrow operation surface, and low mechanization, causing a vicious cycle of "worker concentration, slow progress, and risk concentration"; fourth, the design and construction have low standardization, and the support materials and special construction tools are difficult to realize industrialization, which not only increases the cost but also prolongs the construction period, and the quality control is difficult. The above problems have become a bottleneck restricting the safe, rapid and economic construction of soft surrounding rock tunnels. SUMMARY

[0004] To solve the above technical problems, the application provides a soft surrounding rock tunnel face advance pre-reinforcement construction method suitable for large-section excavation to solve the problems existing in the prior art.

[0005] To achieve the above purpose, in a first aspect, the application provides a soft surrounding rock tunnel face advance pre-reinforcement construction method suitable for large-section excavation, comprising:

[0006] Redefine the large section excavation method and the soft surrounding rock tunnel face pre-reinforcement technology classification, establish a sample database containing multiple soft surrounding rock tunnel historical engineering cases according to the classification results, wherein each sample includes a basic index feature vector and an achievement index feature vector;

[0007] According to the influence degree of each feature in the basic index feature vector on the achievement index, the weight of each feature is assigned to form a weight vector, and the weighted basic index matrix is obtained according to the weight vector;

[0008] Based on the weighted basic index matrix and the Gaussian kernel function, the similarity vector of the basic index feature vector of the new engineering sample and each sample in the sample database is calculated;

[0009] According to the similarity vector and the achievement index construction excavation method matrix in the sample database, the construction excavation method probability vector of the new engineering sample is calculated;

[0010] According to the construction excavation method probability vector, the recommended construction excavation method is determined;

[0011] According to the similarity vector, the achievement index construction excavation method matrix and the achievement index pre-reinforcement measure matrix in the sample database, the construction excavation method condition matrix is constructed;

[0012] According to the recommended construction excavation method and the construction excavation method condition matrix, the advance pre-reinforcement measure activation probability vector is calculated;

[0013] According to the advance pre-reinforcement measure activation probability vector and the preset threshold, the recommended advance pre-reinforcement measure is determined.

[0014] Preferably, the step of establishing a sample database containing multiple soft surrounding rock tunnel historical engineering cases includes:

[0015] Obtain the technical parameters of multiple soft surrounding rock tunnel historical engineering cases;

[0016] Quantitative coding is performed on the technical parameters to generate a basic index feature vector, an achievement index construction excavation method feature vector and an achievement index pre-reinforcement measure feature vector.

[0017] Preferably, in the step of quantitatively coding the technical parameters, the construction excavation method is coded by using single-hot coding mode, and the advance pre-reinforcement measure is coded by using multi-hot coding mode.

[0018] Preferably, the step of calculating the similarity vector of the basic index feature vector of the new engineering sample and each sample in the sample database includes:

[0019] The weight vector is used to weight the sample basic index matrix to obtain a weighted sample basic index matrix;

[0020] weighting the base index feature vector of the new engineering sample by using the weight vector to obtain a weighted new engineering sample feature vector;

[0021] Based on the Gaussian kernel function, the similarity between each sample in the weighted sample base index matrix and the weighted new engineering sample feature vector is calculated to form a similarity vector.

[0022] Preferably, the step of calculating the construction excavation method probability vector of the new engineering sample comprises:

[0023] Multiply the transpose of the similarity vector by the achievement index construction excavation method matrix to obtain a first intermediate result;

[0024] Multiply the transpose of the similarity vector by the all-one vector to obtain a second intermediate result;

[0025] Divide the first intermediate result by the second intermediate result to obtain the construction excavation method probability vector.

[0026] Preferably, the step of determining the recommended construction excavation method is:

[0027] Select the construction excavation method with the maximum probability value in the construction excavation method probability vector as the recommended construction excavation method.

[0028] Preferably, the step of constructing the construction excavation method condition matrix comprises:

[0029] Multiply the transpose of the achievement index construction excavation method matrix by the diagonalized similarity vector to obtain a first matrix;

[0030] Multiply the transpose of the achievement index construction excavation method matrix by the similarity vector, and diagonalize the result to obtain a second matrix;

[0031] Multiply the second matrix by the first matrix to obtain the construction excavation method condition matrix.

[0032] Preferably, the step of calculating the advanced pre-reinforcement measure activation probability vector is:

[0033] Multiply the single-hot encoding vector representing the recommended construction excavation method by the construction excavation method condition matrix to obtain the advanced pre-reinforcement measure activation probability vector.

[0034] Preferably, the step of determining the recommended advanced pre-reinforcement measure is:

[0035] Compare each element in the advanced pre-reinforcement measure activation probability vector with the preset threshold value, and if the element value is greater than or equal to the preset threshold value, use the advanced pre-reinforcement measure corresponding to the element.

[0036] Preferably, the parameters of the weight vector and the Gaussian kernel function are optimized using leave-one-out cross-validation.

[0037] Compared with the prior art, the present invention has the following advantages and technical effects:

[0038] This invention provides a method for pre-reinforcement of tunnel faces in soft rock tunnels with large-section excavation, applicable to such projects. The method includes: First, redefining the classification of large-section excavation methods and pre-reinforcement technologies for tunnel faces in soft rock tunnels; establishing a sample database containing multiple historical engineering cases of soft rock tunnels based on the classification results, where each sample includes a basic indicator feature vector and an outcome indicator feature vector; Second, assigning weights to each feature based on its influence on the outcome indicators to form a weight vector; obtaining a weighted basic indicator matrix based on the weight vector; and Third, calculating the basic indicator feature vector of a new engineering sample and its corresponding Gaussian kernel function. The similarity vectors of each sample in the sample database are used as follows: Next, based on the similarity vectors and the construction excavation method matrix of the achievement indicators in the sample database, the construction excavation method probability vector of the new project sample is calculated; further, based on the construction excavation method probability vectors, a recommended construction excavation method is determined; based on the similarity vectors, the construction excavation method matrix of the achievement indicators, and the pre-reinforcement measure matrix of the achievement indicators in the sample database, a construction excavation method condition matrix is ​​constructed; finally, based on the recommended construction excavation method and the construction excavation method condition matrix, an activation probability vector of advanced pre-reinforcement measures is calculated; based on the activation probability vector of advanced pre-reinforcement measures and a preset threshold, recommended advanced pre-reinforcement measures are determined.

[0039] This invention transforms the subjective decision-making process, which originally relied on the personal experience of engineers, into a structured and process-oriented objective analysis process by establishing a standardized sample database and unified coding rules. This effectively overcomes the drawback that different technicians may make drastically different designs under similar geological conditions, achieving standardization and unification of the decision-making process.

[0040] This invention utilizes the Gaussian kernel function to calculate engineering similarity, enabling it to accurately identify successful samples from historical cases that most closely match the current geological conditions and design parameters of new engineering projects. Based on the data from these similar samples, construction schemes are recommended, resulting in conclusions (excavation methods and pre-reinforcement measures) that are more closely aligned with actual engineering conditions. This significantly improves the adaptability and scientific validity of the schemes to specific engineering conditions, enhancing the scientific rigor and suitability of scheme selection.

[0041] This invention streamlines the tedious work of extensive literature review and case analysis into a sample database, and automatically matches and recommends solutions through algorithms. This significantly reduces reliance on senior experts, lowers the decision-making threshold, and enables the rapid and efficient generation of preliminary design schemes, thereby improving the efficiency and intelligence of the entire tunnel engineering design process and enhancing the overall efficiency and intelligence of engineering design.

[0042] The sample database constructed in this invention is a continuously expandable and optimized knowledge base. As more successful cases are added, the system's recommendations become increasingly accurate. This promotes the accumulation, sharing, and reuse of engineering experience, laying a solid foundation for the continuous development and standardization of soft rock tunnel engineering technology, and contributing to the industry's technological accumulation and progress. Attached Figure Description

[0043] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0044] Figure 1 This is a flowchart illustrating the pre-reinforcement construction method for tunnel faces in weak surrounding rock, as described in an embodiment of the present invention. Detailed Implementation

[0045] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0046] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0047] Example 1

[0048] like Figure 1 As shown, this embodiment provides a method for pre-reinforcement of the tunnel face in soft surrounding rock, suitable for large-section excavation, including:

[0049] Step S1: Redefine the classification of large-section excavation methods and pre-reinforcement technology for tunnel face in weak surrounding rock. Based on the classification results, establish a sample database containing multiple historical engineering cases of tunnels in weak surrounding rock. Each sample includes a basic indicator feature vector and an outcome indicator feature vector.

[0050] Furthermore, the steps to establish a sample database containing multiple historical engineering cases of tunnels in weak surrounding rock include:

[0051] Step S101: Obtain the technical parameters of multiple historical engineering cases of tunnels in weak surrounding rock;

[0052] In this embodiment, the technical parameters include:

[0053] (1) Basic indicators: weak surrounding rock geological type, surrounding rock weathering degree, tunnel span, tunnel height, surrounding rock grade, surrounding rock water richness, adverse geological type, tunnel burial depth, surrounding rock and lining deformation value, tunnel engineering scale, etc.

[0054] (2) Outcome indicators: large-section construction excavation methods, advanced pre-reinforcement measures, etc.; large-section construction excavation methods include: full-section excavation method, two-stage excavation method, three-stage excavation method, etc.; advanced pre-reinforcement measures include: core soil periphery pre-reinforcement measures, core soil pre-reinforcement measures, core soil self-supporting measures, tunnel face temporary support measures, tunnel bottom reinforcement measures, and surface reinforcement measures;

[0055] Specifically, this embodiment redefines the large-section excavation method based on the operating space requirements of general construction machinery in tunnel engineering and the needs of engineering mechanization. It mainly refers to three excavation methods suitable for mechanical construction: full-section, two-stage, and three-stage.

[0056] Literature review reveals that most tunnel instability in weak surrounding rock begins near the tunnel face. The stability of the surrounding rock near the tunnel face directly impacts the safety and stability of the tunnel project. This embodiment redefines the classification of tunnel face pre-reinforcement technologies for weak surrounding rock tunnels based on the relationship and action type between construction auxiliary engineering measures and the pre-core soil at the tunnel face. These technologies include pre-reinforcement measures around the core soil, pre-reinforcement measures for the core soil, self-supporting measures for the core soil, temporary support measures for the tunnel face, tunnel bottom reinforcement measures, and surface reinforcement measures.

[0057] The main pre-reinforcement measures for the perimeter of the core soil are: advanced anchor bolts, advanced pipe roofs, insert plates, advanced small guide pipes, pre-lining, and pipe curtains.

[0058] The core soil pre-reinforcement measures mainly include: long anchor bolts at the working face, curtain grouting, borehole drainage, horizontal jet grouting piles, and wellpoint dewatering.

[0059] The core soil self-support measures are mainly divided into: core soil retention, long steps, short steps, micro steps, and face slopes.

[0060] Temporary support measures for the tunnel face mainly include: central diaphragm wall, double side wall guide tunnel, cross central diaphragm wall, counter-pressure backfill, and tunnel face sealing.

[0061] The main reinforcement measures for the bottom of the tunnel are: arch bottom reinforcement and arch foot reinforcement.

[0062] Surface reinforcement measures mainly include: surface anchor bolts, surface grouting, and surface jet grouting piles.

[0063] The support system mainly refers to the support system consisting of: the pre-reinforcement measures to reinforce the surrounding rock mass near the working face, the initial support composed of anchor bolts, shotcrete, and steel arches, and the secondary support composed of cast-in-place reinforced concrete lining.

[0064] Advanced pre-reinforcement measures mainly control the rate of deformation development at the working face and ensure construction safety.

[0065] The main function of initial support is to provide support force in a timely manner and to form a support-surrounding rock bearing system together with the surrounding rock in close contact. The initial support bears the ground pressure and the redistributed stress that the surrounding rock cannot bear, and controls the surrounding rock from undergoing large deformation due to excessive load.

[0066] Secondary lining is generally used as a safety reserve for tunnel engineering structures. However, for weak surrounding rock that cannot be stabilized for a long time, or for strata with strict deformation control and small allowable deformation values ​​of the surrounding rock, secondary lining needs to be constructed as early as possible to bear part of the surrounding rock pressure and control the final deformation stability of the surrounding rock.

[0067] Step S102: Quantify and encode the technical parameters to generate feature vectors for basic indicators, construction excavation method, and pre-reinforcement measures.

[0068] Furthermore, in the step of quantifying and coding the technical parameters, a single-heat coding method is used to code the construction excavation method, and a multi-heat coding method is used to code the advanced pre-reinforcement measures.

[0069] Specifically, this embodiment summarizes various technical parameters that affect the selection of advanced pre-reinforcement systems and construction methods for tunnel faces in weak surrounding rock.

[0070] Based on the characteristics of each technical parameter, quantitative coding rules (including single-thermal, multi-thermal, binary, numerical, and ordered numerical mapping coding rules) are formulated. Feature vectors for each technical parameter are established, and these vectors are divided into basic indicator feature vectors and outcome indicator feature vectors. Specifically, single-thermal coding is generally used for large-section construction excavation methods, while multi-thermal coding is generally used for advanced pre-reinforcement measures. See Table 1 for examples of sample basic indicators, outcome indicators, and their coding rules.

[0071] Table 1

[0072]

[0073]

[0074] Note: The geological conditions and excavation parameters of the tunnel in Table 1 are only a part of the examples. Other parameters can be coded in a similar way.

[0075] This embodiment uses a literature review approach to establish a sample database of engineering case studies on the pre-reinforcement system and construction methods for large-section tunnels in weak surrounding rock. It statistically analyzes various technical parameters for each engineering case, determines the feature vector of each sample (including basic indicator feature vectors and outcome indicator feature vectors), forms a quantified sample matrix, and divides the sample matrix into a basic indicator matrix A∈R. m×n The matrix of large-section construction excavation methods and performance indicators B∈R m×b , Result Indicator Pre-Reinforcement Measures Matrix C∈R m×c Note that m is the number of samples, n is the nth column of features in the sample basic index matrix, b is the bth column of features in the large section construction excavation method matrix of the results index (b types of large section construction excavation methods), and c is the cth column of features in the pre-reinforcement measures matrix of the results index (c types of advanced pre-reinforcement measures).

[0076] To facilitate calculation, the following factors were selected as the main influencing factors in the geological type of the surrounding rock and excavation parameters: rock weathering degree, surrounding rock type, water-richness, surrounding rock grade, deep / shallow burial, span, width / height, and ln excavation area. The basic indicators of the sample data (including geological conditions of the surrounding rock and excavation parameters, etc.) and their codes are detailed in Table 2. The corresponding result indicators of the sample data (including advanced pre-reinforcement measures, construction excavation methods, etc.) and their codes are detailed in Table 3.

[0077] Table 2

[0078]

[0079]

[0080]

[0081] Table 3

[0082]

[0083]

[0084] Step S2: Based on the degree of influence of each feature in the basic indicator feature vector on the result indicator, assign weights to each feature to form a weight vector, and obtain the weighted basic indicator matrix based on the weight vector;

[0085] Specifically, based on the degree of influence of the eigenvalue basic indicators (various technical parameter indicators) in the sample feature vector on the selection of large-section construction excavation methods and advanced pre-reinforcement measures for the outcome indicators, each eigenvalue basic indicator is assigned a weight ω = [ω1, ω2, ..., ω] to assess the degree of influence. n ];

[0086] In this embodiment, since different parameters have different degrees of influence on the tunnel's pre-reinforcement measures and excavation methods, based on the historical database, the cross-entropy loss rule is used to calculate the feature weights of the surrounding rock weathering degree (4-dimensional), surrounding rock type (6-dimensional), whether it is water-rich (1-dimensional), surrounding rock grade (1-dimensional), deep burial / shallow burial (1-dimensional), span (4-dimensional), width / height (1-dimensional), and ln excavation area (1-dimensional).

[0087] Calculating the optimal feature weights that fit the sample database involves complex processes such as high-order matrices, iterative calculations, and determining the convergence of the loss function. This embodiment uses an expert survey method to determine the feature weight vector (19-dimensional) of the sample database, where the feature weights are represented as follows:

[0088] ω

[0089] =[0.2,0.2,0.2,0.2,0.15,0.15,0.15,0.15,0.15,0.15,0.15,0.2,0.05,0.1,0.1,0.1,0.1,0.1,0.1,0.1,0.1] T ∈R 19×1 .

[0090] Step S3: Based on the weighted basic index matrix and Gaussian kernel function, calculate the basic index feature vector of the new engineering sample and the similarity vector of each sample in the sample database;

[0091] Furthermore, the steps for calculating the basic indicator feature vector of the new engineering sample and the similarity vector of each sample in the sample database include:

[0092] Step S301: Use the weight vector to perform weighted processing on the sample basic index matrix to obtain the weighted sample basic index matrix;

[0093] Step S302: Use the weight vector to weight the basic index feature vector of the new project sample to obtain the weighted feature vector of the new project sample;

[0094] Step S303: Based on the Gaussian kernel function, calculate the similarity between each sample in the weighted sample basic index matrix and the feature vector of the weighted new engineering sample to form a similarity vector.

[0095] Specifically, the process of determining the weight vector ω and the Gaussian kernel function parameters σ includes:

[0096] The sample basic index matrix A is obtained by weighting A using ω. ω The similarity matrix S between samples is calculated based on the Gaussian kernel function κ(x,y). ω ∈R m×m Matrix S ω The element (S)ω ) ij It can be represented as:

[0097]

[0098] Calculate the similarity of basic indicators among samples in the sample database to obtain the Gaussian kernel similarity matrix S. ω ∈R m×n , (S ω ) ij =s ij =к(a) ω,i ,a ω,j ),Right now:

[0099]

[0100] Where, when i = j, the sample self-similarity s ij This method is not applicable in leave-one-out cross-validation, where sij is usually set to 0.

[0101] Then, using the sample database result index matrices B and C, the feature weights and the parameter σ in the Gaussian kernel function are optimized using leave-one-out cross-validation. The parameters that enable the model to generalize the most are obtained through iterative calculation. It should be noted that when the number of samples in the sample database is small or in the early stages of calculation, the Gaussian kernel parameter σ0 and weight ω0 can be initially determined by expert survey. In actual implementation, when the result index does not match the actual situation or the number of samples in the sample database increases by a certain amount, the optimization program for the weight vector ω and the Gaussian kernel function parameter σ is started to redetermine ω and σ.

[0102] In this embodiment, the basic index matrix A (18×19) in the sample library can be obtained from Table 2 as follows:

[0103]

[0104] Based on the optimized feature weight vector, the weighted sample basic index matrix A is Aij. ω New sample basic index vector a new ∈R 1×n The weighted average is a new,ω Then, matrix A is calculated based on the Gaussian kernel function κ(x,y) and the optimized parameter σ. ω The element and the new sample a new,ω The similarity matrix s between new The formula is:

[0105]

[0106] This embodiment is mainly used to analyze the basic index feature vector a of the tunnel face in large-section excavation of soft surrounding rock. newEstablish a mathematical relationship with the basic indicator feature matrix A of the sample database, and calculate the similarity vector s. new .

[0107] This embodiment selects a new sample with basic parameters consistent with those in the sample data, including parameters such as rock weathering degree, surrounding rock type, water-richness, surrounding rock grade, deep / shallow burial, span, width / height, and ln excavation area. A summary table of the geological conditions, excavation parameters, and data codes of the new sample is shown in Table 4.

[0108] Table 4

[0109]

[0110]

[0111] In this embodiment, the new sample basic index vector (1×19) is represented as:

[0112] a new =[0 0 1 1 0 0 0 0 0 1 0 5 1 0 0 1 0 1.1.2153 4.4757]∈R 1×19 .

[0113] The weighted basic index matrix is ​​as follows:

[0114] A ω =A·diag(ω);

[0115] a ω,new =a new ·diag(ω);

[0116] Where, diag(ω) represents diag(ω) with respect to ω i It is a diagonal matrix with diagonal elements.

[0117] Step S4: Calculate the construction excavation method probability vector of the new project sample based on the similarity vector and the construction excavation method matrix of the result index in the sample database.

[0118] Furthermore, the steps for calculating the probability vector of the construction excavation method for the new project sample include:

[0119] Step S401: Multiply the transpose of the similarity vector with the construction excavation method matrix of the result index to obtain the first intermediate result;

[0120] Step S402: Multiply the transpose of the similarity vector with the all-one vector to obtain the second intermediate result;

[0121] Step S403: Divide the first intermediate result by the second intermediate result to obtain the construction excavation method probability vector.

[0122] Specifically, based on the construction excavation method matrix B of the sample database results indicators, combined with the calculated similarity matrix s new The normalized probability vector p of the construction excavation method is calculated using the following formula:

[0123]

[0124] Wherein, the probability vector of the construction excavation method is p = [p1, p2, ... p b ],satisfy 1 m This represents an m-dimensional column vector of all 1s.

[0125] In this embodiment, the kernel similarity matrix s is calculated based on the Gaussian kernel function. new The formula is:

[0126]

[0127] Among them, σ needs to be determined by cross-validation using a sample library, and is tentatively set in this calculation. In subsequent calculations, the value of σ is gradually adjusted as the number of samples in the sample library increases. Therefore:

[0128]

[0129] The formula for calculating the prediction vector is:

[0130]

[0131] Step S5: Determine the recommended construction excavation method based on the probability vector of the construction excavation method;

[0132] Furthermore, the step of determining the recommended construction excavation method is as follows: select the construction excavation method with the highest probability value in the probability vector of the construction excavation methods as the recommended construction excavation method.

[0133] Specifically, the recommended construction excavation method is KW. p The formula for calculating (one-hot encoding) is:

[0134]

[0135] This embodiment mainly uses the similarity matrix s ω A mathematical relationship is established between the probability vector p of the construction excavation method for the large-section excavation face of a tunnel in weak surrounding rock, which requires quantitative calculation, and the sample result index construction excavation method matrix B. The maximum value of p in the construction excavation method probability vector, max(p...), is then taken. j The corresponding construction excavation method is the large-section excavation of the tunnel face in weak surrounding rock, and the resulting indicators of the construction excavation method are as follows:

[0136] Step S6: Construct a construction excavation method condition matrix based on the similarity vector, the result index construction excavation method matrix, and the result index pre-reinforcement measure matrix in the sample database;

[0137] Furthermore, the steps for constructing the construction excavation method condition matrix include:

[0138] Step S601: Multiply the transpose of the construction excavation method matrix of the result index with the diagonalized similarity vector to obtain the first matrix;

[0139] Step S602: Multiply the transpose of the construction excavation method matrix of the result index with the similarity vector, and diagonalize and invert the result to obtain the second matrix;

[0140] Step S603: Multiply the second matrix with the first matrix to obtain the construction excavation method condition matrix.

[0141] Specifically, based on the similarity matrix s new Combining the large-section construction excavation method matrix B and the advanced pre-reinforcement measures matrix C from the sample database results, a construction large-section excavation method condition matrix M is constructed, with the formula as follows:

[0142] M=(diag(B T ·s new )) -1 ·(B T ·(diag(s new )·C)), M∈R b×c ;

[0143] Among them, diag(s) new ) is based on s new It is a diagonal matrix with diagonal elements.

[0144] Based on the single-hot coding features of large-section excavation methods, let the method vector for large-section construction excavation be... When determining the excavation method for the large cross-section in the j-th ∈ {1,2,...,b} All other elements are 0.

[0145] Based on the determined construction excavation method Combining the condition matrix M and the large-section excavation method vector b KW Then in determining Under this large-section excavation method, the formula for the activation probability vector r of the advanced pre-reinforcement measures (multi-heat coding) is:

[0146]

[0147] This is based on the large-section construction excavation method. The pre-reinforcement measures (multi-thermal coding) under the given conditions are as follows:

[0148]

[0149] Where i is an integer from 1 to c; τ1 is the threshold, which needs to be determined as a number between (0,1) during calculation.

[0150] In this embodiment, the activation probability vector *r* of advanced pre-reinforcement measures for large-section tunnel face excavation in weak surrounding rock, which needs to be quantitatively calculated, is established mathematically with the advanced pre-reinforcement measure matrix *C* of the sample database results indicators through the construction excavation method condition matrix *M*. Under the premise of determining the large-section construction excavation method, the activation probability vector *r* of advanced pre-reinforcement measures is calculated. When the element *r* in *r*... i When the value is greater than the threshold τ1, the corresponding advanced pre-reinforcement measures for the tunnel face are the advanced pre-reinforcement measures for the tunnel face of large-section excavation soft surrounding rock tunnels.

[0151] In this embodiment, based on Table 3, the advanced pre-reinforcement measures of the result index matrix in the sample database, and the excavation method matrix C (18×5) are as follows:

[0152]

[0153] In this embodiment, the condition matrix M for the large-section excavation method is formulated as follows:

[0154]

[0155] Step S7: Calculate the activation probability vector of the advanced pre-reinforcement measures based on the recommended construction excavation method and the condition matrix of the construction excavation method;

[0156] Further, the step of calculating the activation probability vector of the advanced pre-reinforcement measures is as follows: multiply the single-hot encoded vector representing the recommended construction excavation method with the condition matrix of the construction excavation method to obtain the activation probability vector of the advanced pre-reinforcement measures.

[0157] Specifically, the scheme comparison in this embodiment is based on the new sample construction excavation method. The corresponding vector is b KW The activation probability vector r of the advanced pre-reinforcement measures (multi-hot coding) is calculated as follows:

[0158] r = b KW ·M;

[0159] Step S8: Determine recommended advanced pre-reinforcement measures based on the activation probability vector of the advanced pre-reinforcement measures and the preset threshold.

[0160] Furthermore, the step of determining the recommended advanced pre-reinforcement measures is as follows: each element in the activation probability vector of the advanced pre-reinforcement measures is compared with the preset threshold. If the element value is greater than or equal to the preset threshold, the advanced pre-reinforcement measures corresponding to that element are adopted.

[0161] Specifically, since the new sample construction excavation method is single-heat coded, the maximum value is taken; the advanced pre-reinforcement measures based on the excavation method are multi-heat coded, and the threshold is set to 0.45. A summary table of the new sample schemes for tunnel faces in large-section excavation of weak surrounding rock is shown in Table 5.

[0162] Table 5

[0163]

[0164] Therefore, the recommended construction excavation method for the new sample is the "step-up and step-down method", and the advanced pre-reinforcement measures are "pipe roof + advanced small guide pipe". If necessary, "advanced pre-reinforcement anchor bolts" can be used.

[0165] The beneficial effects of this embodiment:

[0166] This embodiment redefines the classification of large-section excavation methods and face pre-reinforcement technologies for tunnels in weak surrounding rock, laying a conceptual foundation for the unification and standardization of design parameters and construction technologies for tunnels in weak surrounding rock.

[0167] This embodiment uses mathematical methods and an engineering case sample database to quantitatively calculate the pre-reinforcement parameters and construction methods for tunnel faces suitable for large-section excavation in different soft surrounding rock tunnel projects by establishing similarity relationships. This is closer to engineering practice and more instructive than directly referencing relevant specifications.

[0168] This embodiment establishes an engineering case database of advanced pre-reinforcement system and construction methods for tunnel faces in weak surrounding rock, and uses mathematical methods to establish logical relationships. This eliminates the various drawbacks caused by technicians selecting support parameters and construction techniques based on experience, improves the adaptability of technical parameters to the actual project, and effectively controls project costs.

[0169] Based on numerous engineering cases and applying mathematical principles, this embodiment selects pre-reinforcement parameters for the tunnel face suitable for large-section excavation in specific projects. The conclusions show good adaptability and a high degree of standardization, which is conducive to the standardized construction of tunnels in weak surrounding rock, improves construction efficiency and project quality, ensures construction safety, and can also drive the advancement of tunnel engineering technology in weak surrounding rock.

[0170] This embodiment, based on the design and construction process of tunnels in weak surrounding rock, and using the Gaussian kernel function as a preset condition, designs a method for selecting pre-reinforcement measures and construction techniques for tunnel faces in weak surrounding rock tunnel engineering. This method can reasonably and quickly match pre-reinforcement measures for the tunnel face. This embodiment performs a unified analysis and evaluation of various technical parameter data, and mathematically streamlines the evaluation process to arrive at scientifically sound conclusions. This embodiment guides the standardization and unification of the selection of pre-reinforcement measures and construction techniques for soft rock tunnel faces, promoting the development of soft surrounding rock tunnel engineering technology in China.

[0171] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for pre-reinforcement of the tunnel face in soft surrounding rock, applicable to large-section excavation, characterized in that, Includes the following steps: The classification of large-section excavation methods and pre-reinforcement technology for tunnel face in weak surrounding rock is redefined. Based on the classification results, a sample database containing multiple historical engineering cases of tunnels in weak surrounding rock is established, in which each sample includes a feature vector of basic indicators and a feature vector of outcome indicators. Based on the degree of influence of each feature in the basic indicator feature vector on the result indicator, weights are assigned to each feature to form a weight vector, and the weighted basic indicator matrix is ​​obtained based on the weight vector. Based on the weighted basic index matrix and Gaussian kernel function, the basic index feature vector of the new engineering sample and the similarity vector of each sample in the sample database are calculated. Based on the similarity vector and the construction excavation method matrix of the result indicators in the sample database, calculate the construction excavation method probability vector of the new project sample. Based on the probability vector of the construction excavation method, a recommended construction excavation method is determined; Based on the similarity vector, the construction excavation method matrix of the achievement indicators, and the pre-reinforcement measures matrix of the achievement indicators in the sample database, a construction excavation method condition matrix is ​​constructed. Based on the recommended construction excavation method and the condition matrix of the construction excavation method, calculate the activation probability vector of the advanced pre-reinforcement measures; Based on the activation probability vector of the aforementioned advanced pre-reinforcement measures and the preset threshold, recommended advanced pre-reinforcement measures are determined.

2. The method according to claim 1, characterized in that, The steps to establish a sample database containing multiple historical engineering cases of tunnels in weak surrounding rock include: Obtain the technical parameters of multiple historical engineering cases of tunnels in weak surrounding rock; The technical parameters are quantified and encoded to generate feature vectors for basic indicators, feature vectors for construction excavation methods and feature vectors for pre-reinforcement measures.

3. The method according to claim 2, characterized in that, In the step of quantifying and coding technical parameters, a single-heat coding method is used to code the construction excavation method, and a multi-heat coding method is used to code the advanced pre-reinforcement measures.

4. The method according to claim 1, characterized in that, The steps for calculating the basic indicator feature vector of the new project sample and the similarity vector of each sample in the sample database include: The sample basic index matrix is ​​weighted using the weight vector to obtain a weighted sample basic index matrix. The weighted feature vector of the new engineering sample is weighted by the weight vector to obtain the weighted feature vector of the new engineering sample. Based on the Gaussian kernel function, the similarity between each sample in the weighted sample basic index matrix and the feature vector of the weighted new engineering sample is calculated to form a similarity vector.

5. The method according to claim 1, characterized in that, The steps for calculating the probability vector of construction excavation methods for a new engineering sample include: Multiply the transpose of the similarity vector by the construction excavation method matrix of the result index to obtain the first intermediate result; Multiply the transpose of the similarity vector by the all-one vector to obtain the second intermediate result; Divide the first intermediate result by the second intermediate result to obtain the construction excavation method probability vector.

6. The method according to claim 1, characterized in that, The steps to determine the recommended construction excavation method are as follows: The construction excavation method with the highest probability value in the probability vector of the construction excavation methods is selected as the recommended construction excavation method.

7. The method according to claim 1, characterized in that, The steps for constructing the construction excavation method condition matrix include: The first matrix is ​​obtained by multiplying the transpose of the construction excavation method matrix of the aforementioned achievement indicators with the diagonalized similarity vector. Multiply the transpose of the construction excavation method matrix of the aforementioned achievement indicators by the similarity vector, and then diagonalize and invert the result to obtain the second matrix; Multiplying the second matrix by the first matrix yields the construction excavation method condition matrix.

8. The method according to claim 1, characterized in that, The steps for calculating the activation probability vector of advanced pre-reinforcement measures are as follows: Multiplying the single-hot encoded vector representing the recommended construction excavation method with the construction excavation method condition matrix yields the activation probability vector of the advanced pre-reinforcement measures.

9. The method according to claim 1, characterized in that, The steps for determining recommended pre-reinforcement measures are as follows: Each element in the activation probability vector of the advanced pre-reinforcement measures is compared with the preset threshold. If the element value is greater than or equal to the preset threshold, the advanced pre-reinforcement measure corresponding to that element is adopted.

10. The method according to claim 1, characterized in that, The weight vector and the parameters of the Gaussian kernel function are optimized using leave-one-out cross-validation.

Images