Intelligent optimization method and system for three-dimensional structure design of long-distance water delivery tunnel lining based on physical information network

Abstract

The invention belongs to the field of hydraulic engineering, and discloses a physical information network-based intelligent optimization method and system for a three-dimensional structure design of a long-distance water delivery tunnel lining, and the method comprises the steps: S1, building a three-dimensional geometric model of the long-distance water delivery tunnel lining; s2, establishing a structural mechanical analysis physical model considering surrounding rock stress and a construction excavation process; s3, establishing a multi-parameter physical information network model and performing training; and S4, the lining bearing capacity, the maximum deformation and the crack width serve as constraint boundaries, the engineering cost serves as an optimization target, a set of lining structure design parameters is obtained through optimization, and the lining structure design parameters comprise tunnel shape parameters, lining thickness and construction supporting parameters. The method is simple and practical, the lining bearing capacity, the maximum deformation and the crack width can be efficiently and accurately analyzed and calculated according to existing data, inversion optimization of lining structure design parameters is achieved, and the optimal design parameter set is obtained.

Description

Technical Field

[0001] The present invention belongs to the field of water conservancy projects, and particularly relates to an intelligent optimization method and system for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the cyber-physical system. Background Technique

[0002] Long-distance water conveyance tunnels are important components in water conservancy projects such as water resource allocation, conveyance, flood control, and power generation. With the increasing demands of national construction, long-distance water conveyance tunnels gradually exhibit the following characteristics: First, they pass through various complex geological conditions; second, they have a large burial depth and bear high water pressure; third, the cross-sectional shape of the tunnel varies with the position of the tunnel. The above characteristics determine the complexity of the operating environment of the water conveyance tunnel and its importance in design.

[0003] The lining structure provides necessary support for the water conveyance tunnel, ensuring the safety and stability of the tunnel during long-term operation. In engineering, the usual practice for lining structure design is to draw up several different tunnel cross-section schemes for scheme comparison. However, this method involves complex geological and hydrological conditions and requires the analysis of multiple schemes, increasing the computational workload and the complexity of analysis; and effective scheme comparison requires a large amount of on-site data, increasing the upfront time and economic costs. There is an urgent need for an intelligent optimization method for lining structure design to improve the problems existing in the traditional method and be able to quickly and efficiently realize the lining structure design to meet the requirements of engineering safety and stability.

[0004] With the popularization of the Internet and the development of Internet of Things technology, the cyber-physical system has become an emerging research tool. It can closely connect physical information and digital information through the integration of information and communication technologies, realize the intelligent management and optimization of data, items, and resources, and then achieve efficiency improvement. If the cyber-physical system is applied to the intelligent optimization of the three-dimensional structure design of the lining of a long-distance water conveyance tunnel, it can not only solve the cumbersome work of scheme comparison in the traditional method but also achieve efficient data optimization. Consider using the cyber-physical system to conduct intelligent optimization of the three-dimensional structure design of the lining of a long-distance water conveyance tunnel. Summary of the Invention

[0005] To solve the problems existing in the prior art, the present invention provides an intelligent optimization method and system for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the cyber-physical system. This method is simple and practical and can efficiently and accurately analyze and calculate the bearing capacity, maximum deformation, and crack width of the lining according to the existing data, and realize the inversion optimization of the lining structure design parameters S i (including tunnel shape parameters, lining thickness, construction support parameters (bolt row spacing, spacing, and rock penetration depth)) to obtain the optimal design parameter set.

[0006] To achieve the above object, the present invention provides the following solutions:

[0007] Intelligent optimization method for three-dimensional structure design of lining of long-distance water conveyance tunnel based on cyber-physical network, the method comprising:

[0008] S1: Establish a three-dimensional geometric model of the lining of the long-distance water conveyance tunnel;

[0009] S2: According to the three-dimensional geometric model of the lining of the long-distance water conveyance tunnel, establish a physical model for structural mechanics analysis considering the stress of surrounding rock and the construction excavation process;

[0010] S3: According to the physical model for structural mechanics analysis, establish a multi-parameter cyber-physical network model and train it;

[0011] S4: According to the trained multi-parameter cyber-physical network model, with the bearing capacity, maximum deformation and crack width of the lining as the constraint boundaries and the project cost as the optimization objective, optimize to obtain a set of lining structure design parameters, wherein the lining structure design parameters include: tunnel shape parameters, lining thickness, construction support parameters.

[0012] Preferably, in the S1, the three-dimensional geometric model of the lining of the long-distance water conveyance tunnel includes:

[0013] The three-dimensional terrain of the mountain body within the preset range of the tunnel, geological stratification, fault position, strike, dip, dip angle and thickness, and the spline curve of the three-dimensional tunnel line layout.

[0014] Preferably, in the S2, the physical model for structural mechanics analysis includes: equilibrium equation, physical equation, geometric equation, wherein the equilibrium equation needs to consider the initial in-situ stress, and the excavation and support process is reflected by the time-history change of the constitutive parameters in the physical equation.

[0015] Preferably, in the S3, the multi-parameter cyber-physical network model includes:

[0016] The input is tunnel shape parameters, lining thickness, construction support parameters;

[0017] Sampling points are set for calculating the objective function of the neural network, wherein no less than 20 sampling points are set in the circumferential direction of the lining, no less than 5 sampling points are set in the thickness direction, and the axial sampling point spacing is fixed at 1 / 2 of the excavation footage;

[0018] The output is the displacement in the x, y, and z directions of the sampling points;

[0019] Construct a plurality of cyber-physical networks with the same network structure for predicting the displacement of the sampling points. The hidden layer in the network structure is 10 fully connected layers, with 20,000 neurons in each layer; the objective function for evaluating the displacement prediction error of the sampling points is:

[0020]

[0021] where \(i\) is the sampling point number; \(N\) is the total number of sampling points; \(\omega\) E , \(\omega\) P , \(\omega\) A , \(\omega\) B are the weights corresponding to the elastic component, plastic component, bolt displacement component, initial condition component, and boundary condition component in the objective function, and are taken as 1, 1, 100, 100 respectively; \(M\) (i) is the transformation matrix based on the excavation process. When the sampling point \(i\) is in the excavated area, \(M\) (i) = 0, otherwise, \(M\) (i) = 1; are the elastic component, plastic component, bolt displacement component, initial condition component, and boundary condition component of the objective function respectively.

[0022] Preferably, in the step S4, the set of optimized lining structure design parameters includes:

[0023] Input the set of design parameters \(S\) i The value range of is the interval specified by the user inside;

[0024] Randomly generate \(N\) sets of design parameter sets \(S\) 1 , \(S\) 2 , \(S\) 3 ,... \(S\) N , and the random numbers are uniformly distributed in the interval, and \(N\) takes values between 20 and 30;

[0025] Randomly generate an iteration increment of a set of design parameters where is the maximum value of the increment of \(S\) input by the user i ;

[0026] Use the multi-parameter physical information grid model to calculate the objective function of each design parameter set. The objective function is:

[0027] \(F = C+\lambda\) 1 \(\cdot(\max(0,\delta\) max -\delta\) L )) 2 +\lambda\) 2 (\max(0,w\) max -w\) L )) 2

[0028] where \(\lambda\) 1 and \(\lambda\) 2 are the parameters of the penalty function, and a relatively large positive number is input by the user; \(\delta\) max and \(\delta\)L They are respectively the maximum deformation and allowable deformation of the tunnel under the current design parameter set calculated according to the physical information network; w max and w L are the maximum width and allowable width of the tunnel gap calculated; C is the cost function of the project;

[0029] According to the objective function, the particle swarm optimization algorithm is used to obtain the optimal design parameter set.

[0030] The present invention also provides an intelligent optimization system for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the physical information network. The system is used to implement any one of the above methods. The system includes: an interactive interface and data management module, a calculation and prediction module, and a lining structure optimization module;

[0031] The interactive interface and data management module is used to input and output the design parameters of the tunnel lining structure, import the three-dimensional geometric model of the long-distance water conveyance tunnel lining, and set the optimization objective;

[0032] The calculation and prediction module is used to solve the physical model of structural mechanics analysis, construct and train a multi-parameter physical information network model, and solve the objective function;

[0033] The lining structure optimization module optimizes the objective function through the particle swarm optimization algorithm to obtain the optimized design parameter set.

[0034] Preferably, the interactive interface and data management module includes: a parameter input sub-module, a model import sub-module, and an optimization result storage sub-module;

[0035] The parameter input sub-module is used for the user to input the initial design parameters of the tunnel lining structure and the cost in the optimization objective. The initial design parameters include the lining thickness range, the bolt row spacing range, and the tunnel shape parameters;

[0036] The model import sub-module is used to import the three-dimensional geometric model of the long-distance water conveyance tunnel lining and the geological conditions, including the tunnel line spline curve, the three-dimensional terrain of the mountain, and the geological stratification data;

[0037] The optimization result storage sub-module is used to store the intermediate results and the final design parameter set during the optimization process, and supports export in CSV or JSON format.

[0038] Preferably, the calculation and prediction module includes: a physical constraint analysis sub-module, an objective function calculation sub-module, and an error evaluation sub-module;

[0039] The physical constraint analysis sub-module is used to embed the equilibrium equations, physical equations, and geometric equations related to the tunnel as the constraint conditions of the physical information network. The equilibrium equations need to consider the time history changes of the initial in-situ stress and construction support.

[0040] The objective function calculation sub-module is used to calculate the objective function of the sampling points based on the physical information network. The constraints of the objective function are: the normal displacement of the calculation model boundary is 0, and the stress on the inner side of the tunnel is equal to the internal water pressure.

[0041] The error evaluation sub-module is used to calculate the residual of the control equation of the prediction result of the physical information network. The calculation end condition is that the residual is less than 1N or the difference between two iterations is less than 0.1%.

[0042] Preferably, the lining structure optimization module includes: a parameter constraint sub-module, an optimization objective generation sub-module, and a dynamic optimization feedback sub-module.

[0043] The parameter constraint sub-module is used to set the boundary conditions and optimization constraints of the design parameters.

[0044] The optimization objective generation sub-module is used to generate the target optimization function. The objective function comprehensively considers the constraint conditions of the maximum deformation of the tunnel, the crack width, and the project cost.

[0045] The dynamic optimization feedback sub-module is used to dynamically update the weight distribution of the sampling points during the optimization process. The dynamic adjustment rate of the sampling points is 5%, so as to accelerate the convergence to the global optimal solution.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] The intelligent optimization method and system for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the physical information network according to the present invention optimize the design parameters of the tunnel lining structure by combining artificial intelligence algorithms, avoid the complicated calculation and analysis work in the traditional method, improve the engineering implementation efficiency, can automatically evaluate the advantages and disadvantages of various design schemes during the design process, and can be continuously updated and optimized with the development of technology and the accumulation of data to ensure the reliability and safety of the design, providing the necessary technical support for the three-dimensional structure design of the lining of modern long-distance water conveyance tunnels. Description of the Drawings

[0048] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0049] Figure 1Schematic diagram of the intelligent optimization method for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the cyber-physical network according to an embodiment of the present invention;

[0050] Figure 2 Three-dimensional geometric model diagram of the long-distance water conveyance tunnel according to an embodiment of the present invention;

[0051] Figure 3 Schematic diagram of the calculation process for constructing a cyber-physical network model according to an embodiment of the present invention;

[0052] Figure 4 Schematic diagram of the particle swarm optimization algorithm for the lining structure design parameters according to an embodiment of the present invention;

[0053] Figure 5 Schematic diagram of the modules of the intelligent optimization system according to an embodiment of the present invention. Detailed implementation manners

[0054] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0055] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0056] Embodiment 1

[0057] The intelligent optimization method for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the cyber-physical network proposed in this embodiment is as Figure 1 shown. It includes: S1: According to the known engineering design data, establish a three-dimensional geometric model of the lining of the long-distance water conveyance tunnel, and the model contains the necessary information required for calculation and analysis; S2: Establish a structural mechanics analysis physical model considering the surrounding rock stress and the construction excavation process, and the model contains the equations used for analysis; S3: Establish a multi-parameter cyber-physical network model and perform training, and the model contains network input information and sampling point information; S4: With the lining bearing capacity, maximum deformation, and crack width as the constraint boundaries and the project cost as the optimization goal, optimize to obtain the optimal parameter design set of the lining structure design parameter S i The lining structure design parameter S i shall include the tunnel shape parameter, lining thickness, and construction support parameters (bolt row spacing, spacing, and rock penetration depth). The specific steps are as follows:

[0058] S1. Establish a geometric model of the long-distance water conveyance tunnel;

[0059] First, based on the basic design data of the tunnel project and the results of relevant tests, use 3D modeling software such as CATIA, AutoCAD, Revit, etc. to establish a 3D geometric model of the lining of the long-distance water conveyance tunnel, such as Figure 2 . The tunnel cross-section is distinguished according to the surrounding rock grade of the tunnel body. In this example Figure 2 , the surrounding rock is divided into Class Ⅲ, Class Ⅳ and Class Ⅴ surrounding rock. The initial support of the tunnel cross-section includes C25 shotcrete, Φ8 steel mesh with a size of 25×25 cm, and anchor plate with a size of 15×15×1 cm. In the secondary lining, C30 concrete is used for the invert, arch and side walls of the tunnel. The thicknesses of the reinforced concrete layers corresponding to Class Ⅲ, Class Ⅳ and Class Ⅴ surrounding rock are 80 cm, 90 cm and 100 cm respectively. In addition, a deformation allowance needs to be reserved, which is divided into 5 cm, 7 cm and 12 cm according to the types of surrounding rock.

[0060] S2. Establish a physical model of structural mechanics analysis considering the stress of the surrounding rock and the construction excavation process;

[0061] The established physical model of structural mechanics analysis includes the equilibrium equation (the equilibrium state of the tunnel structure under the action of loads):

[0062]

[0063] where σ is the stress tensor, f is the body force, is the gradient operator.

[0064] Physical equation (the stress-strain relationship of materials, that is, the constitutive relationship):

[0065] σ = E·ε (2)

[0066] where E is the elastic tensor of the material, and ε is the strain tensor.

[0067] Geometric equation (the relationship between displacement and strain):

[0068]

[0069] where u is the displacement vector.

[0070] Among them, the initial in-situ stress needs to be considered in the equilibrium equation, and the excavation and support process is reflected by the time-history change of the constitutive parameters in the physical equation.

[0071] S3. Establish a multi-parameter physics-informed neural network model and conduct training;

[0072] Establish a multi-parameter physics-informed neural network model for predicting the displacement of sampling points. The hidden layer in its network structure is 10 fully connected layers, with 20,000 neurons in each layer. The training calculation process is as Figure 3As shown below. First, input the initial design parameters of the tunnel lining structure, including tunnel shape parameters, lining thickness, and construction support parameters (bolt row spacing, spacing, and rock penetration depth); construct a neural network structure (DNN). Among them, the neural network has 10 fully connected layers, with 20,000 neurons in each layer, and the network weights are initialized later; after completing the above steps, use the neural network structure (DNN) to calculate displacement and stress. Among them, the displacement u is the output of the neural network as follows:

[0073]

[0074] In the formula, x, y, and z are coordinates, and the stress σ is the partial derivative of the displacement with respect to the input coordinates:

[0075]

[0076] In the formula, E is the elastic tensor, and the gradient of the displacement and are the partial derivatives of the output of the neural network with respect to the coordinates.

[0077] Subsequently, calculate the objective function. It should be noted that when setting the sampling points for calculating the objective function of the neural network, there are no less than 20 sampling points in the circumferential direction of the lining, no less than 5 sampling points in the thickness direction, and the axial sampling point spacing is fixed at 1 / 2 of the excavation footage; after obtaining the objective function, it is necessary to calculate the final error and compare it with the set error threshold. If the absolute value of e rr is less than 1 kN, the calculation results are output. If the requirements are not met, the network weights need to be updated and the calculation is performed again until the final error meets the requirements, at which point the weight update is stopped and the calculation results are output. The calculation results include the displacements and stresses in the x, y, and z directions of the sampling points.

[0078] The objective function for evaluating the displacement prediction error of the sampling points is:

[0079]

[0080] In the formula, i is the sampling point number; N is the total number of sampling points; ω E , ω P , ω A , ω B are the weights corresponding to the elastic component, plastic component, bolt displacement component, initial condition component, and boundary condition component in the objective function, and are taken as 1, 1, 100, and 100 respectively; M (i) is the transformation matrix based on the excavation process. When the sampling point i is in the excavated area, M (i) =0, otherwise, M (i) =1; They are the elastic component, plastic component, bolt displacement component, initial condition component, and boundary condition component of the objective function, and are defined as:

[0081]

[0082] In the formula, E is the elastic modulus, and σ is the stress at the sampling point calculated according to the displacement u predicted by the neural network. f (i) is the body force component caused by self-weight, and f g is the body force component caused by the prestress of the bolt; A

[0083]

[0084] In the formula, λ θ , F y are respectively the plastic multiplier and yield function in the Mohr-Coulomb elastic-plastic constitutive model, and are expressed as functions of the displacement u during the solution calculation process;

[0085]

[0086] In the formula, are respectively the elastic objective function component calculated according to the surrounding rock parameters, the plastic objective function component obtained according to the surrounding rock parameters, the elastic objective function component obtained according to the bolt parameters, and the plastic objective function component obtained according to the bolt parameters. The value of the objective function is obtained from equations (6) and (7);

[0087]

[0088] In the formula, u (i) and are respectively the displacement predicted by the neural network and the true displacement of the boundary condition.

[0089] S4. Optimize to obtain the set of lining structure design parameters S i ;

[0090] Optimize to obtain the set of lining structure design parameters S i The set process is as shown in Figure 4 and mainly includes the following steps: First, initialize the design parameter set in the particle swarm optimization algorithm (PSO); input the design parameter set S i ​Determine the solution space of the design parameter set according to the value range; randomly generate N groups of design parameter sets within the value range; randomly determine the initial position and initial update speed of the design parameters; calculate the objective function of each design parameter set through the physics-informed grid model. After obtaining the objective function, the difference needs to be calculated with the previously obtained objective function for iteration. When the difference between any two iterations of the objective function reaches the minimum (less than a certain critical threshold) or tends to a certain stable value, the design parameters of the last iteration are output, that is, the optimized optimal parameter solution; otherwise, it is necessary to update the optimal position and optimal speed of each group of design parameters and update the historical optimal values and positions of all design parameters, and then repeat the iteration process until the optimal design parameter solution is output when the end condition is satisfied.

[0091] Specifically, 4a) Input the design parameter set S i The value range of is the interval specified by the user inside;

[0092] 4b) Randomly generate N groups of design parameter sets S 1 , S 2 , S 3 ,... S N , and the random numbers are uniformly distributed within the interval, and N is taken between 20 and 30;

[0093] 4c) Randomly generate an iterative increment of a group of design parameters where is the maximum value of the increment of S input by the user i ;

[0094] 4d) Use the physics-informed grid model in step (S3) to calculate the objective function of each design parameter set. The objective function is as follows:

[0095] F = C + λ 1 ·(max(0, δ max - δ L )) 2 + λ 2 ·(max(0, w max w L )) 2 (11)

[0096] In the formula, λ 1 and λ 2 are the parameters of the penalty function, and large positive numbers are input by the user; δ max and δ L are the maximum deformation and allowable deformation of the tunnel under the current design parameter set calculated according to the physics-informed network respectively; w max and w Lis the maximum width and allowable width of the tunnel gap calculated; C is the project cost function, defined as follows:

[0097] C = C m (V) + C const. (V, T) + C r (V, σ) (12)

[0098] where C is the material cost of the lining, which is a function of the lining volume V obtained directly from the geometric model; C const. is the user-defined tunnel construction cost function determined by the geometric cross-section type T of the tunnel and the lining volume V; C r is the reinforcement cost function of the lining, calculated based on the internal force calculated by the physical information network and the reinforcement ratio obtained by the stress diagram method;

[0099] 4e) The particle swarm optimization algorithm is adopted, with formula (11) as the objective function, to obtain the optimal design parameter set as follows: the net width is 9.3 m, the maximum net height is 7.4 m, and the crown curvature radius is 12.7 m; for the lining parameters, the reinforced concrete lining is selected in the scheme, the lining thickness is optimized to be 0.47 m, and the concrete strength grade is selected as C25; in terms of construction support, shotcrete is used for the initial support, and its thickness is set to 0.15 m.

[0100] Example 2

[0101] As Figure 5 shown, the present invention also provides an intelligent optimization system for the three-dimensional structure design of the lining of a long-distance water conveyance tunnel based on the physical information network. The system is used to implement any one of the above methods. The system includes: an interactive interface and data management module, a calculation and prediction module, and a lining structure optimization module;

[0102] The interactive interface and data management module is used to input and output the design parameters of the tunnel lining structure, import the three-dimensional geometric model of the long-distance water conveyance tunnel lining, and set the optimization objectives;

[0103] The calculation and prediction module is used to solve the physical model of structural mechanics analysis, construct and train the multi-parameter physical information network model, and solve the objective function;

[0104] The lining structure optimization module optimizes the objective function through the particle swarm optimization algorithm to obtain the optimized design parameter set.

[0105] In this embodiment, the interactive interface and data management module includes: a parameter input sub-module, a model import sub-module, and an optimization result storage sub-module;

[0106] The parameter input sub-module is used for users to input the initial design parameters of the tunnel lining structure and the cost in the optimization objective. The initial design parameters include the lining thickness range (0.5m to 1.5m), the bolt row spacing range (0.5m to 2.0m), and the tunnel shape parameters;

[0107] The model import sub-module is used to import the three-dimensional geometric model of the long-distance water conveyance tunnel lining and the geological conditions, including the tunnel line spline curve, the three-dimensional mountain terrain, and the geological stratification data;

[0108] The optimization result storage sub-module is used to store the intermediate results and the final set of design parameters during the optimization process, and supports export in CSV or JSON format.

[0109] In this embodiment, the calculation and prediction module includes: a physical constraint analysis sub-module, an objective function calculation sub-module, and an error evaluation sub-module;

[0110] The physical constraint analysis sub-module is used to embed the balance equations, physical equations, and geometric equations related to the tunnel as the constraint conditions of the physical information network. The balance equation needs to consider the time-varying initial ground stress and construction support;

[0111] The objective function calculation sub-module is used to calculate the objective function of the sampling points based on the physical information network. The constraints of the objective function include the maximum tunnel deformation (less than 20mm), the crack width (less than 0.2mm), and the material stress distribution conforming to the design specifications. Specifically: the normal displacement of the calculation model boundary is 0, and the stress on the inner side of the tunnel is equal to the internal water pressure;

[0112] The error evaluation sub-module is used to calculate the residual of the control equation of the prediction result of the physical information network. The error calculation uses the mean square error (MSE), and the error threshold is less than 0.1mm. The calculation end condition is that the residual is less than 1N or the difference between two iterations is less than 0.1%.

[0113] In this embodiment, the lining structure optimization module includes: a parameter constraint sub-module, an optimization objective generation sub-module, and a dynamic optimization feedback sub-module;

[0114] The parameter constraint sub-module is used to set the boundary conditions and optimization constraints of the design parameters;

[0115] The optimization objective generation sub-module is used to generate the target optimization function, and the objective function comprehensively considers the constraint conditions of the maximum tunnel deformation, the crack width, and the project cost;

[0116] The dynamic optimization feedback sub-module is used to dynamically update the weight allocation of the sampling points during the optimization process. The dynamic adjustment rate of the sampling points is 5%, so as to accelerate the convergence to the global optimal solution.

[0117] This module can support customizing the lining geometry, allowing the setting of the physical and mechanical properties of different lining materials, and supporting the setting of different types of internal and external loads and fixed and variable boundary conditions. The module optimizes the lining structure design parameter set according to the objective function by using the particle swarm algorithm. The specific optimization steps are as follows: First, input the value range of the lining structure design parameter set S i ; Second, randomly generate N groups of design parameter sets within the value range; Then calculate the objective function of each design parameter set; Finally, use the particle swarm optimization algorithm to optimize the objective function to obtain the optimal lining structure design parameter set.

[0118] The embodiments described above are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. An intelligent optimization method for the three-dimensional structural design of long-distance water tunnel lining based on physical information network, characterized in that: The method comprises: S1: Establish a 3D geometric model of the lining of a long-distance water transfer tunnel; S2: Based on the three-dimensional geometric model of the lining of the long-distance water transfer tunnel, a physical model for structural mechanics analysis considering the surrounding rock stress and the construction excavation process is established; S3: Based on the physical model of structural mechanics analysis, a multi-parameter physical information network model is established and trained; S4: Based on the trained multi-parameter physical information network model, with lining bearing capacity, maximum deformation and crack width as constraint boundaries and project cost as optimization target, the set of lining structure design parameters is optimized, where the lining structure design parameters include: tunnel shape parameters, lining thickness, and construction support parameters.

2. The method according to claim 1, characterized in that In S1, the three-dimensional geometric model of the lining of the long-distance water conveyance tunnel includes: The three-dimensional topography, geological stratification, fault location, strike, dip, inclination and thickness of the mountain within the preset range of the tunnel, and the three-dimensional line layout spline curve of the tunnel.

3. The method according to claim 1, characterized in that In S2, the physical model of structural mechanics analysis includes: equilibrium equations, physical equations, and geometric equations, wherein the equilibrium equations need to consider the initial ground stress, and the excavation and support processes are reflected by the time-course changes of constitutive parameters in the physical equations.

4. The method according to claim 1, characterized in that: In S3, the multi-parameter physical information network model includes: The inputs are tunnel shape parameters, lining thickness, and construction support parameters; Sampling points are set to calculate the objective function of the neural network, where no less than 20 sampling points are set in the lining circumferential direction, no less than 5 sampling points are set in the thickness direction, and the axial sampling point spacing is fixed to 1 / 2 of the excavation footage; The output is the displacement of the sampling point in the x, y, and z directions; Construct multiple physical information networks with the same network structure to predict the displacement of sampling points. The hidden layer in the network structure is 10 fully connected layers, with 20,000 neurons in each layer; the objective function used to evaluate the displacement error of sampling point displacement prediction is for: Where i is the sampling point number; N is the total number of sampling points; ω E ,ω P ,ω A ,ω B are the weights corresponding to the elastic component, plastic component, anchor displacement component, initial condition component, and boundary condition component in the objective function, which are 1, 1, 100, and 100 respectively; M (i) is the transformation matrix based on the excavation process. When the sampling point i is located in the excavated area, M (i) =0, otherwise, M (i) =1; They are the elastic component, plastic component, anchor displacement component, initial condition component, and boundary condition component of the objective function respectively.

5. The method according to claim 1, characterized in that In S4, the set of lining structure design parameters obtained by optimization includes: Input design parameter set S i The value range is the interval specified by the user Inside; Randomly generate N sets of design parameter sets S1, S2, S3, ...S within the value range N , the random number is used in Evenly distributed within the interval, N value is between 20 and 30; Randomly generate a set of iterative increments of design parameters in S is the user input i The maximum value of the increment; The multi-parameter physical information grid model is used to calculate the objective function of each design parameter set. The objective function is: F=C+λ1·(max(0,δ max -d L )) 2 +λ2·(max(0,w max -w L )) 2 Where λ1 and λ2 are the parameters of the penalty function, which are large positive numbers input by the user; δ max and δ L are the maximum deformation and allowable deformation of the tunnel under the current design parameter set calculated according to the physical information network; w max and w L is the calculated maximum width and allowable width of the tunnel gap; C is the cost function of the project; According to the objective function, the particle swarm algorithm is used to obtain the optimal set of design parameters.

6. An intelligent optimization system for three-dimensional structural design of long-distance water tunnel lining based on physical information network, the system is used to implement the method described in any one of claims 1 to 5, characterized in that: The system includes: an interactive interface and data management module, a calculation and prediction module, and a lining structure optimization module; The interactive interface and data management module are used to input and output tunnel lining structure design parameters, import the three-dimensional geometric model of the long-distance water transfer tunnel lining, and set optimization targets; The calculation and prediction module is used to solve the physical model of structural mechanics analysis, build and train a multi-parameter physical information network model, and solve the objective function; The lining structure optimization module optimizes the objective function through a particle swarm optimization algorithm to obtain an optimized design parameter set.

7. The system according to claim 6, characterized in that The interactive interface and data management module includes: a parameter input submodule, a model import submodule, and an optimization result storage submodule; The parameter input submodule is used for the user to input the initial design parameters of the tunnel lining structure and the cost in the optimization target, wherein the initial design parameters include the lining thickness range, the anchor spacing range and the tunnel shape parameters; The model import submodule is used to import the 3D geometric model of the lining of the long-distance water transfer tunnel and the geological conditions, including the tunnel line spline curve, the 3D terrain of the mountain and the geological layering data; The optimization result storage submodule is used to store the intermediate results and the final design parameter set in the optimization process, and supports exporting in CSV or JSON format.

8. The system according to claim 6, characterized in that The calculation and prediction module includes: a physical constraint analysis submodule, an objective function calculation submodule, and an error evaluation submodule; The physical constraint analysis submodule is used to embed tunnel-related equilibrium equations, physical equations and geometric equations as constraint conditions of the physical information network. The equilibrium equation needs to consider the time-course changes of initial ground stress and construction support; The objective function calculation submodule is used to calculate the objective function of the sampling point based on the physical information network. The constraints of the objective function are: the normal displacement of the calculation model boundary is 0, and the stress inside the tunnel is equal to the internal water pressure; The error evaluation submodule is used to calculate the residual of the control equation of the physical information network prediction result, and the calculation end condition is that the residual is less than 1N or the difference between two iterations is less than 0.1%.

9. The system according to claim 6, characterized in that The lining structure optimization module includes: a parameter constraint submodule, an optimization target generation submodule, and a dynamic optimization feedback submodule; The parameter constraint submodule is used to set the boundary conditions and optimization constraints of the design parameters; The optimization target generation submodule is used to generate a target optimization function, which comprehensively considers the constraints of the maximum deformation of the tunnel, the crack width and the engineering cost; The dynamic optimization feedback submodule is used to dynamically update the weight distribution of sampling points during the optimization process, and the dynamic adjustment rate of the sampling points is 5%, thereby accelerating the convergence to the global optimal solution.

Images