Efficient proxy analysis method and system for tunnel engineering multi-field information prediction

Abstract

The invention discloses an efficient agent analysis method and system for tunnel engineering multi-field information prediction, and the method comprises the steps: constructing a surrounding rock parameter inversion efficient agent model, a ground stress parameter inversion efficient agent model and a multi-field information prediction efficient agent model, and forming a digital twinborn body; respectively training the models by using data of a surrounding rock mechanical response simulation database; acquiring surrounding rock parameters, construction parameters and ground stress parameters of a current section of a tunnel construction site, and performing inversion by using the trained surrounding rock parameter inversion efficient proxy model and the ground stress parameter inversion efficient proxy model to obtain the surrounding rock parameters and the ground stress parameters; on the basis of the surrounding rock parameters and the ground stress parameters obtained through inversion, combined with the construction parameters, the trained multi-field information prediction efficient agent model is utilized, and multi-field information of the front non-excavated section is obtained through prediction. According to the method, the multi-field information of the non-excavated section can be efficiently and accurately predicted by using the digital twinborn body.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel multi-field information prediction, and in particular to an efficient proxy analysis method and system for tunnel engineering multi-field information prediction. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The construction of underground space engineering is a typical excavation and unloading process. Due to the excavation and unloading effect of tunnel construction, the stability of the surrounding rock is affected by many factors such as geological structure, ground stress, rock mechanical properties, construction methods, etc. These factors are usually unknown during the design and construction process. Therefore, the analysis results are quite different from the actual results. This leads to many safety hazards in the tunnel construction process, such as surrounding rock collapse, large deformation of surrounding rock, and block drop. Therefore, whether the stability of the surrounding rock can be predicted quickly and accurately using real-time monitoring information is directly related to the safe construction of the tunnel. How to predict the stress field, displacement field and other multi-field information of the tunnel through monitoring information to ensure the stability of the surrounding rock during tunnel excavation has been widely studied in the engineering and academic fields.

[0004] Tunnel monitoring and measurement is a common method to obtain multi-field information. However, the existing monitoring method is to bury sensors in the constructed section after tunnel excavation, which mainly has the following problems:

[0005] ①During the tunnel construction period from blasting to sensor installation, multiple field information data cannot be obtained in time, and there is a data collection vacuum period. Therefore, there are some missing multi-field information data in the monitoring;

[0006] ② The focus of monitoring and measurement is to monitor the excavated sections after tunnel excavation. It is not possible to carry out multiple information monitoring on the unexcavated sections; that is, heavy monitoring and difficult prediction.

[0007] In the prior art, the parameters of the surrounding rock and the ground stress parameters of the unexcavated area are often obtained by on-site drilling. However, this method is expensive and can only obtain the relevant parameters of the drilling position, but cannot obtain continuous geological parameters.

[0008] In recent years, the prediction of multi-field information of tunnels through numerical simulation has been widely used. Numerical simulation can establish a fine numerical calculation model through preliminary geological survey and engineering geological data. By carrying out calculations under different working conditions, geological parameter information at any position of the tunnel can be obtained to predict the mechanical behavior of the target area. However, the efficiency of numerical simulation calculation after fine modeling is low, and each working condition takes a long time, which cannot meet the real-time guidance of construction. Moreover, after the model is established, if the input surrounding rock and ground stress parameters do not match the actual ones, better simulation results need to be obtained through parameter calibration. The parameter calibration process is cumbersome and it is difficult to obtain accurate parameters. The results obtained by numerical simulation cannot be well integrated with the on-site monitoring information. Summary of the invention

[0009] In order to solve the above problems, the present invention proposes an efficient proxy analysis method and system for predicting multi-field information of tunnel engineering. Based on numerical simulation, a surrounding rock mechanical response simulation database is established. Through machine learning combined with optimization algorithms, efficient proxy models for inversion of surrounding rock parameters and ground stress parameters and efficient proxy models for predicting multi-field information are established respectively. The field monitoring data are integrated to realize the prediction of multi-field information of the unexcavated section of the tunnel.

[0010] In some embodiments, the following technical solutions are adopted:

[0011] An efficient proxy analysis method for multi-field information prediction of tunnel engineering, comprising:

[0012] Obtain the surrounding rock parameters, construction parameters and ground stress parameters at the tunnel construction site and establish a refined tunnel model; perform equal probability sampling on the obtained parameters and then randomly combine them into sample data matrices for different working conditions;

[0013] Based on the simulation of each working condition by finite difference software, multiple monitoring sections are set in the tunnel model to monitor the multi-field information of each section. The multi-field information of each section and the corresponding numerical simulation working condition sample data are used as a set of data to construct a surrounding rock mechanical response simulation database and perform normalization; the multi-field information includes displacement field and stress field;

[0014] Construct high-efficiency proxy models for inversion of surrounding rock parameters, high-efficiency proxy models for inversion of ground stress parameters, and high-efficiency proxy models for prediction of multi-field information to form digital twins; train the above models respectively using data from the surrounding rock mechanical response simulation database;

[0015] Obtain surrounding rock parameters, construction parameters and geostress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and geostress parameter inversion efficient proxy model to invert the surrounding rock parameters and geostress parameters respectively;

[0016] Based on the surrounding rock parameters and ground stress parameters obtained by inversion and combined with the construction parameters, the multi-field information of the unexcavated section ahead is predicted using the trained multi-field information prediction efficient agent model.

[0017] Furthermore, each working condition is composed of a variety of parameter data obtained by sampling, and the parameters include: surrounding rock parameters, construction parameters and ground stress parameters.

[0018] Furthermore, the surrounding rock parameters include cohesion, internal friction angle, elastic modulus and Poisson's ratio; the construction parameters include construction footage and construction methods; and the ground stress parameters include tunnel burial depth, surrounding rock lateral pressure coefficient and formation density.

[0019] Furthermore, based on the data in the surrounding rock mechanical response simulation database, the construction parameters, multi-field information parameters, soil density and tunnel burial depth are used as inputs, and the surrounding rock parameters are used as outputs to train an efficient proxy model for surrounding rock parameter inversion;

[0020] Based on the data in the surrounding rock mechanical response simulation database, the efficient proxy model for inversion of geostress parameters is trained with construction parameters, multi-field information parameters, soil density and tunnel depth as input and lateral pressure coefficient as output.

[0021] Furthermore, based on the data in the surrounding rock mechanical response simulation database, the efficient agent model for multi-field information prediction is trained with surrounding rock parameters, ground stress parameters and construction parameters as input and the stress field and displacement field information of the front section as output.

[0022] Furthermore, the surrounding rock parameter inversion efficient proxy model, the ground stress parameter inversion efficient proxy model and the multi-field information prediction efficient proxy model all use the PSO-BP neural network model.

[0023] Furthermore, it also includes:

[0024] Based on the predicted multi-field information of the unexcavated section ahead, if the predicted value is not within the safety range, the construction speed is reduced and the digital twin is used to re-simulate the prediction until the predicted multi-field information meets the safety range requirements; and the corresponding construction speed is fed back to the construction site;

[0025] or,

[0026] Based on the predicted multi-field information of the unexcavated section ahead, if the predicted value is not within the safety range, the three-bench method is selected for construction; if it is within the safety range, the full-section method is used for construction.

[0027] In other embodiments, the following technical solutions are adopted:

[0028] An efficient proxy analysis system for multi-field information prediction of tunnel engineering, comprising:

[0029] The data acquisition module is used to obtain the surrounding rock parameters, construction parameters and ground stress parameters of the tunnel construction site and establish a refined tunnel model; the acquired parameters are sampled with equal probability and then randomly combined into sample data matrices of different working conditions;

[0030] A database construction module is used to simulate each working condition based on finite difference software, set multiple monitoring sections in the tunnel model, monitor the multi-field information of each section, and use the multi-field information of each section and the corresponding numerical simulation working condition sample data as a set of data to construct a surrounding rock mechanical response simulation database and perform normalization; the multi-field information includes displacement field and stress field;

[0031] The model training module is used to construct an efficient proxy model for inversion of surrounding rock parameters, an efficient proxy model for inversion of ground stress parameters, and an efficient proxy model for prediction of multiple field information to form a digital twin; the above models are trained respectively using the data from the surrounding rock mechanical response simulation database;

[0032] The multi-field information prediction module is used to obtain the surrounding rock parameters, construction parameters and ground stress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and ground stress parameter inversion efficient proxy model to invert the surrounding rock parameters and ground stress parameters respectively; based on the inverted surrounding rock parameters and ground stress parameters, combined with the construction parameters, the trained multi-field information prediction efficient proxy model is used to predict the multi-field information of the unexcavated section ahead.

[0033] In other embodiments, the following technical solutions are adopted:

[0034] A terminal device includes a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store multiple instructions, and the instructions are suitable for the processor to load and execute the above-mentioned efficient proxy analysis method for multi-field information prediction of tunnel engineering.

[0035] In other embodiments, the following technical solutions are adopted:

[0036] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the above-mentioned efficient proxy analysis method for predicting multi-field information of tunnel engineering.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] (1) The present invention establishes a database containing surrounding rock parameters, geostress parameters, construction parameters, multi-field information and other characteristics through finite difference software; at the same time, an efficient proxy model for inversion of surrounding rock parameters and geostress parameters and an efficient proxy model for prediction of multi-field information are constructed to form a digital twin. Based on the multi-field information (stress, displacement) of the current section at the tunnel construction site, the digital twin can be used to realize efficient and accurate prediction of the multi-field information of the unexcavated section; the problems of low efficiency, inaccurate parameters and inability to combine numerical simulation with field engineering data in traditional research are solved.

[0039] (2) The present invention feeds back the multi-field information results obtained by the digital twin prediction to the construction site, and can provide guidance on the construction method and construction speed; at the same time, the construction speed can be adjusted based on the prediction results, and then the prediction results are obtained again through the digital twin, so as to obtain a suitable construction speed and feed it back to the construction site; so that the multi-field information results obtained by numerical simulation can be well integrated and utilized with the on-site monitoring information.

[0040] Other features and advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of an efficient agent analysis method for predicting multiple field information of tunnel engineering in an embodiment of the present invention;

[0042] Figure 2 A schematic diagram of refined modeling in an embodiment of the present invention;

[0043] Figure 3 It is a parameter characteristic structure diagram of the surrounding rock mechanical response simulation database in an embodiment of the present invention;

[0044] Figure 4 It is a schematic diagram of an efficient proxy model for inversion of surrounding rock parameters and ground stress parameters in an embodiment of the present invention;

[0045] Figure 5 Schematic diagram of an efficient agent model for multi-field information prediction in an embodiment of the present invention;

[0046] Figure 6 This is a schematic diagram of construction guidance based on a digital twin system in an embodiment of the present invention;

[0047] Figure 7 It is a structural diagram of a multi-field information prediction system based on tunnel construction mechanical behavior agent in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0049] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0050] Embodiment 1

[0051] In one or more embodiments, an efficient proxy analysis method for multi-field information prediction of tunnel engineering is disclosed. According to the on-site geological survey information, a refined numerical calculation model is established using Rhino software. A large number of numerical calculations are carried out on geological parameters that affect the stability of the surrounding rock, surrounding rock parameters, ground stress parameters, construction parameters and other working conditions, and a surrounding rock mechanical response simulation database is established. Next, the BP neural network model is combined with the particle swarm optimization algorithm PSO, with the surrounding rock parameters (cohesion c, internal friction angle φ, elastic modulus E, Poisson's ratio μ) and ground stress parameters (lateral pressure coefficient K) in the database as output parameters, and other parameters as input parameters, to establish and train the optimized surrounding rock parameter inversion efficient proxy model and ground stress parameter inversion efficient proxy model; Then, the BP neural network model is combined with the particle swarm optimization algorithm PSO, with the multi-field information (displacement field, stress field) in the database as output parameters, and the ground stress parameters (tunnel burial depth h, soil density ρ, lateral pressure coefficient K), surrounding rock parameters (cohesion c, internal friction angle φ, elastic modulus E, Poisson's ratio μ), construction parameters (construction speed v, construction method-full section excavation / three-bench excavation) as input parameters, to establish and train the optimized multi-field information prediction efficient proxy model. Within a certain range, the surrounding rock parameters remain unchanged. Then, the stress field, displacement field and other multi-field information of the unexcavated section ahead can be actually and quickly obtained based on the multi-field information monitored on site. Finally, through continuous iteration of data, a digital twin system that interacts with the physical space and the virtual twin model is realized, which is used to guide on-site construction.

[0052] Combination Figure 1 The specific process of the method in this embodiment is as follows:

[0053] S101: Obtain surrounding rock parameters, construction parameters and ground stress parameters at the tunnel construction site, and establish a refined tunnel model; perform equal probability sampling on the obtained parameters, and then randomly combine them into a working condition sample data matrix.

[0054] Specifically, during the geological exploration stage, drilling exploration is carried out on the construction site to obtain on-site construction parameters, including construction speed (excavation speed) v and construction method (full-section excavation / three-bench excavation), etc.; surrounding rock parameters and ground stress parameters are obtained through drilling, among which the surrounding rock parameters include cohesion φ, internal friction angle c, elastic modulus E and Poisson's ratio μ; ground stress parameters include tunnel burial depth h, soil density ρ and lateral pressure coefficient K.

[0055] Since drilling holes at the construction site consumes a lot of manpower and material resources and is costly, in this embodiment, a model is established based on the geotechnical module in the civil software, and a refined tunnel model is established through the Rhino software. Figure 2 Assuming that the tunnel model is 200m long, during the tunnel excavation process, a monitoring section is set up every 10m, so a 200m tunnel has 20 monitoring sections; therefore, multiple sets of data can be obtained after carrying out numerical simulation of each working condition; not only the amount of data obtained is increased, but also different burial depths can better reflect the degree of refinement of the model.

[0056] Based on the initial drilling data, Latin hypercube (LHS sampling) random sampling is performed on surrounding rock parameters (cohesion φ, internal friction angle c, elastic modulus E, Poisson's ratio μ), ground stress parameters (tunnel burial depth h, soil density ρ, lateral pressure coefficient K), and construction parameters (construction speed v, construction method-full-section excavation / three-step excavation) that affect the stability of tunnel surrounding rock. Among them, the tunnel burial depth has been reflected when establishing the refined model, the soil density remains basically unchanged, and the soil density and the construction method do not participate in the LHS sampling. After the LHS sampling of other parameters forms a data matrix, the data are combined with the construction method, and each data combination corresponds to a group of working condition sample data; in this embodiment, there are at least 200 groups of working condition sample data, and numerical analysis can be carried out on each working condition sample data of the model through FLAC3D software.

[0057] S102: Simulate each working condition based on finite difference software, set multiple monitoring sections in the model, monitor the multi-field information of each section, take the multi-field information of each section and the corresponding numerical simulation working condition parameters as a set of data, build a surrounding rock mechanical response simulation database, and normalize it; the multi-field information includes displacement field and stress field.

[0058] In this embodiment, for each set of working condition sample data, multiple monitoring sections can be set in the model to obtain more data. The input parameters considered in the numerical simulation and the stress field, displacement field and other multi-field information data of each monitoring section are combined into a set of data. The amount of sample data should be large enough (at least 1000 sets) as a simulation database of surrounding rock mechanical response; Figure 3A structural diagram of the surrounding rock mechanical response simulation database is given. Each group of data in the database includes four types of data: A, B, C and D. Among them, Class A data includes: soil density ρ, soil lateral pressure coefficient K and tunnel burial depth h; Class B data includes elastic modulus E, Poisson's ratio μ, internal friction angle c and cohesion φ; Class C data includes excavation speed v and construction method (including full-section excavation / three-bench excavation); Class D data includes stress field information and displacement field information.

[0059] This embodiment can simulate a large amount of relevant data through numerical simulation, without the need for a large number of drillings, which can save costs and improve efficiency.

[0060] S103: constructing an efficient proxy model for inversion of surrounding rock parameters, an efficient proxy model for inversion of ground stress parameters, and an efficient proxy model for prediction of multiple field information respectively to form a digital twin; and training the above models respectively using data from a surrounding rock mechanical response simulation database.

[0061] In this embodiment, combined with Figure 4 , based on the data in the surrounding rock mechanical response simulation database, with construction parameters (including excavation speed v and construction method), multi-field information parameters (displacement and stress), soil density ρ and tunnel burial depth h as input, and surrounding rock parameters (including elastic modulus E, Poisson's ratio μ, internal friction angle c and cohesion φ) as output, an efficient proxy model for surrounding rock parameter inversion is trained.

[0062] Similarly, based on the data in the surrounding rock mechanical response simulation database, the construction parameters (including excavation speed v and construction method), multi-field information parameters (displacement and stress), soil density ρ and tunnel burial depth h are used as inputs, and the lateral pressure coefficient K is used as output to train the efficient proxy model for inversion of ground stress parameters.

[0063] The above two models can quickly and accurately output surrounding rock parameters (cohesion c, internal friction angle φ, elastic modulus E, Poisson's ratio μ) and ground stress parameters (lateral pressure coefficient K).

[0064] Combination Figure 5 Based on the data in the surrounding rock mechanical response simulation database, the surrounding rock parameters (including elastic modulus E, Poisson's ratio μ, internal friction angle c and cohesion φ), ground stress parameters (including tunnel burial depth h, soil density ρ, lateral pressure coefficient K) and construction parameters (including excavation speed v and construction method) are used as input, and the stress field and displacement field information of the front section are used as output to train the efficient agent model for multi-field information prediction.

[0065] This model can quickly and accurately output multiple field information such as stress and displacement of the unexcavated section.

[0066] In this embodiment, the high-efficiency proxy model for inversion of surrounding rock parameters, the high-efficiency proxy model for inversion of ground stress parameters and the high-efficiency proxy model for prediction of multiple field information all use the PSO-BP neural network model. The initial weights and thresholds of the BP neural network model are optimized by the particle swarm optimization algorithm (PSO) programmed by matlab to achieve the optimization of the BP neural network model.

[0067] By constructing, learning and training the PSO-BP neural network model in MATLAB, 70% of the samples of the surrounding rock mechanical response simulation database after normalization are used as the data set for training the model, and the model established by the training set is tested to achieve the purpose of verifying the generalization of the model. 30% of the samples of the surrounding rock mechanical response simulation database after normalization are used as the data set for testing the model, and the accuracy of the established model is quantitatively judged through evaluation indicators.

[0068] The digital twin of this embodiment integrates the efficient proxy model for inversion of surrounding rock parameters and geostress parameters and the efficient proxy model for prediction of multi-field information, and realizes the establishment of the digital twin system through continuous data inversion and prediction iteration.

[0069] In the actual construction process, surrounding rock parameters such as elastic modulus E, Poisson's ratio μ, internal friction angle c and cohesion φ, as well as the lateral pressure coefficient K are not easy to obtain. If they are obtained by drilling, the economic efficiency is not high. If numerical simulation is used, it will take a lot of time and the results may be inaccurate. Therefore, this embodiment constructs a neural network model, uses the data from the actual excavation process to invert these parameters, and then predicts the displacement and stress of the unexcavated section ahead based on the inverted parameters, thereby solving the problems of low efficiency, inaccurate parameters and inability to combine numerical simulation with field engineering data in traditional research.

[0070] S104: Obtain surrounding rock parameters, construction parameters and geostress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and geostress parameter inversion efficient proxy model to invert the surrounding rock parameters and geostress parameters respectively.

[0071] In this embodiment, when the tunnel construction reaches the monitoring section S, the multi-field information (displacement, stress) of the monitoring information is obtained in real time and brought into the digital twin system, which can reflect the multi-field information (displacement, stress) of the S+K section ahead in real time, thereby realizing the prediction of the multi-field information (displacement, stress) of the section.

[0072] Specifically, the displacement field and stress field data of the construction site, the excavation speed v and construction method, the soil density ρ and the tunnel burial depth h information are obtained in real time, and they are respectively input into the trained surrounding rock parameter inversion efficient proxy model and the ground stress parameter inversion efficient proxy model to realize the inversion of the elastic modulus E, Poisson's ratio μ, internal friction angle c, cohesion φ and lateral pressure coefficient K.

[0073] Then, the surrounding rock parameters and ground stress parameters obtained by inversion are combined with the on-site construction parameters and input into the trained multi-field information prediction efficient agent model, which can predict the displacement field and stress field and other multi-field information of the unexcavated section ahead.

[0074] S105: The predicted multi-field information (displacement, stress) is fed back to the site, and guidance is given on the construction method and speed.

[0075] Specifically, combined Figure 6 According to the multi-field information prediction (displacement, stress) of the monitored section S and the multi-field information prediction (displacement, stress) of the unexcavated section S+K, there are two cases that can be considered:

[0076] ① When any of the predicted displacement, stress, and displacement change rate is too large and not within the set safety range, the construction speed can be reduced in the digital twin to find a safe speed, and then fed back to the construction site to improve construction safety. Among them, the displacement change rate can be calculated by the predicted displacement value, and the change rate is the tangent value of the change curve.

[0077] ② When any of the predicted displacement, stress, and displacement change rate is too small, the construction speed can be increased in the digital twin to obtain a safe speed, which can be fed back to the construction site to improve construction efficiency.

[0078] It can also provide guidance from the perspective of construction methods. When the predicted value is too large and exceeds the set safety range, the three-step method can be selected to improve construction safety; when the predicted value is too small, the full-section method can be used to improve construction efficiency.

[0079] In this embodiment, the set safety range is determined according to relevant tunnel construction specifications.

[0080] Embodiment 2

[0081] In one or more embodiments, an efficient proxy analysis system for multi-field information prediction of tunnel engineering is disclosed, combined with Figure 7 , specifically including:

[0082] The data acquisition module is used to obtain the surrounding rock parameters, construction parameters and ground stress parameters of the tunnel construction site and establish a refined tunnel model; the acquired parameters are sampled with equal probability and then randomly combined into sample data matrices of different working conditions;

[0083] A database construction module is used to simulate each working condition based on finite difference software, set multiple monitoring sections in the tunnel model, monitor the multi-field information of each section, and use the multi-field information of each section and the corresponding numerical simulation working condition sample data as a set of data to construct a surrounding rock mechanical response simulation database and perform normalization; the multi-field information includes displacement field and stress field;

[0084] The model training module is used to construct an efficient proxy model for inversion of surrounding rock parameters, an efficient proxy model for inversion of ground stress parameters, and an efficient proxy model for prediction of multiple field information to form a digital twin; the above models are trained respectively using the data from the surrounding rock mechanical response simulation database;

[0085] The multi-field information prediction module is used to obtain the surrounding rock parameters, construction parameters and geostress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and geostress parameter inversion efficient proxy model to invert the surrounding rock parameters and geostress parameters respectively; based on the inverted surrounding rock parameters and geostress parameters, combined with the construction parameters, the trained multi-field information prediction efficient proxy model is used to predict the multi-field information of the unexcavated section ahead.

[0086] It also includes a construction guidance module, which is used to reduce the construction speed based on the predicted multi-field information of the unexcavated section ahead. If the predicted value is not within the safety range, the simulation prediction is re-performed using the digital twin until the predicted multi-field information meets the safety range requirements; and the corresponding construction speed is fed back to the construction site;

[0087] or,

[0088] It is used for multiple field information of the unexcavated section ahead based on the prediction. If the predicted value is not within the safe range, the three-step method is selected for construction; if it is within the safe range, the full-section method is used for construction.

[0089] It should be noted that the specific implementation methods of the above modules have been described in Example 1 and will not be described in detail.

[0090] Embodiment 3

[0091] In one or more embodiments, a terminal device is disclosed, including a server, the server including a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the efficient proxy analysis method for multi-field information prediction of tunnel engineering in Example 1 when executing the program. For the sake of brevity, it will not be described here.

[0092] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0093] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0094] In the implementation process, each step of the above method can be completed by an integrated logic circuit of hardware in a processor or an instruction in the form of software.

[0095] Embodiment 4

[0096] In one or more embodiments, a computer-readable storage medium is disclosed, in which a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the efficient agent analysis method for predicting multi-field information of tunnel engineering described in Example 1.

[0097] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. An efficient proxy analysis method for multi-field information prediction of tunnel engineering, characterized in that: include: Obtain the surrounding rock parameters, construction parameters and ground stress parameters at the tunnel construction site and establish a refined tunnel model; perform equal probability sampling on the obtained parameters and then randomly combine them into sample data matrices for different working conditions; Based on the simulation of each working condition by finite difference software, multiple monitoring sections are set in the tunnel model to monitor the multi-field information of each section. The multi-field information of each section and the corresponding numerical simulation working condition sample data are used as a set of data to construct a surrounding rock mechanical response simulation database and perform normalization; the multi-field information includes displacement field and stress field; Construct high-efficiency proxy models for inversion of surrounding rock parameters, high-efficiency proxy models for inversion of ground stress parameters, and high-efficiency proxy models for prediction of multi-field information to form digital twins; train the above models respectively using data from the surrounding rock mechanical response simulation database; Obtain surrounding rock parameters, construction parameters and geostress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and geostress parameter inversion efficient proxy model to invert the surrounding rock parameters and geostress parameters respectively; Based on the surrounding rock parameters and ground stress parameters obtained by inversion and combined with the construction parameters, the multi-field information of the unexcavated section ahead is predicted using the trained multi-field information prediction efficient agent model.

2. An efficient proxy analysis method for multi-field information prediction of tunnel engineering as claimed in claim 1, characterized in that: Each working condition is composed of a variety of parameter data obtained by sampling, and the parameters include: surrounding rock parameters, construction parameters and ground stress parameters.

3. The efficient proxy analysis method for multi-field information prediction of tunnel engineering according to claim 1, characterized in that: The surrounding rock parameters include cohesion, internal friction angle, elastic modulus and Poisson's ratio; the construction parameters include construction footage and construction methods; the ground stress parameters include tunnel burial depth, surrounding rock lateral pressure coefficient and formation density.

4. The efficient proxy analysis method for multi-field information prediction of tunnel engineering according to claim 1, characterized in that: Based on the data in the surrounding rock mechanical response simulation database, the construction parameters, multi-field information parameters, soil density and tunnel depth are used as inputs, and the surrounding rock parameters are used as outputs to train the surrounding rock parameter inversion efficient proxy model; Based on the data in the surrounding rock mechanical response simulation database, the efficient proxy model for inversion of geostress parameters is trained with construction parameters, multi-field information parameters, soil density and tunnel depth as input and lateral pressure coefficient as output.

5. The efficient proxy analysis method for multi-field information prediction of tunnel engineering according to claim 1, characterized in that: Based on the data in the surrounding rock mechanical response simulation database, the high-efficiency proxy model for multi-field information prediction is trained with surrounding rock parameters, ground stress parameters and construction parameters as input and the stress field and displacement field information of the front section as output.

6. The efficient proxy analysis method for multi-field information prediction of tunnel engineering according to claim 1, characterized in that: The surrounding rock parameter inversion efficient proxy model, the ground stress parameter inversion efficient proxy model and the multi-field information prediction efficient proxy model all use the PSO-BP neural network model.

7. The efficient proxy analysis method for multi-field information prediction of tunnel engineering according to claim 1, characterized in that: Also includes: Based on the predicted multi-field information of the unexcavated section ahead, if the predicted value is not within the safety range, the construction speed is reduced and the digital twin is used to re-simulate the prediction until the predicted multi-field information meets the safety range requirements; and the corresponding construction speed is fed back to the construction site; or, Based on the predicted multi-field information of the unexcavated section ahead, if the predicted value is not within the safety range, the three-bench method is selected for construction; if it is within the safety range, the full-section method is used for construction.

8. An efficient agent analysis system for multi-field information prediction of tunnel engineering, characterized in that: include: The data acquisition module is used to obtain the surrounding rock parameters, construction parameters and ground stress parameters of the tunnel construction site and establish a refined tunnel model; the acquired parameters are sampled with equal probability and then randomly combined into sample data matrices of different working conditions; A database construction module is used to simulate each working condition based on finite difference software, set multiple monitoring sections in the tunnel model, monitor the multi-field information of each section, and use the multi-field information of each section and the corresponding numerical simulation working condition sample data as a set of data to construct a surrounding rock mechanical response simulation database and perform normalization; the multi-field information includes displacement field and stress field; The model training module is used to construct an efficient proxy model for inversion of surrounding rock parameters, an efficient proxy model for inversion of ground stress parameters, and an efficient proxy model for prediction of multiple field information to form a digital twin; the above models are trained respectively using the data from the surrounding rock mechanical response simulation database; The multi-field information prediction module is used to obtain the surrounding rock parameters, construction parameters and geostress parameters of the current section of the tunnel construction site, and use the trained surrounding rock parameter inversion efficient proxy model and geostress parameter inversion efficient proxy model to invert the surrounding rock parameters and geostress parameters respectively; based on the inverted surrounding rock parameters and geostress parameters, combined with the construction parameters, the trained multi-field information prediction efficient proxy model is used to predict the multi-field information of the unexcavated section ahead.

9. A terminal device, comprising a processor and a memory, wherein the processor is used to implement instructions; and the memory is used to store multiple instructions, characterized in that: The instructions are suitable for being loaded by a processor and executing the efficient agent analysis method for predicting multi-field information of tunnel engineering as described in any one of claims 1-7.

10. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instructions are suitable for being loaded by a processor of a terminal device and executing the efficient agent analysis method for predicting multi-field information of tunnel engineering as described in any one of claims 1-7.

Images