A tunnel surrounding rock quality grade identification method and device based on an entropy weight method, equipment and a storage medium

Abstract

The application discloses a tunnel surrounding rock quality grade identification method and device based on an entropy weight method, equipment and a storage medium, and relates to the technical field of geological surveying.The method is to first divide all identified sample data into several parts, then apply multiple machine learning algorithms to modeling training for each part to obtain model classification accuracies corresponding to each part and each algorithm, then obtain classification model weight coefficients of each algorithm based on the entropy weight method, apply the multiple machine learning algorithms to modeling training for all identified sample data to obtain tunnel surrounding rock quality grade classification models corresponding to each algorithm, then apply the classification model weight coefficients of each algorithm and probability values obtained based on the classification models and classified in all tunnel surrounding rock quality grades to calculate comprehensive probability values classified in all tunnel surrounding rock quality grades, and finally take the grade corresponding to the maximum comprehensive probability value as a final identification result.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of geological survey, and particularly relates to a tunnel surrounding rock quality grade identification method, device, equipment and storage medium based on an entropy weight method. BACKGROUND

[0002] With the continuous development of new urbanization construction in China, the development of transportation infrastructure has entered a golden period, the construction demand is continuously booming, and the transportation construction project is being promoted to the central and western regions, and a large number of tunnel projects are being or will be built in the mountainous areas of the western region of China. The topography and geological conditions of the mountainous areas in the western region of China are complex, the geological structure is active, the engineering depth is large, and the underground engineering is subjected to the coupling action of high water pressure and high ground stress, which brings great challenges to the survey, design, construction and operation of the tunnel. In the survey, design and construction stages of the tunnel, the most important work is undoubtedly to classify the quality of the surrounding rock of the tunnel, and the quality grade of the surrounding rock of the tunnel is one of the main bases for dynamic design and construction of the tunnel.

[0003] At present, in the tunnel project, the quality of the surrounding rock is mainly classified according to the BQ (Basic Quality) method of the “Engineering Rock Mass Classification Standard”, but the BQ method needs many quantitative test experiments, and it is relatively cumbersome to use, so in actual work, the technical personnel on site more often use the experience method to classify the quality of the surrounding rock, and the classification result is often too large or too small, which is not conducive to the dynamic support and excavation design of the tunnel. SUMMARY

[0004] The purpose of the present application is to provide a tunnel surrounding rock quality grade identification method, device, computer equipment and computer readable storage medium based on an entropy weight method, to solve the problem that the BQ method is relatively cumbersome to use and the experience method has a large classification result error in the existing tunnel surrounding rock quality grade identification technology.

[0005] In order to achieve the above purpose, the present application adopts the following technical scheme:

[0006] In a first aspect, a tunnel surrounding rock quality grade identification method based on an entropy weight method is provided, comprising:

[0007] All identified sample data are obtained, wherein the identified sample data contain an identified tunnel surrounding rock quality grade of a tunnel surrounding rock object and K index values collected on K evaluation indexes of the tunnel surrounding rock object, the K index values correspond to the K evaluation indexes one by one, and K represents a positive integer greater than or equal to 5;

[0008] The all identified sample data are randomly and equally divided into M identified sample data, wherein M represents a positive integer greater than or equal to 5;

[0009] For each of the M pieces of identified sample data, all the K indicator values corresponding thereto are taken as input items, and all the identified tunnel surrounding rock quality grades corresponding thereto are taken as output items, an artificial intelligence model based on N machine learning algorithms is verified and modeled, and N model classification accuracies corresponding thereto and corresponding to the N machine learning algorithms are obtained, wherein N represents a positive integer greater than or equal to 5;

[0010] According to the N model classification accuracies of the respective identified sample data, a classification model weight coefficient of the N machine learning algorithms is obtained based on an entropy weight method;

[0011] For all the identified sample data, all the K indicator values corresponding thereto are taken as input items, and all the identified tunnel surrounding rock quality grades corresponding thereto are taken as output items, the artificial intelligence model based on the N machine learning algorithms is again verified and modeled, and N tunnel surrounding rock quality grade classification models corresponding thereto and corresponding to the N machine learning algorithms are obtained;

[0012] K new indicator values of a target tunnel surrounding rock object collected on the K evaluation indicators are obtained, wherein the K new indicator values correspond one-to-one to the K evaluation indicators;

[0013] For each of the N tunnel surrounding rock quality grade classification models, the K new indicator values are input into the corresponding model, and a probability value corresponding thereto and classifying a tunnel surrounding rock quality grade of the target tunnel surrounding rock object on all tunnel surrounding rock quality grades is output;

[0014] According to the classification model weight coefficient of the N machine learning algorithms and the probability value of the N tunnel surrounding rock quality grade classification models and classifying a tunnel surrounding rock quality grade of the target tunnel surrounding rock object on all tunnel surrounding rock quality grades, a comprehensive probability value of classifying a tunnel surrounding rock quality grade of the target tunnel surrounding rock object on all tunnel surrounding rock quality grades is calculated according to the following formula:

[0015]

[0016] In the formula, S represents the total number of the all tunnel surrounding rock quality grades, s represents a positive integer less than or equal to S, P s represents a comprehensive probability value of classifying a tunnel surrounding rock quality grade of the target tunnel surrounding rock object on the s-th tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades, n represents a positive integer less than or equal to N, A s,nw s,n represents a probability value of the tunnel surrounding rock quality grade classification of the target tunnel surrounding rock object on the s-th tunnel surrounding rock quality grade in the N-th tunnel surrounding rock quality grade classification model, and w n w n represents a classification model weight coefficient of the n-th machine learning algorithm in the N kinds of machine learning algorithms, and the n-th machine learning algorithm has a one-to-one correspondence with the n-th tunnel surrounding rock quality grade classification model;

[0017] The tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object.

[0018] Based on the above invention content, a tunnel surrounding rock quality grade automatic identification scheme based on entropy weight method and multiple machine learning algorithms is provided, that is, on the one hand, all the identified sample data are divided into several parts, and then multiple machine learning algorithms are applied for modeling training for each part to obtain the model classification accuracy corresponding to each part and each algorithm, and then the classification model weight coefficients of each algorithm are obtained based on the entropy weight method, on the other hand, the multiple machine learning algorithms are also applied for modeling training for the all identified sample data to obtain the tunnel surrounding rock quality grade classification model corresponding to each algorithm, then the classification model weight coefficients of each algorithm and the probability values obtained based on the classification model and classified on all tunnel surrounding rock quality grades are applied to calculate the comprehensive probability values classified on all tunnel surrounding rock quality grades, and finally the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is taken as the final identification result, so that the cumbersome problem of BQ method is avoided, the classification result error is effectively reduced compared with the experience method, and the dynamic support and excavation design of the tunnel are facilitated, which is convenient for practical application and popularization.

[0019] In one possible design, the classification model weight coefficients of the N kinds of machine learning algorithms are obtained based on the entropy weight method according to the N model classification accuracies of the identified sample data in each part, including:

[0020] For each kind of machine learning algorithm in the N kinds of machine learning algorithms, the corresponding entropy value is calculated according to the following formula:

[0021]

[0022] In the formula, n represents a positive integer, S n ln represents a function of taking natural logarithm, m and m' represent positive integers respectively, and η mn w m,n represents the n-th model classification accuracy of the m-th identified sample data in the M identified sample data, and η m′na (n th) model classification accuracy of an m'th identified sample data in the M identified sample data;

[0023] For the various machine learning algorithms, the corresponding classification model weight coefficients are calculated according to the following formula:

[0024]

[0025] wherein, w n denotes the classification model weight coefficient of then'th machine learning algorithm, n' denotes a positive integer, S n′ denotes the entropy value of then'th machine learning algorithm in the N machine learning algorithms.

[0026] In one possible design, the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object, including:

[0027] determining the number of tunnel surrounding rock quality grades corresponding to the maximum comprehensive probability value;

[0028] judging whether the number is greater than or equal to 2, if yes, determining the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value and being the highest grade as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object, otherwise, determining the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object.

[0029] In one possible design, the K evaluation indexes include any one or any combination of a rock hardness strength index, a rock mass weathering degree index, a rock mass integrity dimension index, a rock mass structure dimension index, a structure surface combination degree index, an underground water condition dimension index and a ground stress condition dimension index.

[0030] In one possible design, the rock hardness strength index includes a rock saturated uniaxial compressive strength segmentation result or a classification result in hard rock, relatively hard rock, relatively soft rock, soft rock and extremely soft rock;

[0031] The rock mass weathering degree index includes a rock wave velocity ratio segmentation result or a classification result in unweathered state, slight weathering state, moderate weathering state, strong weathering state and fully weathered state;

[0032] The rock mass integrity dimension index includes a rock mass integrity coefficient segmentation result or a classification result in complete state, relatively complete state, relatively broken state, broken state and extremely broken state;

[0033] The rock mass structure dimension index includes classification results in massive structure, layered structure, mosaic structure, cataclastic structure and dispersed structure, or in their subcategories;

[0034] The structure surface combination degree index includes classification results in tight state, relatively tight state, relatively relaxed state and relaxed state;

[0035] The groundwater condition dimension index includes classification results in multiple types of tunnel water inrush;

[0036] The in-situ stress condition dimension index includes segmented results of the ratio of the saturated uniaxial compressive strength of the rock to the maximum initial stress perpendicular to the tunnel axis, or classification results in the state of extremely high level, high level, medium level and low level.

[0037] In one possible design, the N machine learning algorithms include machine learning algorithms based on decision tree, logistic regression, Naive Bayes method, random forest, support vector machine, K-nearest neighbor method, stochastic gradient descent method, multivariate linear regression, multilayer perception, back propagation neural network and / or radial basis function network.

[0038] In a second aspect, an entropy weight method-based tunnel surrounding rock quality grade identification device is provided, which includes a first data acquisition module, a data random equal division module, a first modeling training module, a weight coefficient calculation module, a second modeling training module, a second data acquisition module, an initial probability calculation module, a comprehensive probability calculation module and a surrounding rock grade determination module.

[0039] The first data acquisition module is configured to acquire all identified sample data, wherein the identified sample data includes an identified tunnel surrounding rock quality grade of a tunnel surrounding rock object and K index values collected from the tunnel surrounding rock object on K evaluation indexes, the K index values correspond to the K evaluation indexes one by one, and K represents a positive integer greater than or equal to 5.

[0040] The data random equal division module is in communication connection with the first data acquisition module and is configured to randomly and equally divide the all identified sample data into M portions of identified sample data, wherein M represents a positive integer greater than or equal to 5.

[0041] The first modeling training module is in communication connection with the data random equal division module and is configured to, for each portion of the M portions of identified sample data, take all the corresponding K index values as input items and take all the corresponding identified tunnel surrounding rock quality grades as output items, calibrate and verify modeling of an artificial intelligence model based on N machine learning algorithms, and obtain N model classification accuracies corresponding to the N machine learning algorithms one by one, wherein N represents a positive integer greater than or equal to 5.

[0042] The weight coefficient calculation module is in communication connection with the first modeling training module, and is configured to obtain the classification model weight coefficients of the N machine learning algorithms based on an entropy weight method according to the N model classification accuracies of the respective identified sample data;

[0043] The second modeling training module is in communication connection with the first data acquisition module, and is configured to, for the all identified sample data, take all the K indicator values corresponding thereto as input items, take all the identified tunnel surrounding rock quality grades corresponding thereto as output items, and perform rate verification modeling again on the artificial intelligence model based on the N machine learning algorithms to obtain N tunnel surrounding rock quality grade classification models corresponding to and one-to-one corresponding to the N machine learning algorithms.

[0044] The second data acquisition module is configured to acquire K new indicator values of a target tunnel surrounding rock object collected on the K evaluation indicators, wherein the K new indicator values correspond one-to-one to the K evaluation indicators.

[0045] The initial probability calculation module is in communication connection with the second modeling training module and the second data acquisition module respectively, and is configured to, for each tunnel surrounding rock quality grade classification model in the N tunnel surrounding rock quality grade classification models, input the K new indicator values into the corresponding model to output a probability value corresponding to classification of a tunnel surrounding rock quality grade of the target tunnel surrounding rock object among all the tunnel surrounding rock quality grades.

[0046] The comprehensive probability calculation module is in communication connection with the weight coefficient calculation module and the initial probability calculation module respectively, and is configured to, according to the classification model weight coefficients of the N machine learning algorithms and the probability values of classification of a tunnel surrounding rock quality grade of the target tunnel surrounding rock object among all the tunnel surrounding rock quality grades of the N tunnel surrounding rock quality grade classification models, calculate a comprehensive probability value of classification of a tunnel surrounding rock quality grade of the target tunnel surrounding rock object among all the tunnel surrounding rock quality grades according to the following formula:

[0047]

[0048] In the formula, S represents the total number of all the tunnel surrounding rock quality grades, s represents a positive integer less than or equal to S, P s represents a comprehensive probability value of classification of a tunnel surrounding rock quality grade of the target tunnel surrounding rock object among all the tunnel surrounding rock quality grades, n represents a positive integer less than or equal to N, A s,nw represents a probability value of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object on the s-th tunnel surrounding rock quality level in the N tunnel surrounding rock quality level classification models, and s represents a positive integer; n w represents a classification model weight coefficient of the n-th machine learning algorithm in the N machine learning algorithms, and the n-th machine learning algorithm has a one-to-one correspondence with the n-th tunnel surrounding rock quality level classification model;

[0049] The surrounding rock level determination module is in communication connection with the comprehensive probability calculation module, and is configured to determine the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value as the tunnel surrounding rock quality level of the target tunnel surrounding rock object.

[0050] In one possible design, the weight coefficient calculation module includes a first calculation sub-module and a second calculation sub-module in communication connection;

[0051] The first calculation sub-module is configured to calculate, for each machine learning algorithm in the N machine learning algorithms, a corresponding entropy value according to the following formula:

[0052]

[0053] In the formula, n represents a positive integer, S n ln represents a natural logarithm function, m and m' represent positive integers, and η mn η represents the n-th model classification accuracy of the m-th identified sample data in the M identified sample data, and m represents a positive integer; m′n η represents the n-th model classification accuracy of the m'-th identified sample data in the M identified sample data;

[0054] The second calculation sub-module is configured to calculate, for each machine learning algorithm, a corresponding classification model weight coefficient according to the following formula:

[0055]

[0056] In the formula, w n η represents the classification model weight coefficient of the n-th machine learning algorithm, n' represents a positive integer, and S n′ η represents the entropy value of the n'-th machine learning algorithm in the N machine learning algorithms.

[0057] In a third aspect, the present application provides a computer device comprising a memory, a processor and a transceiver connected in sequence, wherein the memory is configured to store a computer program, the transceiver is configured to transmit and receive data, and the processor is configured to read the computer program and execute the tunnel surrounding rock quality grade identification method according to any possible design of the first aspect.

[0058] In a fourth aspect, the present application provides a computer readable storage medium having instructions stored thereon, wherein the instructions, when executed on a computer, perform the tunnel surrounding rock quality grade identification method according to any possible design of the first aspect.

[0059] In a fifth aspect, the present application provides a computer program product comprising instructions, wherein the instructions, when executed on a computer, cause the computer to perform the tunnel surrounding rock quality grade identification method according to any possible design of the first aspect.

[0060] The above-mentioned scheme has the following beneficial effects:

[0061] (1) The present application creatively provides a tunnel surrounding rock quality grade automatic identification scheme based on entropy weight method and multiple machine learning algorithms, that is, on the one hand, all identified sample data are divided into several parts, and then multiple machine learning algorithms are applied to modeling training for each part to obtain the model classification accuracy corresponding to each part and each algorithm, and then the classification model weight coefficients of each algorithm are obtained based on the entropy weight method; on the other hand, the multiple machine learning algorithms are also applied to modeling training for the all identified sample data to obtain the tunnel surrounding rock quality grade classification model corresponding to each algorithm, and then the classification model weight coefficients of each algorithm and the probability value obtained based on the classification model and classified in all tunnel surrounding rock quality grades are applied to calculate the comprehensive probability value classified in all tunnel surrounding rock quality grades, and finally the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is taken as the final identification result, so that the cumbersome problem of BQ method is avoided, the classification result error is effectively reduced compared with the experience method, and the dynamic support and excavation design of the tunnel are facilitated, which is convenient for practical application and popularization. BRIEF DESCRIPTION OF DRAWINGS

[0062] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0063] Figure 1A flowchart of a tunnel surrounding rock quality grade identification method based on an entropy weight method provided by an embodiment of the present application.

[0064] Figure 2 A calculation flowchart example of a classification model weight coefficient based on an entropy weight method provided by an embodiment of the present application.

[0065] Figure 3 A structural schematic diagram of a tunnel surrounding rock quality grade identification device based on an entropy weight method provided by an embodiment of the present application.

[0066] Figure 4 A structural schematic diagram of a computer device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the present application will be briefly introduced below with reference to the drawings and the description of the embodiments or the prior art. Obviously, the following description of the drawings is only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings. It should be noted that the description of these embodiments is used to help understand the present application, but does not constitute a limitation on the present application.

[0068] It should be understood that although the terms first and second, etc. may be used herein to describe various objects, these objects should not be limited by these terms. These terms are only used to distinguish one object from another object. For example, a first object can be called a second object, and similarly a second object can be called a first object, without departing from the scope of the example embodiments of the present application.

[0069] It should be understood that for the term "and / or" that may appear in the present text, it is only a description of the association relationship of the associated objects, which means that there can be three kinds of relationships, for example, A and / or B, which can mean that there are three cases of A alone, B alone, or A and B together; for example, A, B and / or C, which means that there are any one of A, B and C or any combination thereof; for the term " / and" that may appear in the present text, it is another description of the relationship of another associated object, which means that there can be two kinds of relationships, for example, A / and B, which means that there are two cases of A alone or A and B together; in addition, for the character " / " that may appear in the present text, it generally means that the associated objects before and after are an "or" relationship.

[0070] Embodiment:

[0071] As Figure 1As shown, the tunnel surrounding rock quality grade identification method provided in the first aspect of the present embodiment and based on the entropy weight method can be, but is not limited to, executed by a computer device with certain computing resources, for example, by an electronic device such as a platform server, a personal computer (PC, which refers to a multi-purpose computer suitable for personal use in size, price and performance; desktop computers, notebook computers to small notebook computers and tablet computers, and ultrabooks, etc.), a smart phone, a personal digital assistant (PDA), or a wearable device, etc. As shown in the figure, Figure 1 As shown, the tunnel surrounding rock quality grade identification method can include, but is not limited to, the following steps S1-S9.

[0072] S1. Obtain all identified sample data, wherein the identified sample data includes, but is not limited to, an identified tunnel surrounding rock quality grade of a tunnel surrounding rock object and K index values collected on K evaluation indexes for the tunnel surrounding rock object, the K index values correspond one-to-one to the K evaluation indexes, and K represents a positive integer greater than or equal to 5.

[0073] In the step S1, the identified tunnel surrounding rock quality grade is a historical identification result of the tunnel surrounding rock object (the identification method used is preferably, but not limited to, the traditional BQ method), and the K evaluation indexes are an index system used for participating in the identification of the tunnel surrounding rock quality grade. Through summary and analysis of previous research, the inventors have known that the main factors affecting the tunnel surrounding rock quality are rock strength, rock mass integrity, ground stress and underground water, and based on previous work, it is found that the factors that have a direct or indirect impact on the evaluation of the tunnel surrounding rock quality are as follows: rock hardness strength, rock weathering degree, rock mass integrity, rock mass structure, structure surface combination degree, underground water condition and ground stress condition, etc. (the aforementioned terms are common terms in existing geological exploration technology, and will not be described here), therefore, specifically, the K evaluation indexes include, but are not limited to, any one or any combination of the rock hardness strength index, the rock weathering degree index, the rock mass integrity dimension index, the rock mass structure dimension index, the structure surface combination degree index, the underground water condition dimension index and the ground stress condition dimension index, etc. As shown in the figure, Figure 2 As shown, the K evaluation indexes preferably include the rock hardness strength index, the rock weathering degree index, the rock mass integrity dimension index, the rock mass structure dimension index, the structure surface combination degree index, the underground water condition dimension index and the ground stress condition dimension index (i.e. K = 7).

[0074] In the step S1, since the subsequent machine learning algorithm is essentially a mathematical probability method, its input parameters must all be numerical, while in the K evaluation indexes, part of them are qualitative indexes (such as rock mass weathering degree index, rock mass structure dimension index and groundwater condition dimension index), whose corresponding index values are discrete quantities, and the other part of the index values are continuous quantities, so in order to use the continuous variables in the machine learning algorithm, it is necessary to discretize them or assign them a density function (Jensen & Nielsen, 2007). At the same time, since it is difficult to accurately assign a density function to the input parameters under the condition of limited samples, a discretization method (i.e. dividing continuous quantities into multiple ranges) can be used to process continuous variables. In this embodiment, the inventors specifically use the parameter range recommended by the industry guideline "Engineering Rock Mass Classification Standard" to obtain the index values of the K evaluation indexes, i.e. in detail, the rock hardness strength index includes but is not limited to the segmented results of the saturated uniaxial compressive strength of the rock or the classification results in hard rock, relatively hard rock, relatively soft rock, soft rock and extremely soft rock; the rock mass weathering degree index includes but is not limited to the segmented results of the wave velocity ratio of the rock or the classification results in unweathered state, slight weathering state, moderate weathering state, strong weathering state and fully weathered state; the rock mass integrity dimension index includes but is not limited to the segmented results of the rock mass integrity coefficient or the classification results in complete state, relatively complete state, relatively broken state, broken state and extremely broken state; the rock mass structure dimension index includes but is not limited to the classification results in massive structure, stratified structure, mosaic structure, cataclastic structure and dispersed structure or in their subcategories; the structure surface bonding degree index includes but is not limited to the classification results in tight state, relatively tight state, relatively loose state and loose state; the groundwater condition dimension index includes but is not limited to the classification results in various types of tunnel water; and the in-situ stress condition dimension index includes but is not limited to the segmented results of the ratio of the saturated uniaxial compressive strength of the rock to the maximum initial stress perpendicular to the tunnel axis or the classification results in extremely high level state, high level state, medium level state and low level state.

[0075] In the step S1, the interval ranges and possible states of the above-mentioned various evaluation indexes can be specifically shown in Table 1 as follows:

[0076] Table 1. Interval ranges and possible states of various evaluation indexes

[0077]

[0078] In the above Table 1, σ max represents the maximum initial stress perpendicular to the tunnel axis. In addition, the specific collection method of the K index values is the prior art, and the required identified tunnel surrounding rock quality grade can have but is not limited to five levels such as Grade I, Grade II, Grade III, Grade IV and Grade V.

[0079] S2. Randomly divide all the identified sample data into M equal parts, where M represents a positive integer greater than or equal to 5.

[0080] In step S2, for example, Figure 2 As shown, all the identified sample data were randomly divided into 10 identified sample data: sample 1 to 10, i.e., M = 10.

[0081] S3. For each of the M identified sample data, take all the corresponding K index values ​​as input items and all the corresponding identified tunnel surrounding rock quality grades as output items, and perform calibration and verification modeling on the artificial intelligence model based on N machine learning algorithms to obtain the corresponding N model classification accuracy rates that correspond one-to-one with the N machine learning algorithms, where N represents a positive integer greater than or equal to 5.

[0082] In step S3, the machine learning algorithm is a core algorithm in current artificial intelligence. It can combine known classification experience sample data to obtain a classification model through calibration and verification modeling. Specifically, the N machine learning algorithms include, but are not limited to, those based on decision trees, logistic regression, Naive Bayes, random forests, support vector machines, K-nearest neighbors, stochastic gradient descent, multivariate linear regression, multilayer perceptrons, backpropagation neural networks, and / or radial basis function networks. Decision trees, logistic regression, Naive Bayes, random forests, support vector machines, K-nearest neighbors, stochastic gradient descent, multivariate linear regression, multilayer perceptrons, backpropagation neural networks, and / or radial basis function networks are all common solutions in existing artificial intelligence methods. The specific process of calibration and verification modeling includes a model calibration process and a verification process. That is, first, the model simulation results are compared with the measured data, and then the model parameters are adjusted according to the comparison results to make the simulation results match the actual results. Therefore, a tunnel surrounding rock quality grade classification model and model classification accuracy can be obtained through conventional calibration and verification modeling methods. For example, Figure 2 As shown, the preferred N machine learning algorithms include decision trees, logistic regression, Naive Bayes, random forests, and support vector machines (i.e., N=5), which can yield 50 classification models for tunnel surrounding rock quality levels and the model classification accuracy (i.e., Figure 2 The 10x5 matrix shown, such as S 10,5 This refers to the classification accuracy of the tunnel surrounding rock quality grade classification model trained based on the 5th machine learning algorithm among the N machine learning algorithms and the 10th identified sample data among the M identified sample data.

[0083] S4. Obtain the classification model weight coefficients of the N machine learning algorithms based on the entropy weight method according to the N model classification accuracies of the identified sample data.

[0084] In the step S4, the entropy weight method is an existing means that can calculate the weights of each index by using the tool of information entropy, so as to provide a basis for multi-index comprehensive evaluation. Specifically, the classification model weight coefficients of the N machine learning algorithms are obtained based on the entropy weight method according to the N model classification accuracies of the identified sample data, including but not limited to the following steps S41-S42.

[0085] S41. For each machine learning algorithm in the N machine learning algorithms, the corresponding entropy value is calculated according to the following formula:

[0086]

[0087] In the formula, n represents a positive integer, S n represents the entropy value of the nth machine learning algorithm in the N machine learning algorithms, ln() represents a natural logarithm function, m and m' respectively represent positive integers, and η mn represents the nth model classification accuracy of the mth identified sample data in the M identified sample data, η m′n represents the nth model classification accuracy of the m'th identified sample data in the M identified sample data.

[0088] In the step S41, the concept of entropy (Greek: entropia, English: entropy) was proposed by the German physicist Clausius in 1865, which means "internal change of a system" in Greek origin, that is, "change of internal properties of a system"; in 1948, Shannon extended the concept of entropy in statistical physics to the process of channel communication, that is, Shannon defined "entropy" as "Shannon entropy" or "information entropy", that is:

[0089]

[0090] In the formula, s(p1, p2, …, p I ) represents information entropy, i marks all possible samples in the probability space, p i represents the probability of occurrence of the sample, I represents the total number of samples, and k represents an arbitrary constant related to unit selection, so the calculation formula of the entropy value can be obtained based on the formula.

[0091] S42. For each machine learning algorithm, the corresponding classification model weight coefficient is calculated according to the following formula:

[0092]

[0093] wherein, w n denotes the classification model weight coefficient of the nth machine learning algorithm, n' denotes a positive integer, S n′ denotes the entropy value of the nth machine learning algorithm in the N machine learning algorithms.

[0094] In the step S42, since the entropy value reflects the discrete degree of the N model classification accuracy on the corresponding algorithm, the classification model weight coefficient of the corresponding algorithm can be calculated and determined based on the entropy weight method.

[0095] S5. For all the identified sample data, the corresponding all K indicator values are taken as input items, and the corresponding all identified tunnel surrounding rock quality grades are taken as output items, and the artificial intelligence model based on the N machine learning algorithms is re-verified and modeled, and N tunnel surrounding rock quality grade classification models corresponding to the N machine learning algorithms are obtained.

[0096] In the step S5, specific details can be referred to the foregoing step S3, which will not be described here.

[0097] S6. K new values of the K evaluation indicators collected on the target tunnel surrounding rock object are obtained, wherein the K new values of the K evaluation indicators correspond to the K evaluation indicators one by one.

[0098] In the step S6, the target tunnel surrounding rock object can be but is not limited to the tunnel surrounding rock which needs to dynamically determine the quality grade in the dynamic support and excavation process of the tunnel. In addition, the specific collection method of the K new values is also the prior art.

[0099] S7. For each tunnel surrounding rock quality grade classification model in the N tunnel surrounding rock quality grade classification models, the K new values of the indicators are input into the corresponding model, and the probability value of the tunnel surrounding rock quality grade of the target tunnel surrounding rock object classified in all tunnel surrounding rock quality grades is output.

[0100] In the step S7, all tunnel surrounding rock quality grades can be but are not limited to five levels of Grade I, Grade II, Grade III, Grade IV and Grade V. For example, the probability value of the tunnel surrounding rock quality grade of the target tunnel surrounding rock object classified in the five levels in each tunnel surrounding rock quality grade classification model of the five tunnel surrounding rock quality grade classification models is shown in Table 2 as follows:

[0101] Table 2. Probability values of five tunnel surrounding rock quality grade classification models and five levels

[0102] Class I Class II Class III Class IV Class V Tunnel surrounding rock quality grade classification model 1 corresponding to the decision tree 0.10 0.20 0.30 0.20 0.20 Tunnel surrounding rock quality grade classification model 2 corresponding to the logic regression 0.15 0.15 0.20 0.30 0.20 Tunnel surrounding rock quality grade classification model 3 corresponding to the naive Bayes method 0.05 0.30 0.15 0.40 0.10 Tunnel surrounding rock quality grade classification model 4 corresponding to the random forest 0.00 0.40 0.30 0.20 0.10 Tunnel surrounding rock quality grade classification model 5 corresponding to the support vector machine 0.02 0.05 0.23 0.40 0.30

[0103] S8. The comprehensive probability value of classifying the tunnel surrounding rock quality grade of the target tunnel surrounding rock object in all tunnel surrounding rock quality grades is calculated according to the following formula:

[0104]

[0105] wherein S represents the total number of the all tunnel surrounding rock quality grades, s represents a positive integer less than or equal to S, P s represents the comprehensive probability value of classifying the tunnel surrounding rock quality grade of the target tunnel surrounding rock object in the s-th tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades, n represents a positive integer less than or equal to N, A s,n represents the probability value of classifying the tunnel surrounding rock quality grade of the target tunnel surrounding rock object in the s-th tunnel surrounding rock quality grade in the n-th tunnel surrounding rock quality grade classification model, w n represents the classification model weight coefficient of the n-th machine learning algorithm, and the n-th machine learning algorithm has a one-to-one correspondence with the n-th tunnel surrounding rock quality grade classification model.

[0106] In the step S8, for example, the final obtained comprehensive probability values of classifying the tunnel surrounding rock quality grade of the target tunnel surrounding rock object in the five levels are shown in Table 3 as follows:

[0107] Table 3. Comprehensive probability values in the five levels

[0108] Class I 0.201493 Class II 0.207334 Class III 0.209827 Class IV 0.197399 Class V 0.214312

[0109] S9. The tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object.

[0110] In the step S9, for example, according to the above Table 3, since the maximum comprehensive probability value is 0.214312, the V level can be determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object. In addition, when there are two or more maximum comprehensive probability values, in order to follow the safety conservation principle in tunnel design, the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value is preferably determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object, including but not limited to: first determining the number of tunnel surrounding rock quality levels corresponding to the maximum comprehensive probability value; then judging whether the number is greater than or equal to 2, if yes, the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value and being the highest level is determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object, otherwise the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object.

[0111] Thus, based on the tunnel surrounding rock quality level identification method described in the foregoing steps S1-S9, an automatic tunnel surrounding rock quality level identification scheme based on the entropy weight method and multiple machine learning algorithms is provided, that is, on the one hand, all the identified sample data are first divided into several parts, then multiple machine learning algorithms are applied for modeling training for each part, to obtain the model classification accuracy corresponding to each part and each algorithm, and then the classification model weight coefficients of each algorithm are obtained based on the entropy weight method, on the other hand, the multiple machine learning algorithms are also applied for modeling training for the all identified sample data, to obtain the tunnel surrounding rock quality level classification model corresponding to each algorithm, then the classification model weight coefficients of the algorithms and the probability values obtained based on the classification model and classified in all tunnel surrounding rock quality levels are applied, to calculate the comprehensive probability values classified in all tunnel surrounding rock quality levels, and finally the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value is taken as the final identification result, so as to not only avoid the cumbersome problem of using the BQ method, but also effectively reduce the classification result error relative to the experience method, and thus can be beneficial to the dynamic support and excavation design of the tunnel, and is convenient for practical application and promotion.

[0112] As shown in Figure 3 the second aspect of the present embodiment provides a virtual device for implementing the tunnel surrounding rock quality level identification method of the first aspect, comprising a first data acquisition module, a data random equal division module, a first modeling training module, a weight coefficient calculation module, a second modeling training module, a second data acquisition module, an initial probability calculation module, a comprehensive probability calculation module and a surrounding rock level determination module.

[0113] The first data acquisition module is configured to acquire all identified sample data, wherein the identified sample data comprises an identified tunnel surrounding rock quality grade of a tunnel surrounding rock object and K index values collected on K evaluation indexes for the tunnel surrounding rock object, the K index values correspond to the K evaluation indexes one by one, and K represents a positive integer greater than or equal to 5;

[0114] The data random division module is communicatively connected to the first data acquisition module and is configured to randomly divide the all identified sample data into M portions of identified sample data, wherein M represents a positive integer greater than or equal to 5;

[0115] The first modeling training module is communicatively connected to the data random division module and is configured to, for each portion of identified sample data in the M portions of identified sample data, take all corresponding K index values as input items and all corresponding identified tunnel surrounding rock quality grades as output items, perform rate verification modeling on an artificial intelligence model based on N machine learning algorithms, and obtain N model classification accuracies corresponding to the N machine learning algorithms one by one, wherein N represents a positive integer greater than or equal to 5;

[0116] The weight coefficient calculation module is communicatively connected to the first modeling training module and is configured to obtain classification model weight coefficients of the N machine learning algorithms based on an entropy weight method according to the N model classification accuracies of the each portion of identified sample data;

[0117] The second modeling training module is communicatively connected to the first data acquisition module and is configured to, for the all identified sample data, take all corresponding K index values as input items and all corresponding identified tunnel surrounding rock quality grades as output items, perform rate verification modeling again on an artificial intelligence model based on the N machine learning algorithms, and obtain N tunnel surrounding rock quality grade classification models corresponding to the N machine learning algorithms one by one;

[0118] The second data acquisition module is configured to acquire K new index values on the K evaluation indexes for a target tunnel surrounding rock object, wherein the K new index values correspond to the K evaluation indexes one by one;

[0119] The initial probability calculation module is respectively communicatively connected to the second modeling training module and the second data acquisition module and is configured to, for each tunnel surrounding rock quality grade classification model in the N tunnel surrounding rock quality grade classification models, input the K new index values into the corresponding model and output to obtain a probability value corresponding to a tunnel surrounding rock quality grade of the target tunnel surrounding rock object among all tunnel surrounding rock quality grades;

[0120] The comprehensive probability calculation module is respectively communicatively connected with the weight coefficient calculation module and the initial probability calculation module, and is configured to calculate a comprehensive probability value of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object in all tunnel surrounding rock quality levels according to the classification model weight coefficients of the N machine learning algorithms and the probability values of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object in all tunnel surrounding rock quality levels, and the comprehensive probability value of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object in all tunnel surrounding rock quality levels is calculated according to the following formula:

[0121]

[0122] In the formula, S represents the total number of the all tunnel surrounding rock quality levels, s represents a positive integer less than or equal to S, P s represents the comprehensive probability value of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object in the s-th tunnel surrounding rock quality level in all tunnel surrounding rock quality levels, n represents a positive integer less than or equal to N, A s,n represents the probability value of classifying the tunnel surrounding rock quality level of the target tunnel surrounding rock object in the s-th tunnel surrounding rock quality level in the n-th tunnel surrounding rock quality classification model in the N tunnel surrounding rock quality classification models, w n represents the classification model weight coefficient of the n-th machine learning algorithm in the N machine learning algorithms, and the n-th machine learning algorithm has a one-to-one correspondence with the n-th tunnel surrounding rock quality classification model;

[0123] The surrounding rock level determination module is communicatively connected with the comprehensive probability calculation module, and is configured to determine the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value as the tunnel surrounding rock quality level of the target tunnel surrounding rock object.

[0124] In one possible design, the weight coefficient calculation module includes a first calculation sub-module and a second calculation sub-module which are communicatively connected.

[0125] The first calculation sub-module is configured to calculate a corresponding entropy value according to the following formula for each machine learning algorithm in the N machine learning algorithms:

[0126]

[0127] In the formula, n represents a positive integer, S n represents the entropy value of the n-th machine learning algorithm in the N machine learning algorithms, ln() represents a natural logarithm function, m and m' represent positive integers respectively, and η mn represents the n-th model classification accuracy of the m-th identified sample data in the M identified sample data, and η m′ndenotes the n-th model classification accuracy of the m'-th identified sample data in the M identified sample data;

[0128] The second calculation sub-module is configured to calculate the classification model weight coefficient of each machine learning algorithm according to the following formula:

[0129]

[0130] wherein, w n denotes the classification model weight coefficient of the n-th machine learning algorithm, n' denotes a positive integer, S n′ denotes the entropy value of the n'-th machine learning algorithm in the N machine learning algorithms.

[0131] The working process, working details and technical effects of the foregoing device provided by the second aspect of the embodiment can be referred to the tunnel surrounding rock quality grade identification method described in the first aspect, and will not be repeated here.

[0132] As shown in Figure 4 The third aspect of the embodiment provides a computer device for executing the tunnel surrounding rock quality grade identification method described in the first aspect, which comprises a memory, a processor and a transceiver connected in sequence, wherein the memory is configured to store a computer program, the transceiver is configured to transceive data, and the processor is configured to read the computer program and execute the tunnel surrounding rock quality grade identification method described in the first aspect. Specifically, the memory can include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a flash memory, a first input first output (FIFO) memory and / or a first input last output (FILO) memory, etc.; the processor can be, but is not limited to, a microprocessor with a model number of STM32F105 series. In addition, the computer device can further include, but is not limited to, a power module, a display screen and other necessary components.

[0133] The working process, working details and technical effects of the foregoing computer device provided by the third aspect of the embodiment can be referred to the tunnel surrounding rock quality grade identification method described in the first aspect, and will not be repeated here.

[0134] The fourth aspect of the embodiment provides a computer readable storage medium storing instructions of the tunnel surrounding rock quality grade identification method as described in the first aspect, that is, the computer readable storage medium stores instructions, and when the instructions run on a computer, the tunnel surrounding rock quality grade identification method as described in the first aspect is executed. Wherein, the computer readable storage medium refers to a carrier for storing data, which can include, but is not limited to, floppy disks, optical disks, hard disks, flash memories, USB flash disks and / or memory sticks and other computer readable storage media, and the computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0135] The working process, working details and technical effects of the aforementioned computer readable storage medium provided by the fourth aspect of the embodiment can be referred to the tunnel surrounding rock quality grade identification method as described in the first aspect, which will not be described here.

[0136] The fifth aspect of the embodiment provides a computer program product containing instructions, which, when running on a computer, causes the computer to execute the tunnel surrounding rock quality grade identification method as described in the first aspect. Wherein, the computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices.

[0137] Finally, it should be noted that: the above only describes the preferred embodiments of the present application, and is not used to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for identifying the quality grade of tunnel surrounding rock based on the entropy weight method, characterized in that, include: Acquire all identified sample data, wherein the identified sample data package contains the identified tunnel surrounding rock quality grade of the tunnel surrounding rock object and K index values ​​collected on the tunnel surrounding rock object on K evaluation indicators, wherein the K index values ​​correspond one-to-one with the K evaluation indicators, and K represents a positive integer greater than or equal to 5; All the identified sample data are randomly divided into M equal parts, where M represents a positive integer greater than or equal to 5; For each of the M identified sample data, take all the corresponding K index values ​​as input items and take all the corresponding identified tunnel surrounding rock quality grades as output items to calibrate and verify the artificial intelligence model based on N machine learning algorithms, and obtain the corresponding N model classification accuracy rates that correspond one-to-one with the N machine learning algorithms, where N represents a positive integer greater than or equal to 5. Based on the classification accuracy of N models for each identified sample data, the classification model weight coefficients of the N machine learning algorithms are obtained using the entropy weight method. For all the identified sample data, take all the corresponding K index values ​​as input items and take all the corresponding identified tunnel surrounding rock quality grades as output items, and calibrate and verify the artificial intelligence model based on the N machine learning algorithms again to obtain N tunnel surrounding rock quality grade classification models that correspond one-to-one with the N machine learning algorithms. Obtain new values ​​of K indicators collected from the target tunnel surrounding rock object based on the K evaluation indicators, wherein the new values ​​of the K indicators correspond one-to-one with the K evaluation indicators; For each of the N tunnel surrounding rock quality grade classification models, the new values ​​of the K indicators are input into the corresponding model, and the corresponding probability value that classifies the target tunnel surrounding rock object into all tunnel surrounding rock quality grades is output. Based on the classification model weight coefficients of the N machine learning algorithms and the probability values ​​of the N tunnel surrounding rock quality grade classification models for classifying the target tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades, the comprehensive probability value for classifying the target tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades is calculated according to the following formula: In the formula, S represents the total number of all tunnel surrounding rock quality grades, s represents a positive integer less than or equal to S, and P s This represents the comprehensive probability value for classifying the target tunnel surrounding rock quality object into the s-th tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades, where n represents a positive integer less than or equal to N, and A s,n w represents the probability value of classifying the target tunnel's surrounding rock quality object into the nth tunnel surrounding rock quality grade of the N tunnel surrounding rock quality grade classification models. n This represents the classification model weight coefficient of the nth machine learning algorithm among the N machine learning algorithms, and there is a one-to-one correspondence between the nth machine learning algorithm and the nth tunnel surrounding rock quality grade classification model; The tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object.

2. The method for identifying the quality grade of tunnel surrounding rock as described in claim 1, characterized in that, Based on the classification accuracy of N models for each identified sample data, the classification model weight coefficients of the N machine learning algorithms are obtained using the entropy weight method, including: For each of the N machine learning algorithms, the corresponding entropy value is calculated according to the following formula: In the formula, n represents a positive integer, and S n Let η represent the entropy value of the nth machine learning algorithm among the N machine learning algorithms, ln() represents the function for calculating the natural logarithm, m and m′ represent positive integers, and η represents the entropy value of the nth machine learning algorithm. mn η represents the classification accuracy of the nth model for the mth identified sample among the M identified sample data. m′n This represents the classification accuracy of the nth model for the m′th identified sample data out of the M identified sample data. For each of the aforementioned machine learning algorithms, the corresponding classification model weight coefficients are calculated using the following formula: In the formula, w n S represents the classification model weight coefficient of the nth machine learning algorithm, where n′ represents a positive integer. n′ This represents the entropy value of the n′-th machine learning algorithm among the N machine learning algorithms.

3. The method for identifying the quality grade of tunnel surrounding rock as described in claim 1, characterized in that, The tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object, including: Determine the number of tunnel surrounding rock quality grades corresponding to the maximum comprehensive probability value; If the quantity is greater than or equal to 2, the highest level of tunnel surrounding rock quality corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object; otherwise, the tunnel surrounding rock quality level corresponding to the maximum comprehensive probability value is determined as the tunnel surrounding rock quality level of the target tunnel surrounding rock object.

4. The method for identifying the quality grade of tunnel surrounding rock as described in claim 1, characterized in that, The K evaluation indicators include any one or any combination of the following: rock hardness and strength index, rock weathering degree index, rock integrity dimension index, rock structure dimension index, structural surface bonding degree index, groundwater condition dimension index, and geostress condition dimension index.

5. The method for identifying the quality grade of tunnel surrounding rock as described in claim 4, characterized in that, The rock hardness index includes the saturated uniaxial compressive strength segmentation results of the rock or the classification results in hard rock, relatively hard rock, relatively soft rock, soft rock and extremely soft rock. The rock mass weathering index includes the wave velocity segmentation results of the rock or the classification results in unweathered, slightly weathered, moderately weathered, severely weathered and completely weathered states. The rock mass integrity dimension index includes the segmented results of the rock mass integrity coefficient or the classification results in the intact state, relatively intact state, relatively broken state, broken state and extremely broken state. The rock mass structure dimension index includes classification results in or in their subclasses of massive structure, layered structure, mosaic structure, fractured structure and granular structure. The structural surface bonding degree index includes classification results in tight, relatively tight, relatively loose and loose states. The groundwater condition dimension index includes classification results for various types of water discharge from the tunnel face; The geostress condition dimension index includes segmented results of the ratio of the rock's saturated uniaxial compressive strength to the maximum initial stress perpendicular to the tunnel axis, or classification results in extremely high-level, high-level, medium-level, and low-level conditions.

6. The method for identifying the quality grade of tunnel surrounding rock based on the entropy weight method as described in claim 1, characterized in that, The N types of machine learning algorithms include those based on decision trees, logistic regression, Naive Bayes, random forests, support vector machines, K-nearest neighbors, stochastic gradient descent, multivariate linear regression, multilayer perceptrons, backpropagation neural networks, and / or radial basis function networks.

7. A tunnel surrounding rock quality grade identification device based on the entropy weight method, characterized in that, It includes a first data acquisition module, a data random distribution module, a first modeling and training module, a weight coefficient calculation module, a second modeling and training module, a second data acquisition module, an initial probability calculation module, a comprehensive probability calculation module, and a surrounding rock grade determination module; The first data acquisition module is used to acquire all identified sample data, wherein the identified sample data package contains the identified tunnel surrounding rock quality grade of the tunnel surrounding rock object and K index values ​​collected on the tunnel surrounding rock object on K evaluation indicators, wherein the K index values ​​correspond one-to-one with the K evaluation indicators, and K represents a positive integer greater than or equal to 5; The data random distribution module is communicatively connected to the first data acquisition module and is used to randomly divide all the identified sample data into M parts of identified sample data, where M represents a positive integer greater than or equal to 5. The first modeling and training module is communicatively connected to the data random distribution module. It is used to take all the corresponding K index values ​​as input items and all the corresponding identified tunnel surrounding rock quality grades as output items for each of the M identified sample data. It performs calibration and verification modeling on the artificial intelligence model based on N machine learning algorithms to obtain the corresponding N model classification accuracy rates that correspond one-to-one with the N machine learning algorithms, where N represents a positive integer greater than or equal to 5. The weight coefficient calculation module is communicatively connected to the first modeling and training module, and is used to obtain the classification model weight coefficients of the N machine learning algorithms based on the entropy weight method according to the classification accuracy of the N models of each identified sample data. The second modeling training module is communicatively connected to the first data acquisition module. It is used to take all the corresponding K index values ​​as input items and all the corresponding identified tunnel surrounding rock quality grades as output items for all the identified sample data, and to perform calibration and verification modeling again on the artificial intelligence model based on the N machine learning algorithms to obtain N tunnel surrounding rock quality grade classification models that correspond one-to-one with the N machine learning algorithms. The second data acquisition module is used to acquire new values ​​of K indicators collected from the target tunnel surrounding rock object based on the K evaluation indicators, wherein the new values ​​of the K indicators correspond one-to-one with the K evaluation indicators; The initial probability calculation module is communicatively connected to the second modeling training module and the second data acquisition module, respectively. It is used to input the new values ​​of the K indicators into the corresponding model for each of the N tunnel surrounding rock quality grade classification models, and output the corresponding probability value that classifies the target tunnel surrounding rock object into all tunnel surrounding rock quality grades. The comprehensive probability calculation module is communicatively connected to the weight coefficient calculation module and the initial probability calculation module, respectively. It is used to calculate the comprehensive probability value for classifying the target tunnel surrounding rock quality object into all tunnel surrounding rock quality levels based on the weight coefficients of the classification models of the N machine learning algorithms and the probability values ​​of the N tunnel surrounding rock quality level classification models, according to the following formula: In the formula, S represents the total number of all tunnel surrounding rock quality grades, s represents a positive integer less than or equal to S, and P s This represents the comprehensive probability value for classifying the target tunnel surrounding rock quality object into the s-th tunnel surrounding rock quality grade among all tunnel surrounding rock quality grades, where n represents a positive integer less than or equal to N, and A s,n w represents the probability value of classifying the target tunnel's surrounding rock quality object into the nth tunnel surrounding rock quality grade of the N tunnel surrounding rock quality grade classification models. n This represents the classification model weight coefficient of the nth machine learning algorithm among the N machine learning algorithms, and there is a one-to-one correspondence between the nth machine learning algorithm and the nth tunnel surrounding rock quality grade classification model; The surrounding rock grade determination module is communicatively connected to the comprehensive probability calculation module, and is used to determine the tunnel surrounding rock quality grade corresponding to the maximum comprehensive probability value as the tunnel surrounding rock quality grade of the target tunnel surrounding rock object.

8. The tunnel surrounding rock quality grade identification device as described in claim 7, characterized in that, The weight coefficient calculation module includes a first calculation submodule and a second calculation submodule that are connected in communication. The first calculation submodule is used to calculate the corresponding entropy value for each of the N machine learning algorithms according to the following formula: In the formula, n represents a positive integer, and S n Let η represent the entropy value of the nth machine learning algorithm among the N machine learning algorithms, ln() represents the function for calculating the natural logarithm, m and m′ represent positive integers, and η represents the entropy value of the nth machine learning algorithm. mn η represents the classification accuracy of the nth model for the mth identified sample among the M identified sample data. m′n This represents the classification accuracy of the nth model for the m′th identified sample data out of the M identified sample data. The second calculation submodule is used to calculate the corresponding classification model weight coefficients for each of the various machine learning algorithms according to the following formula: In the formula, w n S represents the classification model weight coefficient of the nth machine learning algorithm, where n′ represents a positive integer. n′ This represents the entropy value of the n′-th machine learning algorithm among the N machine learning algorithms.

9. A computer device, characterized in that, The device includes a memory, a processor, and a transceiver connected in sequence, wherein the memory is used to store a computer program, the transceiver is used to send and receive data, and the processor is used to read the computer program and execute the tunnel surrounding rock quality grade identification method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that... The computer-readable storage medium stores instructions that, when executed on a computer, perform the tunnel surrounding rock quality grade identification method as described in any one of claims 1 to 6.

Images