Decision tree determination method, submarine tunnel surrounding rock impermeability classification method and product

Abstract

The application discloses a decision tree determination method, a submarine tunnel surrounding rock impermeability grading method and a product, relates to the technical field of submarine tunnel surrounding rock impermeability grading, and the determination method comprises the following steps: acquiring training data of a tunnel section; the training data comprises surrounding rock parameter data and corresponding impermeability level data; the surrounding rock parameters comprise rock cover layer thickness, water head height, rock uniaxial saturated compressive strength and surrounding rock volume joint number; the training data is classified by using dichotomy to obtain a plurality of division points and corresponding division results; the information gain of each division point is calculated in combination with an information gain criterion; the division point with the maximum information gain and the corresponding division result are selected, two data sets in the division result are taken as "training data" respectively, the above steps are repeatedly executed, and a decision tree for impermeability grading is obtained. The application can accurately perform surrounding rock impermeability grading, makes the surrounding rock impermeability grading simple and intuitive, and can objectively reflect rock mass impermeability.

Description

Technical Field

[0001] This application relates to the field of subsea tunnel surrounding rock permeability classification technology, and in particular to a decision tree determination method, a subsea tunnel surrounding rock permeability classification method and product. Background Technology

[0002] Subsea tunnels, due to their V-shaped longitudinal slope, cannot allow seepage to drain naturally, making drainage a critical challenge for their construction and operational safety. Looking at underwater tunnels both domestically and internationally, drainage methods can be categorized into two types: fully sealed and drainage-guided, both using secondary lining as the main load-bearing structure. Fully sealed tunnel linings bear water pressure comparable to groundwater head, making them unsuitable for tunnels with high overlying aquifers and deep burial depths. Drainage-guided drainage systems can reduce external water pressure on the lining structure to a certain extent, significantly lowering the probability of deterioration and leakage, making the lining structure more economical and efficient. Currently, subsea tunnels often employ a "water-blocking and drainage-limiting" waterproofing system, controlling drainage volume while considering the stability of the surrounding rock reinforcement ring. However, the surrounding rock reinforcement ring and initial support structure bear corresponding water loads while blocking water, necessitating a comprehensive consideration of the safety and economy of the drainage system. To achieve overall safety in subsea tunnels using a "water-blocking and drainage-limiting" drainage system, an active control drainage system needs to be constructed. Its core features lie in the active determination of the allowable drainage volume and the collaborative design of the water-blocking system.

[0003] Drainage systems are influenced by a combination of factors, including surrounding rock conditions and hydrogeological conditions. The main differences in their design schemes lie in the understanding of the surrounding rock's water-blocking capacity and the determination of allowable drainage volume. Although domestic and international scholars have conducted extensive research on the design theory and technology of drainage systems for submarine tunnels, the allowable drainage volume for submarine tunnels mainly relies on engineering experience and analogy, and a unified standard for reference has not yet been formed. The fundamental reason for this is insufficient understanding of the surrounding rock's water-blocking capacity, resulting in a lack of theoretical basis for drainage design. For submarine tunnels, zoned waterproofing design is often required, and in current engineering projects, this design often relies on the results of surrounding rock classification. Existing surrounding rock classification evaluates the strength of the surrounding rock, mainly determined by the rock's hardness and rock mass integrity. However, surrounding rock permeability classification evaluates the permeability of the surrounding rock, which needs to be determined by the hydrogeological conditions of the tunnel location and the permeability of the rock mass. If the surrounding rock grade is low, but the tunnel is located in an area with a high water head, its permeability grade may be high; if the surrounding rock grade is low and its integrity is good, but the joints form good water-passing channels with strong permeability, its permeability grade may be high. Current research lacks a reliable classification method for directly and objectively evaluating the impermeability of surrounding rock. Summary of the Invention

[0004] The purpose of this application is to provide a decision tree determination method, a method and product for classifying the impermeability of surrounding rock in submarine tunnels, which can accurately classify the impermeability of surrounding rock in submarine tunnels, and make the classification of surrounding rock impermeability simple and intuitive, and can objectively reflect the impermeability of the rock mass.

[0005] To achieve the above objectives, this application provides the following solution:

[0006] Firstly, this application provides a decision tree determination method, the decision tree determination method comprising:

[0007] Acquire training data for the tunnel cross-section; the training data includes: surrounding rock parameter data and corresponding impermeability level data; the surrounding rock parameter data includes: surrounding rock parameters and corresponding engineering data; the surrounding rock parameters include: rock overburden thickness, water head height, uniaxial saturated compressive strength of rock, and number of volumetric joints in the surrounding rock;

[0008] The training data is classified using a binary classification method to obtain several split points and corresponding split results; the split results include: a large training data set and a small training data set; the large training data set is the set of training data that is greater than the split point; the small training data set is the set of training data that is less than the split point.

[0009] The information gain of each of the partition points is calculated using the information gain criterion.

[0010] The first partition point and its corresponding partition result are obtained by selecting the partition point with the largest information gain; the first partition result includes: a first large training data set and a first small training data set; the first large training data set is the set of training data greater than the first partition point; the first small training data set is the set of training data less than the first partition point.

[0011] The first large training dataset and the first small training dataset are respectively combined into "training data", and the step of "classifying the training data by binary division to obtain several split points and corresponding split results" is returned until the step of "selecting the split point with the largest information gain and the corresponding split result to obtain the first split point and the corresponding first split result" is completed. Then, there is only one data in the first large training dataset and the first small training dataset, and a decision tree for the classification of the impermeability of the surrounding rock of the submarine tunnel is obtained.

[0012] Optionally, the formula for calculating the information gain of each partition point using the information gain criterion is as follows:

[0013]

[0014] Where Gain(D,a) represents the information gain at the split point t, and Ent(D) represents the information entropy of the entire training dataset D. This represents the set of partition results corresponding to partition point t. Information entropy This represents the large training data set corresponding to the split point t. This represents the small training data set corresponding to the split point t.

[0015] Optionally, the formula for calculating information entropy is:

[0016]

[0017] Where, p k Let be the proportion of the k-th class of samples in set D, where k = 1, 2, ..., n.

[0018] Optionally, the step of classifying the training data using a binary classification method to obtain several dividing points and corresponding dividing results specifically includes:

[0019] Input the training data into the decision tree model;

[0020] Using a decision tree model, the training data is classified using a binary classification method to obtain several split points and corresponding split results.

[0021] Optionally, the training data is input into the decision tree model, specifically including:

[0022] The training data is organized into a decision tree model matrix; the decision tree model matrix includes: a training dataset and an attribute set; the training dataset D = {(x1,y1),(x2,y2),…,(x...} 42 ,y 42 )}, where x i (i = 1, 2, ..., 42) is a 4-dimensional vector, representing the engineering data for four evaluation indicators: rock overburden thickness, water head height, number of volume joints, and uniaxial saturated compressive strength of the rock. i (i = 1, 2, ..., 42) represents the corresponding impermeability level; the attribute set A = {a1, a2, a3, a4}, where a1, a2, a3, a4 are the rock overburden thickness, water head height, volume joint number, and uniaxial saturated compressive strength of the rock, respectively.

[0023] Input the decision tree model matrix into the decision tree model.

[0024] Optionally, the training data for the tunnel cross-section is training data for 52 typical tunnel cross-sections.

[0025] Secondly, this application provides a method for classifying the impermeability of surrounding rock in submarine tunnels, characterized in that the method includes:

[0026] Obtain the surrounding rock parameters and corresponding engineering data of the surrounding rock to be graded;

[0027] The surrounding rock parameters and corresponding engineering data are input into a decision tree to obtain the classification of the surrounding rock to be classified; the decision tree is determined by any of the decision tree determination methods described above.

[0028] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the decision tree determination method or the subsea tunnel surrounding rock impermeability classification method as described above.

[0029] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the decision tree determination method or the subsea tunnel surrounding rock impermeability classification method described in any one of the above descriptions.

[0030] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the decision tree determination method or the subsea tunnel surrounding rock impermeability classification method described in any one of the above descriptions.

[0031] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0032] This application provides a decision tree determination method, a method for classifying the impermeability of surrounding rock in submarine tunnels, and a product thereof. The decision tree determination method includes: acquiring training data of tunnel cross-sections; the training data includes: surrounding rock parameter data and corresponding impermeability level data; the surrounding rock parameters include: rock overburden thickness, hydraulic head, uniaxial saturated compressive strength of rock, and the number of volumetric joints in the surrounding rock; classifying the training data using a binary classification method to obtain several split points and corresponding classification results; the classification results include: a large training data set and a small training data set; the large training data set is the set of training data greater than the split point; the small training data set is the set of training data less than the split point; calculating the information gain of each split point using the information gain criterion; selecting the split point with the largest information gain and its corresponding classification result. The results are divided to obtain a first split point and a corresponding first split result. The first split result includes a first large training data set and a first small training data set. The first large training data set is the set of training data larger than the first split point. The first small training data set is the set of training data smaller than the first split point. The first large training data set and the first small training data set are respectively combined into "training data", and the step of "classifying the training data using the binary method to obtain several split points and corresponding split results" is returned until the step of "selecting the split point with the largest information gain and the corresponding split result to obtain the first split point and the corresponding first split result" is completed. At this point, there is only one data in the first large training data set and the first small training data set, resulting in a decision tree for permeability classification. The purpose of this application is to provide a decision tree determination method, a method and product for classifying the permeability of surrounding rock in submarine tunnels. The proposed permeability classification indicators include rock overburden thickness, hydraulic head, uniaxial saturated compressive strength of the rock, and the number of volume joints, thus ensuring the accuracy of surrounding rock permeability classification. Furthermore, by performing machine learning on a large amount of data, and combining the dichotomy method, information gain criterion, and information gain rate criterion, a decision tree prediction model is established, enabling the classification prediction of continuous attribute data. Finally, a surrounding rock decision tree is formed, making the permeability classification of surrounding rock simple and intuitive, and objectively reflecting the permeability of the rock mass. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1This is an application environment diagram of a decision tree determination method or a method for classifying the impermeability of surrounding rock in a submarine tunnel according to an embodiment of this application.

[0035] Figure 2 A flowchart illustrating a decision tree determination method provided in an embodiment of this application;

[0036] Figure 3 A schematic diagram of a decision tree for classifying the impermeability of surrounding rock in a submarine tunnel, provided as an embodiment of this application;

[0037] Figure 4 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0039] In fact, due to the complex hydrogeological conditions and varying connectivity with water in the submarine tunnel, in order to proactively determine the allowable drainage volume of the tunnel and provide a design method for the drainage and waterproofing system, it is necessary to objectively evaluate the water-blocking capacity of the surrounding rock of the submarine tunnel, i.e., the impermeability of the surrounding rock, and formulate different drainage volume control standards accordingly. This provides a theoretical basis for the collaborative design of the drainage and waterproofing system of the submarine tunnel and also has certain reference value for the drainage and waterproofing design of water-rich tunnels.

[0040] This application uses rock overburden thickness, water head height, uniaxial saturated compressive strength of rock, and volumetric joint number as evaluation indicators for permeability classification. By using machine learning to establish a decision tree model with analyzable continuous value attributes based on a large amount of tunnel seepage case data, it is possible to classify the permeability of tunnel surrounding rock. The main steps include:

[0041] (1) Based on the prediction formula of the original seepage volume of the fractured rock mass submarine tunnel under high-speed flow conditions during the construction period, and combined with the statistical analysis of a large number of typical underwater tunnel and water-rich tunnel cross-section seepage case data, the thickness of the rock overburden, the water head height, the uniaxial saturated compressive strength of the rock and the number of volume joints were determined as the impermeability classification index. The tunnel water inflow was classified by the four-part method as a reference for the impermeability classification of the surrounding rock.

[0042] (2) The tunnel surrounding rock sample data was divided using the dichotomy method, information gain, and information gain ratio criteria to determine the data division points. Based on these division points, a decision tree prediction model was established, enabling hierarchical prediction of continuous attribute data. Using this decision tree model, the tunnel surrounding rock parameters can be searched and classified to obtain the impermeability classification of the surrounding rock.

[0043] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0044] The decision tree determination method or the subsea tunnel surrounding rock impermeability classification method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on other servers.

[0045] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0046] In one exemplary embodiment, such as Figure 2 As shown, a decision tree determination method is provided. This method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps S1 to S5. Wherein:

[0047] S1. Obtain training data for the tunnel cross-section; the training data includes: surrounding rock parameter data and corresponding impermeability level data; the surrounding rock parameter data includes: surrounding rock parameters and corresponding engineering data; the surrounding rock parameters include: rock overburden thickness, water head height, uniaxial saturated compressive strength of rock, and number of volumetric joints in the surrounding rock.

[0048] In this embodiment, the following indicators are selected for evaluating the impermeability of tunnel surrounding rock: rock overburden thickness, water head height, uniaxial saturated compressive strength of rock, and number of volumetric joints in the surrounding rock. The classification criteria for the impermeability levels of surrounding rock are shown in Table 1.

[0049] Table 1. Standards for Classification of Permeability Resistance Levels

[0050] <![CDATA[Water seepage volume (m 3 ·d -1 ·m -1 )]]> grade 0～3.32 Ⅰ (Damp) 3.32～8.47 II (dripping water) 8.47～22.28 Ⅲ (Linear flow) >22.28 IV (Flooding)

[0051] This embodiment collected surrounding rock parameter data and corresponding impermeability level data for 52 typical tunnel sections. Some statistical data are shown in Table 2 below:

[0052] Table 2. Statistical data of some typical tunnel sections.

[0053]

[0054]

[0055] The four evaluation indicators—rock overburden thickness, water head height, volumetric joint number, and uniaxial saturated compressive strength—are represented as A = {a1, a2, a3, a4}, all of which are continuous attributes. Simultaneously, the surrounding rock permeability is divided into four levels based on the seepage rate, i.e., y i ∈{Ⅰ,Ⅱ,Ⅲ,Ⅳ},(i=1,2,...,n). The statistical data of some typical tunnel cross sections above are split into training set and test set in a 4:1 ratio, that is, the first 42 sets of data are the training set and the last 10 sets of data are the test set.

[0056] S2. The training data is classified using a binary classification method to obtain several dividing points and corresponding dividing results; the dividing results include: a large training data set and a small training data set; the large training data set is the set of training data that is greater than the dividing point; the small training data set is the set of training data that is less than the dividing point.

[0057] Specifically:

[0058] S21. Input the training data into the decision tree model.

[0059] The training data is organized into a decision tree model matrix; the decision tree model matrix includes: a training dataset and an attribute set; the training dataset D = {(x1,y1),(x2,y2),…,(x...} 42 ,y 42 )}, where x i (i = 1, 2, ..., 42) is a 4-dimensional vector, representing the engineering data for four evaluation indicators: rock overburden thickness, water head height, number of volume joints, and uniaxial saturated compressive strength of the rock. i (i = 1, 2, ..., 42) represent the corresponding impermeability levels; the attribute set A = {a1, a2, a3, a4}, where a1, a2, a3, a4 are the rock overburden thickness, hydraulic head, number of volume joints, and uniaxial saturated compressive strength of the rock, respectively. The decision tree model matrix is ​​then input into the decision tree model.

[0060] The following section describes feature selection for each evaluation indicator to generate the root node of the decision tree.

[0061] S22. Using a decision tree model, the training data is classified using a binary classification method to obtain several split points and corresponding split results.

[0062] S3. Calculate the information gain of each partition point using the information gain criterion. The calculation formula is as follows:

[0063]

[0064] Where Gain(D,a) represents the information gain at the split point t, and Ent(D) represents the information entropy of the entire training dataset D. This represents the set of partition results corresponding to partition point t. Information entropy This represents the large training data set corresponding to the split point t. This represents the small training data set corresponding to the split point t.

[0065] The key to feature selection is choosing the optimal features that have the ability to classify the training data, thereby improving the classification efficiency of the decision tree. For each classification of data within an internal node, the goal is to ensure that the data in the next node belong to the same category as much as possible, i.e., higher "purity". Information entropy can measure the uncertainty of random variables and is one of the most commonly used indicators for measuring the purity of a sample set; the smaller its value, the higher the purity of the sample set. Let p be the proportion of samples of the k-th class in the sample set D. k (k=1,2,...,n), then the information entropy Ent(D) of set D is defined as:

[0066]

[0067] For continuous attributes, the number of possible values ​​is infinite, and we cannot directly partition the sample data within a node based on the possible values ​​of the continuous attribute. Therefore, we consider using a bisection method to discretize the continuous attribute. Suppose that the continuous attribute a in the sample set D has k distinct values ​​{m1, m2, m3, ..., mk}. k Sort the k values ​​in ascending order, denoted as {m}. 1 ,m 2 ,m 3 ,...,m k Let the set containing the k-1 partition points t be defined as follows:

[0068]

[0069] Among them, T a Let m be a set containing k-1 partition points t. iGiven the value of continuous attribute 'a' in ascending order, the i-th value is the partition point 't' that divides D into subsets. and in Includes samples where the value of continuous attribute a is greater than t. Includes samples where the value of continuous attribute a is less than or equal to t.

[0070] At this point, t can be considered a discrete attribute value, and the information entropy Ent(D) can be calculated using the aforementioned formula. and Information entropy, considering two subsets and Different subsets are assigned weights based on the number of samples included. Where λ∈{-,+}, Let D be the set of samples greater than and less than the partition point t after being partitioned. and |D| represents the number of samples in set D.

[0071] Therefore, the information gain Gain(D,a) of each partition point t can be calculated, that is:

[0072]

[0073] Where Gain represents information gain and Ent represents information entropy.

[0074] Taking the root node rock overburden thickness property as an example, the set partitioning formula is first used. The thickness of the overburden layer was discretized, and then the information gain of each candidate dividing point for this attribute was calculated using the information entropy calculation formula and the information gain calculation formula. The information gain of the rock overburden layer thickness (in descending order of information gain) is shown in Table 3.

[0075] Table 3 Information Gain Table for Partial Rock Overburden Thickness Division Points

[0076] Division point Information gain 220.00 0.302895712 215.50 0.236562694 412.00 0.215919612 235.00 0.211351419 260.00 0.202191118 329.00 0.200234818 …… …… 18.722 0.027855219 25.3315 0.027847104 19.15 0.024835007

[0077] S4. Select the split point with the largest information gain and the corresponding split result to obtain the first split point and the corresponding first split result; the first split result includes: the first large training data set and the first small training data set; the first large training data set is the set of training data greater than the first split point; the first small training data set is the set of training data less than the first split point.

[0078] The information gain of all evaluation index attributes was calculated using the above method, resulting in 204 candidate partitioning points. Some calculation results are shown in Table 4 (in descending order of information gain rate). The partitioning point with the largest information gain was selected as the optimal partitioning point. Thus, the partitioning point for the root node sample data is defined as the overburden thickness > 220m.

[0079] Table 4 Information Gain Table for Some Candidate Division Points

[0080] property Division point Information gain Cover thickness 220.00 0.302895712 water head height 156.63 0.299002614 water head height 146.43 0.292477282 water head height 134.39 0.280747145 water head height 124.19 0.279537018 Uniaxial saturated compressive strength 107.75 0.265164245 Uniaxial saturated compressive strength 70.00 0.264305872 water head height 118.73 0.272936082 Joint number 35 0.261076105 Uniaxial saturated compressive strength 100.40 0.253066785 …… …… …… water head height 19.9 0.008583831 Uniaxial saturated compressive strength 30.75 0.007550855 Uniaxial saturated compressive strength 33.85 0.005698035 Uniaxial saturated compressive strength 28.5 0.003184926

[0081] S5. Combine the first large training data set and the first small training data set into "training data" respectively, and return to the step of "classifying the training data using the binary method to obtain several split points and corresponding split results" until the step of "selecting the split point with the largest information gain and the corresponding split result to obtain the first split point and the corresponding first split result" is completed. Then, there will be only one data in the first large training data set and the first small training data set, thus obtaining a decision tree for the classification of the impermeability of the surrounding rock of the submarine tunnel.

[0082] Decision trees are recursively generated from the 42 sets of training data using the method described above. Figure 3 As shown, the decision tree model was evaluated using 10 test sets of data. The predicted impermeability level based on the decision tree was compared with the actual impermeability level. The number of correct predictions accounted for 90% of the test set groups, resulting in an accuracy rate of 90%, indicating that the model prediction effect was good.

[0083] In short, the technical solution of this application mainly includes the following three steps:

[0084] (1) The evaluation indicators for the impermeability of the surrounding rock of the tunnel are determined as the thickness of the rock overburden, the water head, the uniaxial saturated compressive strength of the rock, and the number of volumetric joints in the surrounding rock. At the same time, the permeability of the surrounding rock is classified as a standard for classifying the impermeability of the surrounding rock.

[0085] (2) The original sample dataset, namely the seepage case data of typical seabed and water-rich tunnel sections, is split into training set and test set in a 4:1 ratio. The training set data is analyzed and processed according to the dichotomy method, information gain criterion and information gain rate criterion to form the optimal split point. The decision tree is recursively generated, and the test set data is used to evaluate the generated decision tree model.

[0086] (3) Input the evaluation index parameters of the surrounding rock impermeability of the specified tunnel section into the decision tree. By classifying the nodes of the decision tree, the impermeability level of the surrounding rock of the tunnel section can be obtained.

[0087] This embodiment proposes rock overburden thickness, hydraulic head, uniaxial saturated compressive strength of rock, and volumetric joint number as permeability grading indicators, and also grades tunnel seepage as a reference for surrounding rock permeability grading. Furthermore, this embodiment utilizes machine learning on a large amount of data, combining the dichotomy method, information gain criterion, and information gain rate criterion to establish a decision tree prediction model. This achieves graded prediction of continuous attribute data, ultimately forming a decision tree for grading the permeability of surrounding rock in submarine tunnels. This makes the permeability grading of surrounding rock simple and intuitive, and can objectively reflect the permeability of the rock mass.

[0088] In one exemplary embodiment, such as Figure 2 As shown, a method for classifying the permeability resistance of surrounding rock in submarine tunnels is provided, the method comprising:

[0089] Step A1: Obtain the surrounding rock parameters and corresponding engineering data of the surrounding rock to be graded.

[0090] Step A2: Input the surrounding rock parameters and corresponding engineering data into the decision tree to obtain the classification of the surrounding rock to be classified; the decision tree is determined by the decision tree determination method described above.

[0091] Specifically, the surrounding rock parameters and corresponding engineering data of a certain tunnel section obtained in this embodiment are shown in Table 5:

[0092] Table 5. Rock surrounding parameters and corresponding engineering data for a tunnel.

[0093]

[0094] Input this data into the decision tree obtained above, and refer to... Figure 3 The decision tree shown indicates that the impermeability grade of the surrounding rock in the tunnel section is Class II.

[0095] This has led to the development of a classification method for the impermeability of surrounding rock in submarine tunnels, which has significant practical engineering value.

[0096] Compared with existing surrounding rock classification methods, this embodiment comprehensively considers the hydrogeological conditions of the tunnel location and the permeability of the rock mass, and proposes rock overburden thickness, hydraulic head, uniaxial saturated compressive strength of the rock, and volume joint number as evaluation indicators for the classification of surrounding rock permeability. The selected indicators are relatively easy to collect during the field investigation stage and have certain practicality for the subsequent design and construction stages. A decision tree is formed by combining machine learning methods with the dichotomy method, information gain criterion, and information gain rate criterion. The decision tree is simple and intuitive, and can directly and objectively classify the input surrounding rock parameters and thus predict the permeability level of the surrounding rock.

[0097] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 4 As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and databases. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a decision tree determination method or a method for classifying the impermeability of surrounding rock in a submarine tunnel.

[0098] Those skilled in the art will understand that Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0099] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0100] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0101] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0102] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0103] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0104] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0105] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0106] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method of determining a decision tree, characterized by, The decision tree determination method comprises: Obtaining training data of a tunnel section; the training data comprises surrounding rock parameter data and corresponding impermeability level data; the surrounding rock parameter data comprises surrounding rock parameters and corresponding engineering data; the surrounding rock parameters comprise rock cover layer thickness, water head height, rock uniaxial saturated compressive strength, and surrounding rock volume joint number; Classifying the training data by using dichotomy to obtain a plurality of division points and corresponding division results; the division results comprise a large training data set and a small training data set; the large training data set is a set of training data greater than the division points in the training data; the small training data set is a set of training data less than the division points in the training data; Calculating information gain of each division point in combination with an information gain criterion; Selecting a division point with maximum information gain and corresponding division results to obtain a first division point and corresponding first division results; the first division results comprise a first large training data set and a first small training data set; the first large training data set is a set of training data greater than the first division point; the first small training data set is a set of training data less than the first division point; Respectively taking the first large training data set and the first small training data set as "training data", and returning to the "classifying the training data by using dichotomy to obtain a plurality of division points and corresponding division results" step until the "selecting a division point with maximum information gain and corresponding division results to obtain a first division point and corresponding first division results" step is executed, and only one data is left in the first large training data set and the first small training data set, thereby obtaining a decision tree for sea tunnel surrounding rock impermeability classification.

2. The method of claim 1, wherein, The calculation formula for calculating information gain of each division point in combination with the information gain criterion is: wherein Gain(D, a) represents information gain of the division point t, Ent(D) represents information entropy of the entire training data set D, represents information entropy of the division result set corresponding to the division point t, represents the large training data set corresponding to the division point t, represents the small training data set corresponding to the division point t.​ 3. The method of claim 2, wherein, The calculation formula for information entropy is: where p k is the proportion of the kth class of samples in the set D, k = 1, 2,..., n.

4. The method of claim 1, wherein, The "classifying the training data by using dichotomy to obtain a plurality of division points and corresponding division results" specifically comprises: Inputting the training data into a decision tree model; Classifying the training data by using dichotomy to obtain a plurality of division points and corresponding division results by using the decision tree model.

5. The method of claim 4, wherein, The "inputting the training data into a decision tree model" specifically comprises: The training data is sorted into a decision tree model matrix; the decision tree model matrix comprises: a training data set and an attribute set; the training data set D={(x1, y1), (x2, y2), …, (x 42 ,y 42 )} wherein x i (i=1, 2, …, 42) is a 4-dimensional vector, respectively, the engineering data of the four evaluation indexes of rock cover layer thickness, water head height, volume joint number and rock uniaxial saturated compressive strength, y i (i=1, 2, …, 42) is the corresponding impermeability level; the attribute set A={a1, a2, a3, a4}, a1, a2, a3, a4 are rock cover layer thickness, water head height, volume joint number and rock uniaxial saturated compressive strength in turn; Inputting the decision tree model matrix into the decision tree model.

6. The method of claim 1, wherein, The training data of the tunnel section is training data of 52 typical tunnel sections.

7. A method of classifying the impermeability of the surrounding rock of a subsea tunnel, characterized by The sea tunnel surrounding rock impermeability classification method comprises: Obtaining surrounding rock parameters and corresponding engineering data of a surrounding rock to be classified; Inputting the surrounding rock parameters and corresponding engineering data into a decision tree to obtain classification of the surrounding rock to be classified; the decision tree is determined by the decision tree determination method in any one of claims 1-6.

8. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the decision tree determination method in any one of claims 1-6 or the sea tunnel surrounding rock impermeability classification method in claim 7.

9. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the decision tree determination method in any one of claims 1-6 or the sea tunnel surrounding rock impermeability classification method in claim 7.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the decision tree determination method in any one of claims 1-6 or the sea tunnel surrounding rock impermeability classification method in claim 7.

Images