Large-diameter shield tunneling parameter optimization method and system based on digital twinning

Abstract

The invention discloses a large-diameter shield tunneling parameter optimization method and system based on digital twinning, and relates to the technical field of shield tunnel numerical simulation, machine learning and digital twinning, and the method comprises the steps: obtaining shield tunnel construction site data, and constructing a virtual digital twinning body; building a training sample set based on virtual numerical twinborn generation data; training an agent model by using the training sample set; during construction, actual stratum response data and tunneling parameter data are obtained in real time, the real-time data are input into the soil mass parameter inversion agent model, and the optimal value of the soil mass parameter is rapidly output; the optimal value is input into the virtual digital twin, and dynamic real-time updating of numerical simulation is carried out; and on the basis of the updated soil parameter data and the tunneling parameter data in the next construction step, stratum response prediction data are obtained through the stratum response prediction agent model, so that the tunneling parameter data are adjusted and optimized. According to the invention, more accurate numerical simulation and safe and efficient tunneling of the shield tunnel can be realized.

Description

Technical Field

[0001] The present invention relates to the fields of shield tunnel numerical simulation, machine learning and digital twin technology, and in particular to a large-diameter shield tunnel excavation parameter optimization method and system based on digital twin. Background Art

[0002] With the gradual advancement of urbanization, underground space engineering construction, mainly tunnel engineering, has entered a stage of rapid development. Tunnel engineering construction has gradually attracted attention because of its advantages of occupying less ground resources and solving the current congestion problems on the urban ground. As the planning and construction of tunnel engineering become more and more diversified, the difficulties faced in the survey and construction stages of tunnel engineering construction are also gradually increasing, and different construction methods have emerged during the construction of tunnel engineering. The main methods for urban underground space engineering construction include shield method, New Austrian Tunneling Method and open-cut method. Among them, shield method is safe in construction operation and has a high degree of mechanization in the construction process. In recent years, it has been widely used in urban and rural tunnel construction. Since urban underground space engineering construction is mainly concentrated in places with high building density and population density, surrounded by high-rise buildings and complex distribution of urban pipelines, a little carelessness will cause huge losses, especially the excavation section of shield tunnel construction is getting larger and larger, and disasters are more likely to occur during large-diameter shield construction. Therefore, with the trend of increasingly widespread application of large-diameter shield machines, how to predict and control the risks in the construction process (such as controlling risks by optimizing excavation parameters) during the construction of large-diameter shield machines to ensure the economy and safety of the entire tunnel construction cycle has become an urgent need for the efficient and safe development of current underground space projects.

[0003] At present, numerical simulation is usually used to carry out preliminary analysis in the tunnel design stage. Through numerical simulation analysis of dynamic changes in the process of large-diameter shield tunnel excavation, detailed understanding of each step of surface settlement and surrounding soil disturbance deformation such as excavation, shield tail grouting, upper segment lining and grouting layer hardening can be obtained, providing relevant basis for design, monitoring measurement, prediction and regulation. At present, the commonly used numerical simulation methods for large-diameter shield tunnel excavation include discrete element method, finite element method, finite difference method, etc. The finite element method is based on continuous media and generally assumes small deformation; the discrete element method studies non-continuous media from a microscopic scale and is often used to study large-grained sandstone and broken rock mass; the finite difference method is more suitable for elastic-plastic problem analysis and large deformation assumptions. However, the above-mentioned existing numerical simulation methods have certain problems, namely: the fidelity of numerical simulation determines the effectiveness of numerical simulation. The existing numerical simulation methods have large data volume, slow calculation efficiency, and long time consumption. The actual tunnel formation data changes in real time. The existing numerical simulation has a certain time lag and cannot achieve real-time numerical simulation. Moreover, due to the large amount of calculation, the fidelity of numerical simulation is poor, which leads to poor effectiveness of numerical simulation, and ultimately affects the real-time and accuracy of shield tunnel excavation prediction analysis, and then affects the real-time and accuracy of the final large-diameter shield tunnel excavation parameter optimization adjustment. Summary of the invention

[0004] In order to address the deficiencies of the above-mentioned prior art, the present invention provides a large-diameter shield tunnel excavation parameter optimization method and system based on digital twins. By introducing digital twin technology, a virtual digital twin of the physical space of the large-diameter shield tunnel construction site is constructed to generate a large amount of sample data. An efficient stratum response proxy model is constructed through a machine learning algorithm, and then a model dynamic update algorithm is generated in combination with an intelligent optimization algorithm. The dynamic update is applied to the virtual digital twin to achieve dynamic real-time update of numerical simulation and real-time and accurate prediction of stratum response data. Based on the predicted stratum response data and combined with relevant specifications for large-diameter shield tunneling, an optimization algorithm is used to optimize the excavation parameters, thereby achieving safe and efficient excavation of large-diameter shield tunnels.

[0005] In a first aspect, the present invention provides a large-diameter shield tunnel excavation parameter optimization method based on digital twin.

[0006] A large-diameter shield tunneling parameter optimization method based on digital twin, including:

[0007] Obtain field survey data and engineering design information at the shield tunnel construction site to build a virtual digital twin;

[0008] Based on the virtual digital twin, the corresponding stratum response data is calculated according to the stratum soil parameter data and excavation parameter data to construct a training sample set;

[0009] The training sample set is used to train the soil parameter inversion and formation response prediction proxy models respectively;

[0010] During shield tunnel construction, the actual stratum response data and excavation parameter data are obtained in real time, and the real-time data is input into the soil parameter inversion proxy model to quickly output the optimal values ​​of stratum soil parameters;

[0011] The optimal values ​​of the soil parameters are input into the virtual digital twin to perform dynamic real-time updates of the numerical simulation;

[0012] Based on the updated soil parameter data and the excavation parameter data of the next construction step, the formation response prediction data is obtained through the formation response prediction agent model;

[0013] Based on the formation response prediction data and combined with the large-diameter shield specifications, the excavation parameter data is adjusted and optimized with construction efficiency and construction cost as the goals.

[0014] In the second aspect, the present invention provides a large-diameter shield tunnel excavation parameter optimization system based on digital twin.

[0015] A large-diameter shield tunneling parameter optimization system based on digital twin, including:

[0016] A virtual digital twin construction module is used to obtain field survey data and engineering design information at the shield tunnel construction site to build a virtual digital twin;

[0017] The sample data construction module is used to calculate the corresponding formation response data based on the virtual digital twin, according to the formation soil parameter data and the excavation parameter data, and construct a training sample set;

[0018] An agent model training module is used to train soil parameter inversion and formation response prediction agent models respectively using training sample sets;

[0019] The soil parameter inversion module is used to obtain the actual stratum response data and excavation parameter data in real time during shield tunnel construction, input the real-time data into the soil parameter inversion proxy model, and quickly output the optimal values ​​of stratum soil parameters;

[0020] Numerical simulation dynamic update module, used to input the optimal values ​​of stratum soil parameters into the virtual digital twin, and perform dynamic real-time update of numerical simulation;

[0021] A stratum response prediction module is used to obtain stratum response prediction data through a stratum response prediction agent model based on updated stratum soil parameter data and excavation parameter data at the next construction step;

[0022] The tunneling parameter optimization module is used to adjust and optimize the tunneling parameter data based on the formation response prediction data and the large-diameter shield specifications, with the construction efficiency and construction cost as the goals.

[0023] In a third aspect, the present invention further provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps of the method described in the first aspect are completed.

[0024] In a fourth aspect, the present invention further provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the method described in the first aspect.

[0025] One or more of the above technical solutions have the following beneficial effects:

[0026] 1. The present invention provides a large-diameter shield tunnel excavation parameter optimization method and system based on digital twins. By introducing digital twin technology, the problem of poor fidelity of numerical models in traditional numerical simulations is solved. In the present invention, a virtual digital twin is constructed for the physical space (i.e., physical entity) of the shield tunnel construction site to generate a large amount of sample data. Then, a machine learning algorithm is used to construct an efficient soil parameter inversion and stratum response prediction agent model. The model dynamic update algorithm is generated in combination with the intelligent optimization algorithm. The optimal values ​​of stratum soil parameters are quickly obtained through dynamic model update, and the obtained optimal values ​​of stratum soil parameters are directly input into the In the virtual numerical twin (i.e., the numerical simulation model), dynamic real-time updating of numerical simulation results is achieved to avoid problems such as low computational efficiency, large amount of computation and inaccuracy, thereby ensuring the fidelity of numerical simulation results and the effectiveness of numerical simulation, and providing stable and reliable data support for the predictive analysis of shield tunneling. The dynamic update is applied to the virtual digital twin to achieve dynamic real-time updating of numerical simulation and real-time and accurate prediction of stratum response data. Finally, based on the predicted stratum response data and combined with relevant specifications for large-diameter shield tunneling, an optimization algorithm is used to optimize the excavation parameters, thereby achieving safe and efficient excavation of large-diameter shield tunnels.

[0027] 2. The present invention constructs an agent model based on a large amount of sample data and a machine learning algorithm. By combining an intelligent optimization algorithm to construct a model dynamic update algorithm, the problem of slow computing efficiency is solved, the time of numerical calculation is greatly shortened, and the rapid and real-time dynamic update of the numerical simulation of the physical entity-virtual twin of the shield tunnel is realized.

[0028] 3. The present invention provides a means of real-time prediction for on-site disaster prediction, that is, through the digital twin technology, it realizes the real-time and accurate prediction of the formation response. Based on the predicted formation response data such as surface settlement and face ground stress, combined with the construction specifications, it provides a basis for making safety decisions in advance on site, thereby ensuring the safe construction of shield tunnels. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0030] Figure 1 It is a framework diagram of the interaction between a physical entity and a virtual twin in an embodiment of the present invention;

[0031] Figure 2 Schematic diagram of a large-diameter shield tunnel excavation parameter optimization method based on digital twin in an embodiment of the present invention;

[0032] Figure 3 It is an overall flow chart of a large-diameter shield tunnel excavation parameter optimization method based on digital twin in an embodiment of the present invention;

[0033] Figure 4 The figure is a flow chart of constructing an agent model in an embodiment of the present invention. DETAILED DESCRIPTION

[0034] It should be noted that the following detailed descriptions are exemplary only, are intended to describe specific embodiments, are intended to provide further explanation of the present invention, and are not intended to limit exemplary embodiments according to the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those of ordinary skill in the art to which the present invention belongs. In addition, it should also be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0035] Embodiment 1

[0036] In view of the problem that the existing numerical simulation methods have large data volume, poor calculation accuracy, slow calculation efficiency, and long time consumption, which leads to poor fidelity of numerical simulation, this embodiment proposes a large-diameter shield tunnel excavation parameter optimization method based on digital twins, which introduces digital twin technology and machine learning algorithms. Considering that a large amount of construction data has also been accumulated in the current shield tunneling construction process, the machine learning algorithm can use the massive data to provide effective solutions for the problems encountered in the shield tunnel construction process, and use the machine learning algorithm to further improve the effectiveness and efficiency of data inversion; at the same time, the digital twin technology establishes a corresponding digital virtual model, namely a virtual digital twin, based on the physical entity, and uses the Internet of Things, communication technology and other means to realize the real-time transmission of data between the physical entity and the virtual digital twin, and finally meets the high-fidelity simulation of the physical entity by numerical simulation. That is, the digital twin method proposed in the present invention can realize the real-time data interaction between the physical entity and the digital twin, and further realize the accurate mapping of the physical entity and the digital twin, and ensure the fidelity of the numerical simulation.

[0037] The present embodiment proposes a method for optimizing the excavation parameters of a large-diameter shield tunnel based on digital twins. First, based on field survey data and engineering design data, a refined numerical simulation model, namely a virtual digital twin, is established. The range of stratum soil parameters (hereinafter referred to as soil parameters) and construction excavation parameter data are obtained, and a random parameter sample library is established by a sampling method. All the obtained random parameter working condition combinations are input into the virtual digital twin for numerical simulation. By setting monitoring points on the twin, stratum response data is generated, so as to establish a simulation database with working condition combinations and stratum response as data. Because the stratum distribution of the initial virtual digital twin is obtained by limited survey drilling data, the actual stratum distribution has a strong spatial difference. Therefore, a model dynamic update algorithm is established to realize the rapid acquisition of the optimal values ​​of the stratum soil parameters, so as to ensure that the numerical calculation results and the actual monitored data are highly matched, that is, based on a large amount of sample data and machine learning The algorithm constructs an agent model, and constructs a model dynamic update algorithm by combining an intelligent optimization algorithm. The mapping relationship between the virtual digital twin input parameters and the stratum response data is established through a machine learning algorithm, that is, the stratum soil parameters and construction excavation parameters are used as independent variables X and the simulation calculation result data of the stratum response is used as the dependent variable Y. Then, the actual construction monitoring data of the physical space obtained on site is used as the final optimization goal, and the optimal values ​​of the stratum soil parameters are quickly obtained through the model dynamic update algorithm. After that, the obtained optimal values ​​of the stratum soil parameters are substituted into the virtual digital twin to achieve dynamic real-time update of the numerical simulation results, ensure the fidelity of the numerical simulation results, and realize real-time and accurate prediction of the stratum response in the unexcavated area of ​​the shield tunnel. Finally, according to the stratum response prediction data, combined with the relevant specifications and requirements of large-diameter shield machines, with the construction efficiency and construction cost as the goals, the optimization algorithm is used to adjust and optimize the excavation parameter data.

[0038] The large-diameter shield tunnel excavation parameter optimization method based on digital twin proposed in this embodiment specifically includes the following steps: Figure 1 The three parts shown in the figure are: on-site construction physical space, virtual digital twin, and dynamic real-time update of the model. The specific process of this method is as follows Figure 2 As shown, the following steps are included:

[0039] Step S1: Before the shield tunnel is constructed, obtain the on-site survey data and engineering design information of the shield tunnel construction site and build a virtual digital twin. Figure 3 As shown, based on the construction engineering design data and on-site survey data, a refined two-dimensional numerical analysis model (i.e., a refined numerical model) is established by using finite difference software, and then a virtual digital twin is constructed.

[0040] Furthermore, the field survey data is used to describe the distribution of the strata, including soil physical and mechanical index parameters, groundwater level parameters, pore water pressure parameters and soil constitutive model parameters, etc. Among them, soil physical and mechanical index parameters include soil layer thickness, relative density, saturated weight, cohesion, Poisson's ratio, water content, porosity, soil pressure coefficient and permeability coefficient (i.e. permeability), etc.; soil constitutive model parameters include structural strength, elastic modulus, internal friction angle, etc.; groundwater level parameters refer to monitoring the changes in groundwater level, and the rise and fall of groundwater level will affect the pore water pressure and stability of the soil; pore water pressure parameters reflect the changes in pore water pressure during shield construction. The above-mentioned field survey data is obtained by interpolation of data obtained by the drilling method based on a limited number of survey points, and the actual geological conditions have strong spatial differences; wherein, the arrangement of the survey points on site is determined according to the actual situation, mainly considering the engineering geology and hydrogeology conditions, as well as the relationship between the surface environment and the tunnel axis. Based on a limited number of survey points, the drilling method is used to obtain the corresponding measurement data, and reasonable standards are established to pre-process the drilling data. In this embodiment, the obtained measurement data is interpolated to obtain the field survey data and construct a real stratigraphic model.

[0041] Furthermore, the above-mentioned engineering design data include detailed drawings of tunnel cross-sections, longitudinal sections, lining designs, support structures, etc., as well as construction drawings describing construction steps, construction methods, construction materials, equipment layout, etc.

[0042] Based on the geological and hydrological data obtained from the on-site survey and the engineering design data of the shield tunnel construction project and other real data, a refined two-dimensional numerical analysis model is established using finite difference software (in fact, this model is a virtual digital twin), and the model is given real geological parameters (including soil type, internal friction angle, cohesion, groundwater parameters, elastic modulus, etc.), boundary conditions (including groundwater position, boundary displacement, etc.), units (i.e., solid units) and excavation simulation methods (i.e., step-by-step excavation) to construct a finite difference numerical analysis model. The constructed finite difference numerical analysis model can be used as the initial virtual digital twin.

[0043] The above-mentioned process of calculation using finite difference software (this embodiment uses the existing FLAC3D software) includes: finite difference grid generation, selection of difference format for approximating the well-defined solution problem of the equation, simultaneous solution of algebraic equations, imposition of boundary conditions and initial conditions, iterative solution, and post-processing analysis.

[0044] Based on the refined numerical model (i.e., virtual digital twin) constructed above, certain initial conditions are imposed, such as changing the speed of shield advancement, changing soil conditions, changing groundwater conditions, etc., and different surface settlement values ​​and other formation response data can be obtained through solving.

[0045] However, in actual engineering projects, due to the disturbance caused by tunnel construction, the physical and mechanical properties of the strata evolve continuously, and the refined calculation model has a large amount of calculation and low calculation efficiency. This will cause the monitoring stress and displacement results (i.e., stratum response data) after simulation by the virtual digital twin to be inconsistent with the actual data of the on-site construction, resulting in low fidelity of the numerical simulation, that is, the virtual digital twin is not ideal in the process of realizing real-time data interaction. To this end, this embodiment builds a model that can be calculated, namely, a proxy model, to quickly obtain the optimal values ​​of the stratum soil parameters, and then substitutes the obtained parameters into the virtual digital twin, so as to achieve dynamic real-time update of the numerical simulation results and ensure the fidelity of the numerical simulation results. Specifically, such as Figure 4 As shown, the proxy model is constructed and trained as follows:

[0046] Step S2: Based on the virtual digital twin, the corresponding stratum response data is calculated according to the stratum soil parameter data and the excavation parameter data, and a training sample set is constructed. Specifically, on the basis of using the simulation software to construct the virtual digital twin, the stratum soil parameter and the excavation parameter data are used as variable data, multiple groups of variable data are extracted based on the random sampling method, and each variable data is substituted or input into the virtual digital twin of the simulation software for calculation to obtain the corresponding stratum response data; based on multiple groups of mutually corresponding stratum soil parameters, excavation parameters and stratum response data, a training sample set is constructed.

[0047] Step S3: using the training sample set to train soil parameter inversion and formation response prediction proxy models respectively.

[0048] Among them, the soil parameter inversion proxy model is trained using the training sample set, with the stratum response data and excavation parameter data as independent variables and the soil parameter data as the dependent variable, to learn the mapping relationship among the stratum response data, excavation parameter data and soil parameter data.

[0049] Similarly, the training sample set is used to train the stratum response prediction agent model, with soil parameter data and excavation parameter data as independent variables and stratum response data as dependent variables, to learn the mapping relationship between stratum soil parameters, excavation parameters and stratum response data.

[0050] In the above step S2, training sample data is generated based on the virtual digital twin. The real stratum soil parameters and excavation parameter data of each layer of soil of the virtual digital twin are used as independent variables. The stratum soil parameters include permeability, cohesion, Poisson's ratio, soil density, soil pressure coefficient, elastic modulus, internal friction angle, etc. The excavation parameters include shield excavation speed, total thrust, grouting elastic modulus, grouting thickness, soil bin / mud bin pressure, cutter head torque, etc. After obtaining the independent variable data, assuming that all selected independent variable data obey normal distribution, select the Latin sampling method to extract 1000 groups of random samples to obtain sample data.

[0051] Then, the selected sample data are substituted into the virtual digital twin to carry out numerical calculations (which can be automatically calculated through software), and the formation response data such as surface settlement values ​​and horizontal ground stress values ​​of the tunnel face in the numerical calculation results are exported and used as the dependent variables of the training sample data.

[0052] Through the above method, the independent variables and dependent variables of the training samples are composed of 1000 sets of stratum soil parameters, excavation parameters and stratum response data respectively.

[0053] As another implementation method, in the above process, in order to avoid the influence of null values ​​and extreme values, the real soil parameter data is preprocessed, including: first, removing the null values ​​in the data; then, setting upper and lower limits to filter the data, that is, to avoid too little data, 1.2 times and 0.8 times the average value of each independent variable are set as the upper and lower limits of sampling to remove the calculated values.

[0054] In step S3, based on the stratum soil parameter-excavation parameter-stratum response data sample data constructed in step S2, the stratum soil parameter and excavation parameter data are taken as independent variables X, and the stratum response data is taken as the dependent variable Y. A mapping relationship between X and Y is established through a variety of machine learning intelligent algorithms such as random forest, convolutional neural network, multivariate adaptive regression, deep learning, etc., so as to construct a stratum response prediction agent model.

[0055] Similarly, the formation response data and excavation parameter data are taken as independent variables X, and the formation soil parameters are taken as dependent variables Y. The mapping relationship between X and Y is established through a variety of machine learning intelligent algorithms such as random forest, convolutional neural network, multivariate adaptive regression, deep learning, etc., thereby constructing a soil parameter inversion agent model.

[0056] The above-mentioned proxy model adopts one of the most accurate machine learning algorithms for numerical simulation. This embodiment adopts convolutional neural network (CNN), which has good prediction accuracy and generalization ability. Further, in the process of training the proxy model using this algorithm, Adam and other optimization intelligent algorithms are used to achieve the matching of monitoring data and real data in the numerical analysis results, and find the one with the smallest error between the two. That is, this embodiment adopts Adam-CNN intelligent algorithm to train the data set obtained by numerical simulation, wherein the training accuracy of CNN model mainly comes from learning rate (Learning Rate), batch size (Batch Size), number of iterations (Number ofEpochs), convolution kernel size (Filter Size), number of convolution kernels (Number ofFilters), stride (Stride), padding (Padding), pooling layer specifications (Pooling Layer Specifications), activation function (ActivationFunction), optimizer (Optimizer) these ten hyperparameters, and the Adam optimization algorithm is used to select the optimal parameters, so that the CNN model can better realize the construction of associations between data after optimization, making its prediction more accurate.

[0057] In the above steps, a data sample is constructed and used in combination with deep learning and intelligent optimization algorithms to train a proxy model. The proxy model is used to reduce the time required for numerical calculations and ensure the timeliness of real-time dynamic updates of the proxy model.

[0058] Step S4: During shield tunnel construction, actual stratum response data and excavation parameter data are acquired in real time, and the real-time data is input into the soil parameter inversion proxy model to quickly output the optimal values ​​of stratum soil parameters.

[0059] In this embodiment, the actual measurement function of the physical space of the on-site construction is composed of the Internet of Things and sensors. Among them, the sensors are deployed at monitoring points of key on-site monitoring locations such as the surface and the shield machine, and multi-scale and multi-dimensional massive construction monitoring data are obtained through the sensors. The data includes formation response data, excavation parameter data, such as surface settlement data, soil moisture content, groundwater level, total thrust, propulsion speed, grouting elastic modulus, grouting thickness, soil bin / mud bin pressure, cutter head torque, etc. The Internet of Things is built using a 5G network to transmit multi-scale and multi-dimensional massive sensor data obtained from the construction site to the receiving room in real time, and the receiving room processes the data.

[0060] The ultimate goal of optimization is to obtain actual monitoring data of the physical space on site. The acquired real-time actual construction monitoring data is input into the soil parameter inversion proxy model. The optimal values ​​of the formation soil parameters are quickly obtained through the dynamic real-time update algorithm of the model, and the formation soil parameters such as internal friction angle, elastic modulus, cohesion, Poisson's ratio, soil density, and earth pressure coefficient are output.

[0061] Step S5: input the optimal values ​​of the stratum soil parameters into the virtual digital twin to perform dynamic real-time updating of the numerical simulation.

[0062] Specifically, the optimal values ​​of the obtained stratum soil parameters are substituted into the virtual digital twin to achieve dynamic real-time update of the numerical simulation results and ensure the fidelity of the numerical simulation results. The stratum soil parameters are dynamically updated here. During the shield tunneling process, the stratum parameters need to be inverted once for each excavation step. The inverted parameters are used to update the stratum parameters of the current excavation section of the simulation, thereby realizing the dynamic update of the virtual digital twin. For example, the stratum response data monitored on-site is real-time. When the Nth ring of the shield tunnel is excavated, the real-time stratum response data is obtained. The parameter inversion Adam-CNN model of step S4 is first used to carry out soil parameter inversion to obtain the current stratum soil parameters, realize real-time update of the stratum parameters, and then use the stratum parameters to dynamically update the virtual digital twin. The interaction process from the physical world to the virtual world is to display the monitored stratum response data on the virtual twin, and the interaction process from the virtual world to the physical world is the real-time inversion and update of the stratum parameters of the physical world. The above process is repeated continuously to realize the real-time dynamic interaction between the monitoring data of the physical world and the virtual twin data in the digital twin.

[0063] Step S6: Based on the updated stratum soil parameter data and the excavation parameter data of the next construction step, the stratum response prediction data is obtained through the stratum response prediction agent model.

[0064] With the dynamic real-time update of the virtual twin model, it can provide theoretical support for the safe construction of shield tunnels. Specifically, first in the database, the updated stratum soil parameters, the excavation parameters of the next construction step, such as shield excavation speed, soil bin / mud bin pressure, cutter head torque, total thrust, grouting elastic modulus and grouting thickness, are used as input, and the stratum response data is used as output. The stratum response prediction agent model established above is used to predict the stratum response characteristic data during future excavation, such as surface settlement and face ground stress, based on the actual parameters of shield construction.

[0065] Furthermore, the predicted formation response characteristics can be combined with construction specifications to provide a basis for making safety decisions in advance on site. For example, when the predicted formation response characteristics show a dangerous state, different construction plans can be previewed in advance through the proxy model, and the excavation parameter values ​​of safe and efficient construction plans can be quickly given. Due to the high efficiency of the proxy model, effective plans can be quickly provided to the site to assist in the adjustment of construction excavation parameters and provide reliable theoretical support for safe construction on site.

[0066] That is, step S7, according to the formation response prediction data, combined with the large-diameter shield specification, with construction efficiency and construction cost as the goal, adjust and optimize the excavation parameter data.

[0067] First, parameter analysis and specification integration are carried out in advance. Specifically, the excavation parameters involved (such as shield excavation speed, total thrust, grouting elastic modulus, grouting thickness, soil bin / mud bin pressure, cutter head torque, etc.) are analyzed in detail, and expert consultation and data statistical analysis are used to clarify the impact mechanism of each parameter on stratum response, construction efficiency and construction cost; at the same time, the relevant specifications for large-diameter shield construction, including stratum settlement control standards, construction safety requirements, engineering quality specifications, etc., are comprehensively collected and sorted out, and a database of the correlation between parameters and specifications is established to provide a basic basis for subsequent optimization.

[0068] Secondly, based on the rapid prediction value of the formation response data, combined with the above-mentioned pre-integrated relevant specifications for large-diameter shields, a multi-objective optimization model is constructed with construction efficiency and construction cost as the goals, so as to achieve the adjustment and optimization of the excavation parameters. Specifically, construction efficiency and construction cost are the main optimization goals, where construction efficiency can be quantified as the excavation length or earth excavation volume per unit time, and construction cost can be quantified as a comprehensive cost covering equipment operation cost, material consumption cost, labor cost, etc., and the calculation method and weight coefficient of each cost item can be determined based on actual engineering data; on this basis, a multi-objective optimization mathematical model (abbreviated as optimization model) is constructed with the formation response data (such as surface settlement and face ground stress, etc.) and the excavation parameter data meeting the specification requirements as constraints.

[0069] For the multi-objective optimization mathematical model constructed above, optimization algorithms such as genetic algorithm and particle swarm optimization algorithm are used to optimize the excavation parameters to find the parameter combination with the highest construction efficiency and the lowest cost under the requirements of stratum stability, so as to adjust and optimize the excavation parameter data. On this basis, the trained stratum response prediction agent model is used to preview different excavation parameter adjustment schemes, simulate the stratum response and construction effect under each scheme, and then provide construction personnel with multiple sets of feasible parameter adjustment schemes and corresponding prediction results, including information such as stratum deformation trend, construction progress change, cost estimation, etc., to assist them in making scientific and reasonable decisions. Preferably, in the decision-making process, the actual situation of the project site, construction experience and risk preference can be comprehensively considered to select the optimal excavation parameter adjustment strategy, and the implementation effect can be continuously monitored to further optimize the parameters according to the actual situation.

[0070] As an implementation method, during the shield tunnel construction process, the stratum response data of the unexcavated area, excavation parameters and other field information are judged in real time through the Internet of Things and sensors, soil layer parameter inversion model and stratum response prediction model, and updated to the optimization model in a timely manner. In this way, the parameters and constraints in the optimization model can be dynamically adjusted according to the latest data feedback, and the optimization calculation can be re-performed to achieve real-time optimization and adjustment of the excavation parameters. For example, when it is monitored that the surface settlement is close to the critical value allowed by the specification, the variables that have a greater impact on the stratum settlement, such as the shield excavation speed and grouting parameters, are adjusted first to ensure construction safety and compliance with the specifications, while taking into account the balance between construction efficiency and construction cost.

[0071] Preferably, the optimization results of the final excavation parameters are evaluated and fed back, that is, a complete optimization result evaluation mechanism is established to evaluate the actual effects of the excavation parameters after adjustment and optimization, including whether the formation response is stable within the specification range, whether the construction efficiency is improved, whether the cost is effectively controlled, etc. The evaluation results are then fed back to the optimization model, and the parameters and algorithms of the model are corrected and improved, so as to continuously improve the accuracy and reliability of the optimization model and achieve continuous improvement in the optimization of excavation parameters.

[0072] Embodiment 2

[0073] This embodiment provides a large-diameter shield tunnel excavation parameter optimization system based on digital twin, which specifically includes:

[0074] A virtual digital twin construction module is used to obtain field survey data and engineering design information at the shield tunnel construction site to build a virtual digital twin;

[0075] The sample data construction module is used to calculate the corresponding formation response data based on the virtual digital twin, according to the formation soil parameter data and the excavation parameter data, and construct a training sample set;

[0076] An agent model training module is used to train soil parameter inversion and formation response prediction agent models respectively using training sample sets;

[0077] The soil parameter inversion module is used to obtain the actual stratum response data and excavation parameter data in real time during shield tunnel construction, input the real-time data into the soil parameter inversion proxy model, and quickly output the optimal values ​​of stratum soil parameters;

[0078] Numerical simulation dynamic update module, used to input the optimal values ​​of stratum soil parameters into the virtual digital twin, and perform dynamic real-time update of numerical simulation;

[0079] A stratum response prediction module is used to obtain stratum response prediction data through a stratum response prediction agent model based on updated stratum soil parameter data and excavation parameter data at the next construction step;

[0080] The tunneling parameter optimization module is used to adjust and optimize the tunneling parameter data based on the formation response prediction data and the large-diameter shield specifications, with the construction efficiency and construction cost as the goals.

[0081] Embodiment 3

[0082] This embodiment provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, the steps in the large-diameter shield tunnel excavation parameter optimization method based on digital twins as described above are completed.

[0083] Embodiment 4

[0084] This embodiment also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps in the large-diameter shield tunnel excavation parameter optimization method based on digital twins as described above are completed.

[0085] The steps involved in the above embodiments 2 to 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0086] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0087] The above description is only a preferred embodiment of the present invention. Although the specific implementation mode of the present invention is described in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Technical personnel in the relevant field should understand that on the basis of the technical solution of the present invention, various modifications or deformations that can be made by technical personnel in this field without creative work are still within the scope of protection of the present invention.

Claims

1. A large-diameter shield tunneling parameter optimization method based on digital twin, characterized in that: include: Obtain field survey data and engineering design information at the shield tunnel construction site to build a virtual digital twin; Based on the virtual digital twin, the corresponding stratum response data is calculated according to the stratum soil parameter data and excavation parameter data to construct a training sample set; The training sample set is used to train the soil parameter inversion and formation response prediction proxy models respectively; During shield tunnel construction, the actual stratum response data and excavation parameter data are obtained in real time, and the real-time data is input into the soil parameter inversion proxy model to quickly output the optimal values ​​of stratum soil parameters; The optimal values ​​of the soil parameters are input into the virtual digital twin to perform dynamic real-time updates of the numerical simulation; Based on the updated soil parameter data and the excavation parameter data of the next construction step, the formation response prediction data is obtained through the formation response prediction agent model; Based on the formation response prediction data and combined with the large-diameter shield specifications, the excavation parameter data is adjusted and optimized with construction efficiency and construction cost as the goals.

2. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 1, characterized in that: The field survey data is used to describe the distribution of strata, including: soil physical and mechanical index parameters, groundwater level parameters, pore water pressure parameters and soil constitutive model parameters; the soil physical and mechanical index parameters include soil layer thickness, relative density, saturated weight, cohesion, Poisson's ratio, water content, porosity, soil pressure coefficient and permeability coefficient; the soil constitutive model parameters include structural strength, elastic modulus and internal friction angle; The acquisition of the on-site survey data includes: Based on a limited number of survey points, the measurement data is obtained by drilling method; The acquired measurement data is interpolated to obtain the on-site survey data.

3. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 1, characterized in that: The construction of the virtual digital twin includes: Based on the engineering design data and on-site survey data of the shield tunnel construction project, a refined two-dimensional numerical analysis model was established using the finite difference method; Endow the two-dimensional numerical analysis model with real geological parameters, boundary conditions, entity units and excavation simulation methods, construct a finite difference numerical analysis model, and use the finite difference numerical analysis model as the initial virtual digital twin; The geological parameters include soil type, internal friction angle, cohesion, groundwater parameters, and elastic modulus; the boundary conditions include groundwater position and boundary displacement.

4. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 3 is characterized in that: The construction of the training sample set includes: Taking the soil parameters and tunneling parameters as variable data, multiple groups of variable data are extracted based on the random sampling method, and each variable data is substituted into the virtual digital twin for calculation to obtain the corresponding stratum response data; Based on multiple sets of corresponding stratum soil parameters, excavation parameters and stratum response data, a training sample set is constructed.

5. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 1, characterized in that: The soil parameters of the formation include permeability, cohesion, Poisson's ratio, soil density, earth pressure coefficient, elastic modulus, and internal friction angle; The excavation parameters include shield excavation speed, total propulsion force, grouting elastic modulus, grouting thickness, soil bin / mud bin pressure, and cutter head torque; The formation response data include surface settlement values ​​and horizontal ground stress values ​​at the tunnel face.

6. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 1, characterized in that: The training of soil parameter inversion proxy model and formation response prediction proxy model includes: The soil parameter inversion proxy model is trained using the training sample set, with the formation response data and the excavation parameter data as independent variables and the soil parameter data as the dependent variable, to learn the mapping relationship among the formation response data, the excavation parameter data and the soil parameter data; The training sample set is used to train the stratum response prediction agent model, with soil parameter data and excavation parameter data as independent variables and stratum response data as dependent variable, to learn the mapping relationship among stratum soil parameters, excavation parameters and stratum response data.

7. The large-diameter shield tunnel excavation parameter optimization method based on digital twin according to claim 1, characterized in that: The proxy model is built using any machine learning algorithm including random forest algorithm, convolutional neural network, multivariate adaptive regression algorithm, and deep learning algorithm; In the process of using algorithms to build and train agent models, genetic algorithms are used to optimize the training process.

8. A large-diameter shield tunneling parameter optimization system based on digital twin, characterized in that: include: A virtual digital twin construction module is used to obtain on-site survey data and engineering design information at the shield tunnel construction site to build a virtual digital twin; The sample data construction module is used to calculate the corresponding formation response data based on the virtual digital twin, according to the formation soil parameter data and the excavation parameter data, and construct a training sample set; An agent model training module is used to train soil parameter inversion and formation response prediction agent models respectively using training sample sets; The soil parameter inversion module is used to obtain the actual stratum response data and excavation parameter data in real time during shield tunnel construction, input the real-time data into the soil parameter inversion proxy model, and quickly output the optimal values ​​of stratum soil parameters; Numerical simulation dynamic update module, used to input the optimal values ​​of stratum soil parameters into the virtual digital twin, and perform dynamic real-time update of numerical simulation; A stratum response prediction module is used to obtain stratum response prediction data through a stratum response prediction agent model based on updated stratum soil parameter data and excavation parameter data at the next construction step; The tunneling parameter optimization module is used to adjust and optimize the tunneling parameter data based on the formation response prediction data and the large-diameter shield specifications, with the construction efficiency and construction cost as the goals.

9. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor. When the computer instructions are executed by the processor, the steps of a large-diameter shield tunnel excavation parameter optimization method based on digital twin are completed as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the steps of a large-diameter shield tunneling parameter optimization method based on digital twins as described in any one of claims 1-7.

Images