A tunnel intelligent support decision method and device based on data knowledge joint enhancement

Abstract

The application discloses a tunnel intelligent support decision-making method and device based on data and knowledge joint enhancement. The device comprises a tunnel support data collector, which is used for collecting experience knowledge data and numerical simulation data; an excavation support scheme pre-selection module, which is used for pre-selecting an excavation support scheme suitable for tunnel environment parameters to obtain a pre-selected scheme; an intelligent support decision-making module, which is used for optimizing the pre-selected scheme to obtain an optimized excavation support scheme; a data and knowledge enhancement learning module, which is used for generating an excavation support scheme based on experience knowledge data, fusing the excavation support scheme with the optimized excavation support scheme, and obtaining a tunnel intelligent support decision-making scheme based on data and knowledge joint enhancement; and a support decision-making interactive output module, which is used for interactively outputting the tunnel intelligent support decision-making scheme to a user. The application can realize tunnel intelligent support decision-making based on data and knowledge joint enhancement, and provides technical support for assisting tunnel construction designers to realize intelligent decision-making.

Description

Technical Field

[0001] This invention relates to the field of intelligent tunnel construction technology, and in particular to a method and apparatus for intelligent tunnel support decision-making that combines data and knowledge enhancement. Background Technology

[0002] Intelligent tunnel construction refers to a method of tunnel construction that integrates various technologies such as intelligence, automation, and information technology, utilizing advanced technological means. Through comprehensive perception, ubiquitous interconnection, fusion, processing, proactive learning, and scientific decision-making regarding the "ground-tunnel-machine-information-human" system and its internal and external environment, it significantly improves construction quality and efficiency, reduces construction costs, and creates a safer, more reliable, and more cost-effective next-generation tunnel construction technology system.

[0003] The overall goal of intelligent tunnel construction at this stage revolves around the informatization and digitalization of construction, including the modeling of engineering data and experience. Specific goals include achieving interconnection and feedback of multi-source information such as geology, structure, and equipment; realizing collaborative feedback intelligent decision-making in tunnel construction based on big data fusion and in-depth mining; and digitizing and algorithmizing operational behaviors under data-driven approaches.

[0004] In tunnel construction, excavation and support play a crucial role in the entire project, and effective excavation and support methods are an important way to ensure the overall safety of the tunnel project. Therefore, adopting a reasonable tunnel excavation and support scheme has become the focus of tunnel construction.

[0005] The diverse range of equipment models and manufacturers involved in tunnel construction, particularly drill-and-blast tunneling, creates significant barriers to digital interconnection of equipment. This limits data analysis and fusion, hindering the development of intelligent tunnel construction. During tunnel design and construction, uncertainties regarding surrounding rock properties necessitate dynamic adjustments to tunnel support schemes. Furthermore, as construction progresses, abnormal phenomena such as deformation or even loosening and collapse may occur. Therefore, dynamic feedback of tunnel information is crucial for timely adjustments to support schemes and the implementation of effective measures. All these issues require data integration and timely, reliable support decision-making support. Summary of the Invention

[0006] To address the aforementioned problems, this invention aims to provide a method and apparatus for intelligent tunnel support decision-making that combines data and knowledge enhancement.

[0007] The technical solution of the present invention is as follows:

[0008] On the one hand, a data-knowledge-enhanced intelligent tunnel support decision-making method is provided, including the following steps:

[0009] S1: Collect and acquire empirical knowledge data and numerical simulation data;

[0010] S2: Determine tunnel environmental parameters and excavation support scheme parameters, and construct a tunnel support scheme knowledge table, a multivariate numerical simulation dataset, and an excavation support scheme library based on numerical simulation data;

[0011] S3: Based on the tunnel environment parameters, perform uncertainty reasoning on the excavation and support schemes in the excavation and support scheme library;

[0012] If no excavation and support scheme that fits the tunnel environment parameters is found in the uncertainty reasoning results, proceed to step S4;

[0013] If the uncertainty reasoning results include an excavation and support scheme that is compatible with the tunnel environment parameters, then the compatible excavation and support scheme is selected as the pre-selected scheme, and the process proceeds to step S5.

[0014] S4: Construct a first model for generating excavation and support schemes based on numerical simulation data, and use the multivariate numerical simulation dataset as training parameters to obtain a trained first model for generating excavation and support schemes.

[0015] Based on the trained excavation and support scheme generation model, an excavation and support scheme adapted to the tunnel environment parameters is generated, and this excavation and support scheme is used as a pre-selected scheme to proceed to step S5.

[0016] S5: Construct an intelligent support decision model, and optimize the pre-selected scheme according to the intelligent support decision model to obtain the optimal excavation support scheme;

[0017] S6: Construct a second excavation and support scheme generation model based on experiential knowledge data, and obtain an excavation and support scheme based on experiential knowledge data by combining the tunnel support scheme knowledge table.

[0018] S7: The preferred excavation and support scheme and the excavation and support scheme based on experience knowledge data are integrated to obtain a data knowledge-enhanced intelligent tunnel support decision-making scheme.

[0019] Preferably, in step S4, the excavation and support scheme generation model based on numerical simulation data includes an improved self-attention mechanism Transformer encoder module and a variational inference module; the improved self-attention mechanism Transformer encoder module includes an improved attention layer, a residual connection layer, a layer normalization layer, and a fully connected layer; the variational inference module includes an inference network and a generation network.

[0020] Preferably, the improved formula for calculating the attention layer is:

[0021]

[0022]

[0023]

[0024] In the formula: Q' and K' are the query matrix and key matrix obtained after normalization and matrix transformation; Q and K are the original query matrix and key matrix; μ Q and μ K These are the mean values ​​of Q and K over the time dimension, respectively; the superscript T indicates transpose; σ x d is the standard deviation of the original input; model The dimensions of Model 1 are generated for the excavation and support scheme.

[0025] Preferably, the inference network is used to perform variational inference on the input to generate the probability distribution of the latent variables, as shown in the following formula:

[0026]

[0027] In the formula: q φ (z|x) represents the posterior probability distribution of the latent variables learned by the network given the input x. It follows a Gaussian distribution; μ, σ 2 These represent the mean and variance of the inference network output, respectively.

[0028] The generative network is used to receive latent variables and reconstruct the original input data by mapping them back to the data space, as shown in the following formula:

[0029]

[0030] In the formula: p θ (x'|z) represents the data distribution learned by the generative network given the latent variable z.

[0031] Preferably, in step S5, the intelligent support decision model includes a mechanical evaluation parameter prediction model, a safety factor assessment model, and an evaluation function; the mechanical evaluation parameter prediction model is used to predict the mechanical evaluation parameters corresponding to the pre-selected scheme.

[0032] The safety factor assessment model includes a surrounding rock displacement safety factor K1 assessment model, a support structure safety factor K2 assessment model, a support structure safety factor K3 assessment model, and a support structure safety factor K4 assessment model; wherein...

[0033] The evaluation model for the surrounding rock displacement safety factor K1 is as follows:

[0034]

[0035] In the formula: UL For the ultimate displacement; U p This is the average of the convergent values ​​of the three displacements: arch crown displacement, arch waist displacement, and arch bottom displacement.

[0036] The safety factor K2 evaluation model for the support structure is as follows:

[0037]

[0038] Where: σ L σ represents the ultimate stress of the steel arch frame. p For the stress of the steel arch frame;

[0039] The evaluation model for the safety factor K3 of the support structure is as follows:

[0040]

[0041] In the formula: f c This represents the ultimate compressive strength of the shotcrete. This represents the minimum principal stress convergence value at the crown of the arch;

[0042] The safety factor K4 evaluation model for the support structure is as follows:

[0043]

[0044] In the formula: f h d represents the ultimate compressive strength of concrete; d represents the thickness of the secondary lining in the support scheme; N1-N5 represent the axial force convergence values ​​in the five directions of the tunnel arch crown, arch shoulder, arch waist, arch foot, and arch bottom, respectively.

[0045] The evaluation function is:

[0046]

[0047]

[0048] Where: K i γ represents the magnitude of the i-th safety factor; i For the corresponding safety factor K i The value of the evaluation function under the given condition; γ is the evaluation function value for all safety factors;

[0049] When optimizing the pre-selected scheme based on the intelligent support decision model, the larger the value of the evaluation function γ, the safer the support structure.

[0050] Preferably, in step S6, the second excavation and support scheme generation model based on experiential knowledge data includes a module for obtaining candidate frequent itemsets, an association rule extraction and reasoning module.

[0051] Preferably, step S6 specifically includes the following sub-steps:

[0052] S61: Using the module for obtaining candidate frequent itemsets, a breadth-first search is used to construct candidate frequent itemsets in the tunnel support scheme knowledge table, obtaining the candidate frequent itemset Ec = {e1, e2, ..., e...} m ,…,e n} and perform support counting;

[0053] S62: Using the candidate frequent itemset, the association rule extraction and inference module employs the Apriori algorithm to mine nodes with a minimum confidence threshold, and retains nodes higher than the manually set minimum confidence threshold according to the needs of the real scenario; the confidence level is calculated using the following formula:

[0054]

[0055] In the formula: c(e) m →e n ) for e m and e n confidence level; e m and e n For frequently selected items; count(e m ∪e n ) contains e in K m and e n The number of K represents the knowledge table of the tunnel support scheme; count(e m ) contains e in K m The number of itemsets; m and n are the numbers of the frequent itemsets; L is the number of frequent itemsets;

[0056] S63: Search for data items related to the tunnel environment through steps S61 and S62, and repeat steps S61 and S62 until the tunnel support scheme knowledge table is completely traversed; use the result obtained after traversing the tunnel support scheme knowledge table as the excavation support scheme based on experience knowledge data.

[0057] Preferably, step S7, which integrates the preferred excavation support scheme and the excavation support scheme based on empirical knowledge data, specifically includes the following sub-steps:

[0058] If the excavation and support scheme based on experience knowledge data consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation and support scheme one.

[0059] If the preferred excavation support scheme consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation support scheme two.

[0060] The excavation support scheme 1 and the excavation support scheme 2 are fused together according to the principle of matching and weighting the key parameter items to obtain the data knowledge joint enhancement tunnel intelligent support decision scheme.

[0061] On the other hand, a data-knowledge-enhanced intelligent tunnel support decision-making device is also provided, comprising:

[0062] A tunnel support data acquisition device is used to collect and acquire the empirical knowledge data and the numerical simulation data.

[0063] The excavation and support scheme pre-selection module is used to pre-select an excavation and support scheme that is suitable for the tunnel environment parameters, and obtain the pre-selected scheme;

[0064] The intelligent support decision module is used to optimize the pre-selected scheme to obtain the optimal excavation support scheme;

[0065] The data knowledge enhancement learning module is used to generate an excavation and support scheme based on experiential knowledge data, and integrate it with the preferred excavation and support scheme to realize intelligent tunnel support decision-making through joint data and knowledge enhancement learning, thereby obtaining the intelligent tunnel support decision-making scheme jointly enhanced by data and knowledge.

[0066] The support decision interactive output module is used to interactively output the intelligent tunnel support decision scheme enhanced by the data knowledge to the user.

[0067] Preferably, the pre-selection module for excavation and support schemes includes an uncertainty reasoning module based on deep learning and an excavation and support scheme generation module based on numerical simulation data; the intelligent support decision-making module includes a mechanical evaluation parameter prediction module, a safety factor assessment module, and an evaluation function module.

[0068] The beneficial effects of this invention are:

[0069] 1. This invention integrates data on tunnel environment and construction equipment models and categories, and uses deep learning model algorithms to achieve potential correlation mining, thus solving the technical problems of diverse, complex, and isolated data.

[0070] 2. This invention realizes intelligent support decision-making in tunnel construction, and transforms the decision-making process that previously relied solely on human experience into a model-based process, making industry behavior digital and algorithmic, and achieving a timely and effective decision-making process.

[0071] 3. This invention realizes a data-knowledge-enhanced intelligent tunnel support decision-making method, which enables collaborative decision-making based on engineering data and experience, breaks down information barriers, enhances the security of decision-making, and achieves more reliable intelligent support decision-making, providing technical support for tunnel construction designers to make intelligent decisions. Attached Figure Description

[0072] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0073] Figure 1 This is a schematic diagram of the architecture of the data knowledge-enhanced intelligent tunnel support decision-making method and device described in this invention.

[0074] Figure 2 This is a schematic diagram of the excavation and support scheme generation model one based on numerical simulation data of the present invention;

[0075] Figure 3 This is a schematic diagram of the scheme generation process for Model 2 of the excavation and support scheme generation based on empirical knowledge data of the present invention.

[0076] Figure 4 This is a schematic diagram of the structure of the intelligent tunnel support decision-making device with data knowledge joint enhancement as described in this invention. Detailed Implementation

[0077] The present invention will be further described below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and technical features described in this application can be combined with each other. It should also be pointed out that, unless otherwise indicated, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. The terms "comprising" or "including" and similar words used in this invention refer to elements or objects preceding the word that encompass the elements or objects listed following the word and their equivalents, without excluding other elements or objects.

[0078] On the one hand, such as Figure 1 As shown, this invention provides a data-knowledge-enhanced intelligent tunnel support decision-making method, comprising the following steps:

[0079] S1: Collect and acquire experiential knowledge data and numerical simulation data.

[0080] In one specific embodiment, the empirical knowledge data includes groundwater volume, rock layer thickness, burial depth, geological structure, construction methods, shotcrete parameters, anchor bolt parameters, and steel arch frame parameters, etc.; the numerical simulation data includes cross-section type, excavation mode, excavation advance, step height, arch frame model, arch frame spacing, anchor bolt diameter, anchor bolt angle, small guide diameter, small guide ring spacing, initial support bearing ratio, arch crown displacement, arch waist displacement, and minimum principal stress at the arch crown, etc.

[0081] In one specific embodiment, after acquiring the experiential knowledge data and the numerical simulation data, the method further includes a statistical preprocessing step. Specifically: statistical preprocessing is performed on tunnel environment data based on real tunnel excavation and support scenarios and experiential knowledge-based support data under corresponding conditions; statistical preprocessing is also performed on numerical simulation data obtained from a simulated tunnel system. The expert experience and knowledge of technicians and engineers are summarized and statistically analyzed, and the different construction equipment and models applied under different tunnel environment conditions are organized and processed into a standardized knowledge table. Evolutionary analysis is performed on the tunnel environment conditions and applied tunnel equipment and models based on the simulated tunnel system, and the different evolutionary results under different tunnel environment conditions and construction equipment conditions are summarized and preprocessed.

[0082] S2: Determine tunnel environmental parameters and excavation support scheme parameters, and construct a tunnel support scheme knowledge table, a multivariate numerical simulation dataset, and an excavation support scheme library based on numerical simulation data.

[0083] In a specific embodiment, the tunnel support scheme knowledge table is the experience knowledge summarized and statistically analyzed in step S1, organized into multiple Excel files; the multivariate numerical simulation dataset is a collection of different tunnel environment data and different excavation and support scheme data and their evolutionary mechanical parameters obtained based on the simulated tunnel simulation system, organized into multiple Excel files; the excavation and support scheme library based on numerical simulation data is a safe excavation and support scheme obtained from the simulated tunnel simulation system based on the tunnel environment and its adapted evolution results, including tunnel environment parameters, excavation scheme parameters composed of construction equipment models and categories, support scheme parameters, and mechanical evaluation parameters.

[0084] In one specific embodiment, the knowledge table of tunnel support schemes includes data on hardness, integrity, rock layer thickness, groundwater volume, burial depth, geological structure, construction method, internal contour form, shotcrete + steel mesh, anchor bolts, steel arches, secondary lining, and advanced support. The multivariate numerical simulation dataset includes tunnel environmental data such as elastic modulus, Poisson's ratio, cohesion, internal friction angle, tensile strength, initial support load ratio, secondary lining load ratio, cross-sectional type, and ground stress.

[0085] S3: Based on the tunnel environment parameters, perform uncertainty reasoning on the excavation and support schemes in the excavation and support scheme library; if there is no excavation and support scheme that matches the tunnel environment parameters in the uncertainty reasoning results, proceed to step S4; if there is an excavation and support scheme that matches the tunnel environment parameters in the uncertainty reasoning results, use the matching excavation and support scheme as the pre-selected scheme, and proceed to step S5.

[0086] In one specific embodiment, an uncertainty reasoning model based on deep learning is used for uncertainty reasoning, which is existing technology, and the specific method will not be described in detail here. During reasoning, one or more excavation and support schemes that are suitable for the current tunnel environmental parameters are obtained by fuzzy matching using the industry-given range of surrounding rock grades and sensitive environmental parameters such as elastic modulus, Poisson's ratio, cohesion, and ground stress.

[0087] S4: Construct a first excavation and support scheme generation model based on numerical simulation data, and use the multivariate numerical simulation dataset as training parameters to obtain a trained first excavation and support scheme generation model; generate an excavation and support scheme adapted to the tunnel environment parameters according to the trained first excavation and support scheme generation model, and use the excavation and support scheme as a pre-selected scheme, and proceed to step S5.

[0088] In a specific embodiment, such as Figure 2 As shown, the excavation and support scheme generation model based on numerical simulation data includes an improved self-attention mechanism Transformer encoder module and a variational inference module; the improved self-attention mechanism Transformer encoder module includes an improved attention layer, a residual connection layer, a layer normalization layer, and a fully connected layer; the variational inference module includes an inference network and a generation network.

[0089] In the above embodiments, the improved self-attention mechanism of the Transformer encoder module and the variational inference module can reduce the distribution differences between levels, improve the robustness of the model, and thus uncover the potential correlation between the tunnel environment and the excavation and support scheme. By improving the self-attention mechanism, the model's ability to model non-stationary time-series data can be enhanced.

[0090] In one specific embodiment, the calculation formula for the improved attention layer is:

[0091]

[0092]

[0093]

[0094] In the formula: Q' and K' are the query matrix and key matrix obtained after normalization and matrix transformation; Q and K are the original query matrix and key matrix; μ Q and μ K These are the mean values ​​of Q and K over the time dimension, respectively; the superscript T indicates transpose; σ x d is the standard deviation of the original input; model The dimensions of Model 1 are generated for the excavation and support scheme.

[0095] In the above embodiments, the improved attention layer utilizes the normalized input and the statistics stored during normalization to approximate the original attention layer before normalization, thus optimizing the overly stable data after normalization and the model's temporal modeling capability. Furthermore, when using this invention, one can... Kμ Q These parameters are defined as destationary factors τ and Δ, and a multilayer perceptron (MLP) is used to adaptively learn them, thus allowing the non-stationary information of the time series data to be passed on within the model.

[0096] In one specific embodiment, the inference network is used to perform variational inference on the input to generate a probability distribution of the latent variables, as shown in the following formula:

[0097]

[0098] In the formula: q φ (z|x) represents the posterior probability distribution of the latent variables learned by the network given the input x. It follows a Gaussian distribution; μ, σ 2 These represent the mean and variance of the inference network output, respectively.

[0099] The generative network is used to receive latent variables and reconstruct the original input data by mapping them back to the data space, as shown in the following formula:

[0100]

[0101] In the formula: p θ (x'|z) represents the data distribution learned by the generative network given the latent variable z.

[0102] In the above embodiments, when generating an excavation and support scheme adapted to the tunnel environmental parameters based on the trained excavation and support scheme generation model one, the tunnel environmental parameters are used as input to the trained excavation and support scheme generation model one to predict and generate one or more excavation and support schemes adapted to the tunnel environmental parameters. The tunnel environmental parameters include elastic modulus, Poisson's ratio, cohesion, internal friction angle, tensile strength, and ground stress, etc. The data involved in the excavation and support scheme include excavation advance, step height, step length, small guide pipe diameter, initial concrete support thickness, initial concrete support strength, steel arch grade, steel arch spacing, anchor diameter, and secondary lining strength grade, etc.

[0103] S5: Construct an intelligent support decision model, and optimize the pre-selected scheme according to the intelligent support decision model to obtain the optimal excavation support scheme.

[0104] In one specific embodiment, the intelligent support decision-making model includes a mechanical evaluation parameter prediction model, a safety factor assessment model, and an evaluation function; the mechanical evaluation parameter prediction model is used to predict the mechanical evaluation parameters corresponding to the pre-selected scheme; the safety factor assessment model includes a surrounding rock displacement safety factor K1 assessment model, a support structure safety factor K2 assessment model, a support structure safety factor K3 assessment model, and a support structure safety factor K4 assessment model; wherein...

[0105] The evaluation model for the surrounding rock displacement safety factor K1 is as follows:

[0106]

[0107] In the formula: U L For the ultimate displacement; U p This is the average of the convergent values ​​of the three displacements: arch crown displacement, arch waist displacement, and arch bottom displacement.

[0108] The safety factor K2 evaluation model for the support structure is as follows:

[0109]

[0110] Where: σ L σ represents the ultimate stress of the steel arch frame. p For the stress of the steel arch frame;

[0111] The evaluation model for the safety factor K3 of the support structure is as follows:

[0112]

[0113] In the formula: f c This represents the ultimate compressive strength of the shotcrete. This represents the minimum principal stress convergence value at the crown of the arch;

[0114] The safety factor K4 evaluation model for the support structure is as follows:

[0115]

[0116] In the formula: f h d represents the ultimate compressive strength of concrete; d represents the thickness of the secondary lining in the support scheme; N1-N5 represent the axial force convergence values ​​in the five directions of the tunnel arch crown, arch shoulder, arch waist, arch foot, and arch bottom, respectively.

[0117] The evaluation function is:

[0118]

[0119]

[0120] Where: K iγ represents the magnitude of the i-th safety factor; i For the corresponding safety factor K i The value of the evaluation function under the given condition; γ is the evaluation function value for all safety factors;

[0121] When optimizing the pre-selected scheme based on the intelligent support decision model, the larger the value of the evaluation function γ, the safer the support structure.

[0122] In the above embodiments, the architecture of the mechanical evaluation parameter prediction model is the same as that of the excavation and support scheme generation model based on numerical simulation data, and both are trained using the multivariate numerical simulation dataset. The difference lies in the output of the models. In use, the tunnel environment parameters and parameters of each pre-selected scheme are input to obtain the mechanical evaluation parameters of each pre-selected scheme. These mechanical evaluation parameters include surrounding rock displacement, axial force, bending moment, principal stress, etc., which are used as input to the subsequent safety factor assessment model to calculate the safety factor.

[0123] When evaluating the safety factor using the aforementioned safety factor assessment model, the tunnel environmental parameters, pre-selected scheme parameters, and corresponding mechanical evaluation parameters are input to obtain the surrounding rock displacement safety factor K1, support structure safety factor K2, support structure safety factor K3, and support structure safety factor K4 for each pre-selected scheme. It should be noted that if the safety factors K1-K4 do not meet the industry-specified safety factor range, the process returns to step S4 to regenerate an excavation and support scheme adapted to the tunnel environmental parameters as a new pre-selected scheme, until the safety factors K1-K4 meet the industry-specified safety factor range, thus obtaining an excavation and support scheme that meets the industry-specified safety factor range.

[0124] In a specific embodiment, when optimizing the pre-selected schemes according to the intelligent support decision model, 1-5 safe and cost-effective excavation support schemes are selected. Specifically, when γ is less than 1.5, the corresponding scheme is an unsafe scheme; when γ is between 1.5 and 8, the corresponding scheme is a safe and cost-effective scheme; when γ is greater than 8, the corresponding scheme is a safe scheme but involves material waste. Therefore, the scheme with a safety factor closer to 1.5 is preferred.

[0125] S6: Construct a second excavation and support scheme generation model based on experiential knowledge data, and obtain an excavation and support scheme based on experiential knowledge data by combining the tunnel support scheme knowledge table.

[0126] In a specific embodiment, such as Figure 3 As shown, the second excavation and support scheme generation model based on experiential knowledge data includes a module for obtaining candidate frequent itemsets, an association rule extraction and reasoning module; step S6 specifically includes the following sub-steps:

[0127] S61: Using the module for obtaining candidate frequent itemsets, a breadth-first search is used to construct candidate frequent itemsets in the tunnel support scheme knowledge table, obtaining the candidate frequent itemset Ec = {e1, e2, ..., e...} m ,…,e n} and perform support counting;

[0128] S62: Using the candidate frequent itemset, the association rule extraction and inference module employs the Apriori algorithm to mine nodes with a minimum confidence threshold, and retains nodes higher than the manually set minimum confidence threshold according to the needs of the real scenario; the confidence level is calculated using the following formula:

[0129]

[0130] In the formula: c(e) m →e n ) for e m and e n confidence level; e m and e n For frequently selected items; count(e m ∪e n ) contains e in K m and e n The number of K represents the knowledge table of the tunnel support scheme; count(e m ) contains e in K m The number of itemsets; m and n are the numbers of the frequent itemsets; L is the number of frequent itemsets;

[0131] S63: Search for data items related to the tunnel environment through steps S61 and S62, and repeat steps S61 and S62 until the tunnel support scheme knowledge table is completely traversed; use the result obtained after traversing the tunnel support scheme knowledge table as the excavation support scheme based on experience knowledge data.

[0132] In the above embodiments, the second excavation and support scheme generation model based on empirical knowledge data, which combines empirical knowledge with the Apriori algorithm, can realize the relationship between tunnel environmental parameters and support schemes, including the hardness of surrounding rock, secondary lining parameters, advanced support parameters, shotcrete, anchor bolt parameters, etc., thus solving the problems of scattered, diverse, complex, and isolated data in existing technologies.

[0133] In one specific embodiment, 1-5 excavation and support schemes based on empirical knowledge data are generated through the excavation and support scheme generation model 2.

[0134] S7: The preferred excavation and support scheme and the excavation and support scheme based on experience knowledge data are integrated to obtain a data knowledge-enhanced intelligent tunnel support decision-making scheme.

[0135] In a specific embodiment, the fusion of the preferred excavation support scheme and the excavation support scheme based on empirical knowledge data specifically includes the following sub-steps:

[0136] If the excavation and support scheme based on experience knowledge data consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation and support scheme one.

[0137] If the preferred excavation support scheme consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation support scheme two.

[0138] The excavation support scheme 1 and the excavation support scheme 2 are fused together according to the principle of matching and weighting the key parameter items to obtain the data knowledge joint enhancement tunnel intelligent support decision scheme.

[0139] On the other hand, such as Figure 4 As shown, the present invention also provides a data knowledge-enhanced intelligent tunnel support decision-making device, comprising:

[0140] A tunnel support data acquisition device is used to collect and acquire the empirical knowledge data and the numerical simulation data.

[0141] The excavation and support scheme pre-selection module is used to pre-select an excavation and support scheme that is suitable for the tunnel environment parameters, and obtain the pre-selected scheme;

[0142] The intelligent support decision module is used to optimize the pre-selected scheme to obtain the optimal excavation support scheme;

[0143] The data knowledge enhancement learning module is used to generate an excavation and support scheme based on experiential knowledge data, and integrate it with the preferred excavation and support scheme to realize intelligent tunnel support decision-making through joint data and knowledge enhancement learning, thereby obtaining the intelligent tunnel support decision-making scheme jointly enhanced by data and knowledge.

[0144] The support decision interactive output module is used to interactively output the intelligent tunnel support decision scheme enhanced by the data knowledge to the user.

[0145] In one specific embodiment, the tunnel support data acquisition device includes Excel software, Word software, a tunnel simulation system, etc.

[0146] In one specific embodiment, the excavation and support scheme pre-selection module includes an uncertainty reasoning module based on deep learning and an excavation and support scheme generation module based on numerical simulation data; the intelligent support decision module includes a mechanical evaluation parameter prediction module, a safety factor assessment module, and an evaluation function module.

[0147] On the other hand, the present invention also provides a data-knowledge-enhanced intelligent tunnel support decision-making system device. This system device integrates the intelligent tunnel support decision-making device described in this invention and an intelligent decision-making edge computing terminal. The intelligent tunnel support decision-making device is deployed on the intelligent decision-making edge computing terminal. The intelligent decision-making edge computing terminal includes the physical equipment and system environment required by the intelligent tunnel support decision-making device. The physical equipment includes input / output devices, a processor, a memory, and a power supply. The input / output devices include a keyboard, mouse, and monitor, responsible for data interaction. The processor is used to implement the method of the present invention. The memory is used to store the instructions and data executed by the processor and the system environment required by the system. The power supply can be a portable power source or a fixed power source to provide energy to the device. The system environment includes a Python environment, a deep learning environment, and a MATLAB environment.

[0148] All steps of the intelligent tunnel support decision-making method described in this invention can be completed by hardware related to program instructions. The intelligent tunnel support decision-making system device can be stored in a computer-readable storage medium. When executed, the intelligent tunnel support decision-making system device performs the steps including those of the intelligent tunnel support decision-making method. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, magnetic disks, or server hard drives.

[0149] When the aforementioned integrated system device is implemented as a software functional unit and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This computer software product, stored in a storage medium, includes several instructions to cause the intelligent decision-making edge computing terminal to execute all or part of the method and device of this invention. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, magnetic disks, or server hard drives.

[0150] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A data-knowledge-enhanced intelligent tunnel support decision-making method, characterized in that, Includes the following steps: S1: Collect and acquire empirical knowledge data and numerical simulation data; S2: Determine tunnel environmental parameters and excavation support scheme parameters, and construct a tunnel support scheme knowledge table, a multivariate numerical simulation dataset, and an excavation support scheme library based on numerical simulation data; S3: Based on the tunnel environment parameters, perform uncertainty reasoning on the excavation and support schemes in the excavation and support scheme library; If no excavation and support scheme that fits the tunnel environment parameters is found in the uncertainty reasoning results, proceed to step S4; If the uncertainty reasoning results include an excavation and support scheme that is compatible with the tunnel environment parameters, then the compatible excavation and support scheme is selected as the pre-selected scheme, and the process proceeds to step S5. S4: Construct a first model for generating excavation and support schemes based on numerical simulation data, and use the multivariate numerical simulation dataset as training parameters to obtain a trained first model for generating excavation and support schemes. Based on the trained excavation and support scheme generation model, an excavation and support scheme adapted to the tunnel environment parameters is generated, and this excavation and support scheme is used as a pre-selected scheme to proceed to step S5. S5: Construct an intelligent support decision model, and optimize the pre-selected scheme according to the intelligent support decision model to obtain the optimal excavation support scheme; S6: Construct a second excavation and support scheme generation model based on experiential knowledge data, and obtain an excavation and support scheme based on experiential knowledge data by combining the tunnel support scheme knowledge table. S7: The preferred excavation and support scheme and the excavation and support scheme based on experience knowledge data are integrated to obtain a data knowledge-enhanced intelligent tunnel support decision-making scheme.

2. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 1, characterized in that, In step S4, the excavation and support scheme generation model based on numerical simulation data includes an improved self-attention mechanism Transformer encoder module and a variational inference module; the improved self-attention mechanism Transformer encoder module includes an improved attention layer, a residual connection layer, a layer normalization layer, and a fully connected layer; the variational inference module includes an inference network and a generation network.

3. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 2, characterized in that, The improved calculation formula for the attention layer is as follows: （1） （2） （3） In the formula: , These are the query matrix and key matrix obtained after normalization and matrix transformation; , These are the original query matrix and key matrix; and They are respectively in the time dimension , Mean; superscript Indicates transpose; The standard deviation of the original input; The dimensions of Model 1 are generated for the excavation and support scheme.

4. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 2, characterized in that, The inference network is used to perform variational inference on the input to generate the probability distribution of the latent variables, as shown in the following formula: （4） In the formula: To give input Under the given conditions, infer the posterior probability distribution of the latent variables learned by the network; It follows a Gaussian distribution; , These represent the mean and variance of the inference network output, respectively. The generative network is used to receive latent variables and reconstruct the original input data by mapping them back to the data space, as shown in the following formula: （5） In the formula: Indicates that given latent variables Under certain conditions, the data distribution learned by the generative network.

5. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 1, characterized in that, In step S5, the intelligent support decision model includes a mechanical evaluation parameter prediction model, a safety factor assessment model, and an evaluation function; the mechanical evaluation parameter prediction model is used to predict the mechanical evaluation parameters corresponding to the pre-selected scheme. The safety factor assessment model includes the surrounding rock displacement safety factor. Evaluation model, support structure safety factor Evaluation model, support structure safety factor Evaluation model, support structure safety factor Evaluation model; among which, The surrounding rock displacement safety factor The evaluation model is as follows: （6） In the formula: This is the ultimate displacement; This is the average of the convergent values ​​of the three displacements: arch crown displacement, arch waist displacement, and arch bottom displacement. The safety factor of the support structure The evaluation model is as follows: （7） In the formula: This represents the ultimate stress of the steel arch frame. For the stress of the steel arch frame; The safety factor of the support structure The evaluation model is as follows: （8） In the formula: This represents the ultimate compressive strength of the shotcrete. This represents the minimum principal stress convergence value at the crown of the arch; The safety factor of the support structure The evaluation model is as follows: （9） In the formula: This represents the ultimate compressive strength of concrete. The thickness of the secondary lining in the support scheme; - These are the axial force convergence values ​​in five directions: the tunnel arch crown, arch shoulder, arch waist, arch foot, and arch bottom. The evaluation function is: （10） （11） In the formula: Let i be the size of the i-th safety factor; For the corresponding safety factor The value of the evaluation function under; The evaluation function values ​​for all safety factors; When optimizing the pre-selected scheme based on the intelligent support decision model, the evaluation function The higher the value, the safer the support structure.

6. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 1, characterized in that, In step S6, the second excavation and support scheme generation model based on experiential knowledge data includes a module for obtaining candidate frequent itemsets, an association rule extraction and reasoning module.

7. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to claim 6, characterized in that, Step S6 specifically includes the following sub-steps: S61: Using the module for obtaining candidate frequent itemsets, a breadth-first search is used to construct candidate frequent itemsets in the tunnel support scheme knowledge table, obtaining the candidate frequent itemset Ec={e1,e2, …,e m ,…,e n } and perform support counting; S62: Using the candidate frequent itemset, the association rule extraction and inference module employs the Apriori algorithm to mine nodes with a minimum confidence threshold, and retains nodes higher than the manually set minimum confidence threshold according to the needs of the real scenario; the confidence level is calculated using the following formula: （12） In the formula: for and Confidence level; and These are frequently selected items; K contains and The number of tunnel support schemes, K is the knowledge table of the tunnel support scheme; K contains The number; For the number of frequent items; The number of frequent itemsets; S63: Search for data items related to the tunnel environment through steps S61 and S62, and repeat steps S61 and S62 until the tunnel support scheme knowledge table is completely traversed; use the result obtained after traversing the tunnel support scheme knowledge table as the excavation support scheme based on experience knowledge data.

8. The data-knowledge joint enhancement method for intelligent tunnel support decision-making according to any one of claims 1-7, characterized in that, Step S7, which integrates the preferred excavation support scheme and the excavation support scheme based on empirical knowledge data, specifically includes the following sub-steps: If the excavation and support scheme based on experience knowledge data consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation and support scheme one. If the preferred excavation support scheme consists of multiple schemes, they are merged into one scheme according to the weighted average principle to obtain excavation support scheme two. The excavation support scheme 1 and the excavation support scheme 2 are fused together according to the principle of matching and weighting the key parameter items to obtain the data knowledge joint enhancement tunnel intelligent support decision scheme.

9. A tunnel intelligent support decision-making device based on data knowledge joint enhancement according to any one of claims 1-8, characterized in that, include: A tunnel support data acquisition device is used to collect and acquire the empirical knowledge data and the numerical simulation data. The excavation and support scheme pre-selection module is used to pre-select an excavation and support scheme that is suitable for the tunnel environment parameters, and obtain the pre-selected scheme; The intelligent support decision module is used to optimize the pre-selected scheme to obtain the optimal excavation support scheme; The data knowledge enhancement learning module is used to generate an excavation and support scheme based on experiential knowledge data, and integrate it with the preferred excavation and support scheme to realize intelligent tunnel support decision-making through joint data and knowledge enhancement learning, thereby obtaining the intelligent tunnel support decision-making scheme jointly enhanced by data and knowledge. The support decision interactive output module is used to interactively output the intelligent tunnel support decision scheme enhanced by the data knowledge to the user.

10. The data-knowledge-enhanced intelligent tunnel support decision-making device according to claim 9, characterized in that, The pre-selection module for excavation and support schemes includes an uncertainty reasoning module based on deep learning and an excavation and support scheme generation module based on numerical simulation data; the intelligent support decision-making module includes a mechanical evaluation parameter prediction module, a safety factor assessment module, and an evaluation function module.

Images