A method for evaluating the health of a diversion tunnel structure based on a digital twin model

Abstract

The application discloses a kind of based on digital twinborn model's diversion tunnel structure health assessment method, it is related to water conservancy engineering health monitoring field, including: obtaining the geometric parameter and material parameter of diversion tunnel, constructs initial digital twinborn model;Carrying out numerical simulation including fluid-structure coupling effect, obtain the simulation physical response data of tunnel structure under preset load condition;Simulation physical response data are compared with actual monitoring data, and model error index is calculated;When model error index exceeds preset threshold, construct fitness function with minimizing model error index as target, iteratively optimize model key parameters using particle swarm optimization algorithm, obtain optimal parameter combination;Based on updated digital twinborn model, diversion tunnel structure health assessment is carried out.It realizes accurate simulation and health assessment to diversion tunnel structure response by high-precision digital twinborn model after parameter calibration.

Description

Technical Field

[0001] This invention relates to the field of health monitoring of water conservancy projects, and more specifically to a method for assessing the structural health of water diversion tunnels based on digital twin models. Background Technology

[0002] As a key structure in the water conveyance system of a water conservancy project, the water diversion tunnel is subjected to the coupled effects of multiple loads, such as internal water pressure, surrounding rock constraints, and nonlinearity of the lining material, during long-term operation. Its structural response is complex and has significant temporal and spatial dynamics. Accurately assessing the displacement, stress, and strain state of the tunnel lining is of great significance for ensuring regional water supply security and extending the service life of the project.

[0003] Existing methods for analyzing the structure of water diversion tunnels mostly rely on simplified mechanical assumptions, empirical formulas, or local point monitoring data. For example, traditional analytical methods cannot simultaneously consider fluid-structure interaction effects, the non-uniformity of surrounding rock constraints, and the constitutive relationship of lining materials, resulting in significant deviations between simulation results and actual structural responses. At the same time, conventional numerical simulation models, such as static analysis based on the finite element method, usually use idealized boundary conditions and initial material parameters, which cannot dynamically reflect the actual stress characteristics of the tunnel under different water levels, geological changes, and operating conditions.

[0004] Furthermore, while some studies in the field of tunnel engineering have attempted to introduce the concept of digital twins, most of them focus on safety early warning during construction or single performance indicators. For water diversion tunnels in operation, there is a lack of a systematic accuracy assessment and parameter calibration mechanism that can integrate multi-physics simulation and multi-source monitoring data. Existing digital twin models often ignore the uncertainties of key parameters such as the equivalent elastic support stiffness of the surrounding rock, the external load correction coefficient, and the elastic modulus of the lining concrete. This results in significant deviations between the model output and the measured data in terms of fitting local displacement, stress, or strain, making it difficult to meet the needs of long-term structural health management.

[0005] Therefore, how to design a health assessment method for water diversion tunnels based on digital twin models, so as to achieve accurate assessment of the health status of tunnel structures and the structural response under future loads, is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a method for structural health assessment of water diversion tunnels based on digital twin models. It aims to solve the problems that existing traditional analysis methods are unable to accurately capture the complex structural response of water diversion tunnels under the coupling of multiple load sources, and lack a system parameter calibration mechanism. By constructing an automatically calibrated digital twin model, high-precision simulation of tunnel structure displacement, stress and strain can be achieved, thereby providing a reliable basis for operational status assessment, abnormal area identification and response prediction under future loads.

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] A method for structural health assessment of water diversion tunnels based on digital twin models includes the following steps: S1. Obtain the geometric and material parameters of the water diversion tunnel and construct an initial digital twin model; S2. Based on the initial digital twin model, perform numerical simulations including fluid-structure interaction effects to obtain simulated physical response data of the tunnel structure under preset load conditions; including: The pressure distribution on the tunnel inner wall obtained from the steady-state flow field sub-model in the initial digital twin model is used as a surface load and is unidirectionally transferred to the lining inner wall of the static structural sub-model to form a fluid pressure equivalent nodal load vector. The displacement, stress and strain distribution of the lining structure under the preset load conditions are obtained by solving the dynamic equilibrium equation of the structural domain. S3. Compare the simulated physical response data with the actual monitoring data to calculate the model error index; S4. When the model error index exceeds the preset threshold, a fitness function is constructed with the goal of minimizing the model error index, and the particle swarm optimization algorithm is used to iteratively optimize the key parameters of the model to obtain the optimal parameter combination. S5. Update the initial digital twin model based on the optimal parameter combination, and conduct a structural health assessment of the water diversion tunnel based on the updated digital twin model.

[0009] Preferably, S1 includes: A three-dimensional solid geometric model of a representative section of the water diversion tunnel is constructed based on the aforementioned geometric parameters; The three-dimensional solid geometric model is lightweighted by removing local redundant geometric details and regularizing the solid topology to obtain a computable three-dimensional geometric model. Based on the computable three-dimensional geometric model and the material parameters, a property model including a static structural sub-model and a steady-state flow field sub-model is constructed to obtain an initial digital twin model.

[0010] Preferably, in the static structural sub-model, the lining structure of the water diversion tunnel is taken as the analysis object, the lining material parameters are set, and constant earth pressure, hydrostatic distribution pressure and equivalent elastic support are applied as boundary conditions. The equilibrium equations of the static structural sub-model are:

[0011] in, The overall stiffness matrix of the structure. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. Let be the nodal displacement vector. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

[0012] Preferably, the constant earth pressure equivalent nodal load vector Equivalent nodal load vector of hydrostatic pressure They are represented as follows:

[0013]

[0014] in, For the unit shape function matrix transpose, and These are the surfaces acting under earth pressure and hydrostatic pressure, respectively. The constant earth pressure acting on the outer wall of the lining, It is the hydrostatic pressure.

[0015] Preferably, in the steady-state flow field sub-model, the computational domain is the lightweight tunnel cavity geometry, the inlet adopts the mass flow inlet boundary, the outlet adopts the mass flow outlet boundary, and the tunnel inner wall is set as a non-slip wall boundary. Solving the mass conservation equation for incompressible fluids and momentum conservation equation The pressure field distribution of the water flow inside the cave was obtained; among which, For Hamiltonian operators, For fluid velocity vector, For fluid density, For fluid pressure, For fluid dynamic viscosity, This is the gravitational acceleration vector.

[0016] Preferably, in S2, the displacement, stress, and strain distribution results under the preset load conditions are expressed as follows:

[0017] Where M is the mass matrix and C is the damping matrix. Let be the nodal displacement vector. and These are the acceleration and velocity vectors of the node, respectively. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

[0018] Preferably, S3 includes: At the arch top of multiple preset monitoring sections in the water diversion tunnel, actual monitoring data is collected by sensors. The actual monitoring data includes measured displacement values, measured stress values, and measured strain values. From the simulated physical response data, extract the simulated displacement value, simulated stress value, and simulated strain value at the same location as multiple preset monitoring sections; For each monitoring section, the displacement error, stress error, and strain error are calculated separately. Based on the displacement error, stress error, and strain error of all monitoring sections, a comprehensive model error index is calculated.

[0019] Preferably, in step S4, the fitness function is expressed as:

[0020] in, To monitor the total number of cross sections, , , Let be the absolute error of displacement, absolute error of stress, and absolute error of strain at the j-th cross-section, respectively. , , These are the corresponding weighting coefficients.

[0021] Preferably, in step S4, the iterative optimization of key model parameters using a particle swarm optimization algorithm includes: Initialize the particle swarm, and the position vector of the i-th particle at the t-th iteration is represented as:

[0022] in, The equivalent elastic support stiffness of the surrounding rock. This is the correction factor for external loads. The elastic modulus of the lining concrete; The position vector of each particle is used as the current parameter combination and substituted into the initial digital twin model. Steps S2 and S3 are executed to obtain the corresponding simulation physical response data and model error index, and the fitness function value of each particle is calculated. Update the historical best position of each particle based on the fitness function value. And the global optimal position of the entire particle swarm, gbest; The velocity and position of each particle are updated using the velocity and position update formulas of the particle swarm optimization algorithm until the termination condition is met, thus obtaining the optimal parameter combination.

[0023] Preferably, the velocity and position update formula is expressed as follows:

[0024]

[0025] in, and Let be the velocity and position of the i-th particle in generation t, respectively, and w be the inertial weight. and As a learning factor, and A random number in the range [0,1]. This represents the historical best position of the i-th particle. This represents the historical global optimal position of the entire particle swarm.

[0026] As can be seen from the above technical solution, compared with the prior art, the technical solution of the present invention has the following beneficial effects: 1. This method constructs an initial digital twin model comprising a static structural sub-model and a steady-state flow field sub-model, and introduces unidirectional fluid-structure interaction simulation. It can simultaneously consider the mechanical behavior of the lining structure under the coupled effects of multiple loads, including constant earth pressure, hydrostatic pressure distribution, tunnel water pressure, and equivalent elastic support of the surrounding rock. The obtained displacement, stress, and strain distribution results clearly reflect the non-uniform response characteristics of the tunnel along the axial and radial directions, providing a refined digital characterization method for the operational status of water diversion tunnels.

[0027] 2. Using measured displacement, stress, and strain data from preset monitoring sections, a fitness function that minimizes weighted errors was constructed based on model error indices. Particle swarm optimization algorithm was then used to iteratively optimize the equivalent elastic support stiffness of the surrounding rock, the external load correction coefficient, and the elastic modulus of the lining concrete. The calibrated digital twin model showed high consistency between the simulation results and measured data on multiple typical sections in terms of overall trends and local peak values, enhancing the model's ability to fit the actual structural response.

[0028] 3. Based on the calibrated high-precision digital twin model, a systematic assessment of the structural health status of the water diversion tunnel can be performed. By comparing the displacement, stress, and strain distribution output by the model with the structural safety threshold point by point, sections with abnormal responses can be identified. At the same time, by inputting the predicted future load conditions and re-executing the fluid-structure interaction simulation, the response changes and safety margin of the structure can be predicted, thereby supporting tunnel operation and maintenance decisions. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0030] Figure 1 A flowchart of a method for assessing the structural health of a water diversion tunnel based on a digital twin model, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a three-dimensional solid geometric model of a representative section of a water diversion tunnel provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the earth pressure model of the static model sub-model of the water diversion tunnel provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the hydrostatic pressure model of the static model sub-model of the water diversion tunnel provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the elastic support model of the static model sub-model of the water diversion tunnel provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the unidirectional transmission of fluid pressure to the lining structure provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of the structural response distribution under the initial parameter model provided in the embodiments of the present invention; Figure 8 This is a schematic diagram comparing cross-sectional twins and measured displacements provided in an embodiment of the present invention; Figure 9 This is a schematic diagram comparing cross-sectional twinning and measured strain provided in an embodiment of the present invention; Figure 10 This is a schematic diagram comparing cross-sectional twinning and measured stress provided in an embodiment of the present invention; Figure 11 This is a schematic diagram comparing the optimized cross-sectional twin with the measured displacement provided in an embodiment of the present invention; Figure 12 This is a schematic diagram comparing the optimized cross-sectional twin with the measured strain provided in an embodiment of the present invention. Figure 13 This is a schematic diagram comparing the optimized cross-sectional twin with the measured stress provided in an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] like Figure 1 As shown in the figure, this embodiment provides a method for structural health assessment of water diversion tunnels based on digital twin models, including the following steps: S1. Obtain the geometric and material parameters of the water diversion tunnel and construct an initial digital twin model; S2. Based on the initial digital twin model, perform numerical simulation including fluid-structure interaction effect to obtain the simulated physical response data of the tunnel structure under preset load conditions. S3. Compare the simulated physical response data with the actual monitoring data to calculate the model error index; S4. When the model error index exceeds the preset threshold, a fitness function is constructed with the goal of minimizing the model error index, and the particle swarm optimization algorithm is used to iteratively optimize the key parameters of the model to obtain the optimal parameter combination. S5. Update the initial digital twin model based on the optimal parameter combination, and conduct a structural health assessment of the water diversion tunnel based on the updated digital twin model.

[0033] It obtains the displacement, stress, and strain distribution of the lining under multi-source loads by constructing a digital twin model containing static structure and steady-state flow field and performing unidirectional fluid-structure interaction simulation. Then, based on the measured monitoring data, it uses a particle swarm optimization algorithm to iteratively calibrate the equivalent elastic support stiffness of the surrounding rock, the external load correction coefficient, and the elastic modulus of the lining concrete based on the fitness function that minimizes weighted error. This makes the simulation output highly consistent with the measured data in terms of overall trend and local peak values. The calibrated high-precision model can be reliably applied to the structural health status assessment and forward-looking prediction of water diversion tunnels.

[0034] The following provides further explanation of each step and key technical feature of the above method; In this embodiment, S1, the geometric and material parameters of the water diversion tunnel are obtained, and an initial digital twin model is constructed; including: A three-dimensional solid geometric model of a representative section of the water diversion tunnel is constructed based on the aforementioned geometric parameters; The three-dimensional solid geometric model is lightweighted by removing local redundant geometric details and regularizing the solid topology to obtain a computable three-dimensional geometric model. Based on the computable three-dimensional geometric model and the material parameters, a property model including a static structural sub-model and a steady-state flow field sub-model is constructed to obtain an initial digital twin model.

[0035] like Figure 2 As shown, in this embodiment, a three-dimensional solid geometric model is constructed based on the axis, cross-sectional dimensions, excavation radius, inner radius, and lining thickness of the representative section of the water diversion tunnel. This three-dimensional solid geometric model is used to characterize the spatial orientation, lining boundary, and internal cavity structure of the representative section of the water diversion tunnel, providing a unified geometric basis for subsequent material property assignment, boundary condition setting, flow field calculation, and structural response analysis.

[0036] Furthermore, in the static structural sub-model, the lining structure of the water diversion tunnel is taken as the analysis object, the lining material parameters are set, and constant earth pressure, hydrostatic distribution pressure and equivalent elastic support are applied as boundary conditions. like Figure 3 , Figure 4 , Figure 5 As shown, the static structural sub-model takes the lining structure as the analysis object, applies constant earth pressure and hydrostatic distribution pressure to the outer wall of the lining, and uses equivalent elastic support to characterize the constraint effect of the surrounding rock on the lining structure. Among them, constant earth pressure is used to simulate the continuous squeezing effect of the surrounding rock and overlying soil on the outer wall of the lining; hydrostatic distribution pressure is used to simulate the distribution effect of groundwater pressure on the outer wall of the lining as the water depth changes; equivalent elastic support is used to simulate the constraint effect of the surrounding rock on the radial displacement of the lining structure. Through the above boundary conditions, the static structural sub-model can reflect the stress state of the lining structure under the combined action of external load and surrounding rock constraint during the operation of the water diversion tunnel.

[0037] The equilibrium equations of the static structural sub-model are:

[0038] in, The overall stiffness matrix of the structure. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. Let be the nodal displacement vector. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

[0039] Furthermore, the equivalent nodal load vector of constant earth pressure Equivalent nodal load vector of hydrostatic pressure They are represented as follows:

[0040]

[0041] in, For the unit shape function matrix transpose, and These are the surfaces acting under earth pressure and hydrostatic pressure, respectively. The constant earth pressure acting on the outer wall of the lining, It is the hydrostatic pressure.

[0042] Furthermore, in the steady-state flow field sub-model, the lightweight tunnel cavity geometry is used as the computational domain, the inlet adopts the mass flow inlet boundary, the outlet adopts the mass flow outlet boundary, and the tunnel inner wall is set as a non-slip wall boundary. Solving the mass conservation equation for incompressible fluids and momentum conservation equation The pressure field distribution of the water flow inside the cave was obtained; among which, For Hamiltonian operators, For fluid velocity vector, For fluid density, For fluid pressure, For fluid dynamic viscosity, This is the gravitational acceleration vector.

[0043] In constructing the initial digital twin model, this embodiment focuses on a detailed model of a representative water diversion tunnel with a total length of approximately 2 km. The starting elevation of this representative section is 1675 m, the ending elevation is 1637 m, the longitudinal elevation difference is 38 m, the tunnel excavation radius is 2.7 m, the inner radius is 2.3 m, and the lining thickness is 0.4 m. First, using a 3D geometric modeling tool, the tunnel axis and cross-sectional profile were constructed based on the aforementioned geometric parameters, forming a 3D solid model reflecting the actual spatial morphology. Subsequently, this model was imported into the geometric processing module for lightweighting. While ensuring that the overall tunnel outline, cross-sectional dimensions, and lining boundaries did not change substantially, redundant local geometric details were removed, resulting in a computable 3D geometric model suitable for subsequent numerical analysis. Based on the lightweighted geometric model, an attribute model was constructed, including a static structural sub-model and a steady-state flow field sub-model. The static structural sub-model uses the lining as the analysis object and employs an equivalent linear elastic material; its key parameters are shown in Table 1. Table 1

[0044] Where density is taken The initial value of Young's modulus is Poisson's ratio is 0.18, and the compressive strength is... ultimate tensile strength This provides a reasonable basis for subsequent calibration. Reference range; The steady-state flow field sub-model uses the tunnel cavity as the computational domain, with the inlet set as the mass flow inlet and the outlet as the mass flow outlet. The tunnel wall is set as a no-slip wall boundary. By solving the mass and momentum conservation equations of the incompressible fluid, the pressure field distribution of the water flow inside the tunnel is obtained, providing the fluid load boundary for subsequent fluid-structure interaction analysis.

[0045] In this embodiment S2, based on the initial digital twin model, numerical simulation including fluid-structure interaction effects is performed to obtain simulated physical response data of the tunnel structure under preset load conditions; including: The pressure distribution on the tunnel inner wall obtained from the steady-state flow field sub-model in the initial digital twin model is used as a surface load and unidirectionally transferred to the lining inner wall of the static structural sub-model, forming an equivalent nodal load vector of fluid pressure. :

[0046] in, It is a fluid-structure interaction interface. For the pressure distribution on the inner wall, For the unit shape function matrix Transpose of; like Figure 6 As shown, the pressure distribution on the tunnel inner wall calculated by the steady-state flow field sub-model is extracted as a fluid side load and mapped to the lining inner wall of the structural sub-model through the fluid-structure interaction interface. This transfer method is unidirectional, that is, the fluid pressure acts on the lining structure, and the structural deformation does not update the flow field calculation domain in the reverse direction. Through this processing method, the fluid pressure is transformed into a surface load in the structural domain, thereby participating in the solution of the displacement, stress and strain response of the lining structure.

[0047] The displacement, stress, and strain distribution of the lining structure under preset load conditions are obtained by solving the dynamic equilibrium equations of the structural domain.

[0048] Where M is the mass matrix and C is the damping matrix. Let be the nodal displacement vector. and These are the acceleration and velocity vectors of the node, respectively. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

[0049] In the solution of the steady-state flow field sub-model, the inlet mass flow rate is set to 26300 kg / s, which corresponds to the full-load water delivery flow rate under the design conditions; the no-slip boundary condition on the inner wall ensures the velocity continuity between the fluid and the lining; the combination of key parameters used is shown in Table 2. Table 2

[0050] By solving the mass and momentum conservation equations of incompressible fluids using the finite volume method, the distribution characteristics of pressure decreasing linearly along the tunnel axis can be obtained, with a significant increase in pressure gradient at bends or diameter changes. Furthermore, a sequential solution strategy is adopted for unidirectional fluid-structure interaction: first, the flow field calculation is completed and the pressure at the inner wall nodes is output; then, the pressure load is mapped to the corresponding element surface of the structural model through shape function interpolation. The mass matrix M and damping matrix C in the structural dynamic equilibrium equations can be constructed based on the density of the lining material and the Rayleigh damping assumption. The stiffness matrix K includes the bending and shear resistance contributions of the lining itself. In actual solution, the transient terms are ignored, which can simplify the solution to a quasi-static analysis, thereby reducing the computational cost. like Figure 7 As shown, after completing the unidirectional fluid-structure interaction calculation, the initial digital twin model before the update can output the displacement distribution, stress distribution and strain distribution results of the lining structure. The above structural response distribution results are used to characterize the spatial response characteristics of the tunnel lining under the combined action of internal water pressure, external earth pressure, hydrostatic distribution pressure and equivalent elastic support of the surrounding rock, and serve as the simulation data source for subsequent monitoring section data extraction, model error calculation and parameter calibration.

[0051] In this embodiment, S3, the simulated physical response data is compared with the actual monitoring data to calculate the model error index; including: At the crown of multiple pre-set monitoring sections in the water diversion tunnel, actual monitoring data is collected by sensors. The actual monitoring data includes measured displacement values, measured stress values, and measured strain values. The multiple pre-set monitoring sections include at least a starting section, a middle section, and an ending section distributed along the tunnel axis. Sensors for collecting displacement, stress, and strain data are arranged at the crown of each monitoring section. From the simulated physical response data, extract the simulated displacement value, simulated stress value, and simulated strain value at the same location as multiple preset monitoring sections; For each monitoring section, the displacement error, stress error, and strain error are calculated separately. Based on the displacement error, stress error, and strain error of all monitoring sections, a comprehensive model error index is calculated.

[0052] The comprehensive model error indices include mean absolute error (MAE), root mean square error (RMSE), bias (BIAS), and coefficient of determination. :

[0053]

[0054]

[0055]

[0056] in, and Let be the simulated value and the measured value of the p-th sample point, respectively. The mean of the measured values ​​is n, and the total number of sample points is n; the sample points include displacement, stress or strain data at different monitoring sections.

[0057] In this step, the selection of monitoring sections follows the principle of reflecting the overall situation while taking into account local conditions: Section T1 is located at the tunnel starting point and is significantly affected by the transition section between the entrance and exit; Section T2 is located in the stable middle region; Section T3 is near the end point; vibrating wire strain gauges and displacement gauges are installed at the arch position of each section, and the measured data are continuously collected on an hourly basis, with the average value under stable operating conditions selected as the comparison benchmark; based on all sample points of the three sections, the comprehensive model error index is calculated, including mean absolute error (MAE), root mean square error (RMSE), bias (BIAS), and coefficient of determination (R²); specifically, calculations are performed based on the initial parameter combination, which includes: the equivalent elastic support stiffness of the surrounding rock. External load correction factor Elastic modulus of lining concrete Under this initial parameter combination, a unidirectional fluid-structure interaction simulation was performed based on the initial digital twin model. Simulated displacement, stress, and strain values ​​were extracted at the arch positions of the starting section (T1), the middle section (T2), and the ending section (T3), and compared with the measured displacement, stress, and strain values ​​collected by sensors at the same locations. For any type of response quantity among displacement, stress, and strain, the measured value at the m-th monitoring section was denoted as... The simulation value is The error is In this embodiment, the sample size n=3, corresponding to three monitoring sections T1, T2, and T3 respectively. The measured and simulated values ​​of the three monitoring sections are substituted into MAE, RMSE, BIAS, and R... 2 The calculation formula and the model accuracy evaluation index under the initial parameter combination are shown in Table 3 below; Table 3

[0058] Overall, the initial model fits displacement, stress, and strain well, with MAE and RMSE both within acceptable ranges, BIAS close to zero, and R² close to 1, indicating that the model can effectively capture the overall response characteristics of the representative tunnel section. However, the local deviation of section T1 also indicates that there is still room for improvement in the model's accuracy, and it is necessary to introduce a calibration mechanism to optimize the key parameters.

[0059] In embodiment S4, when the model error index exceeds a preset threshold, a fitness function is constructed with the goal of minimizing the model error index. A particle swarm optimization algorithm is then used to iteratively optimize the key parameters of the model to obtain the optimal parameter combination. The fitness function is expressed as:

[0060] in, To monitor the total number of cross sections, , , Let be the absolute error of displacement, absolute error of stress, and absolute error of strain at the j-th cross-section, respectively. , , These are the corresponding weighting coefficients; , , , , , These represent the simulated displacement, stress, and strain values ​​for the j-th cross-section, respectively. , , These are the corresponding measured values.

[0061] Furthermore, the particle swarm optimization algorithm is used to iteratively optimize the key parameters of the model, including: Initialize the particle swarm, and the position vector of the i-th particle at the t-th iteration is represented as:

[0062] in, The equivalent elastic support stiffness of the surrounding rock. This is the correction factor for external loads. The elastic modulus of the lining concrete; The position vector of each particle is used as the current parameter combination and substituted into the initial digital twin model. Steps S2 and S3 are executed to obtain the corresponding simulation physical response data and model error index, and the fitness function value of each particle is calculated. Update the historical best position of each particle based on the fitness function value. And the global optimal position of the entire particle swarm, gbest; The velocity and position of each particle are updated using the velocity and position update formulas of the particle swarm optimization algorithm until the termination condition is met, thus obtaining the optimal parameter combination.

[0063] Furthermore, the velocity and position update formulas are expressed as follows:

[0064]

[0065] in, and Let be the velocity and position of the i-th particle in generation t, respectively, and w be the inertial weight. and As a learning factor, and A random number in the range [0,1]. This represents the historical best position of the i-th particle. This represents the historical global optimal position of the entire particle swarm; the inertial weight w at this point can be updated according to the following formula: ,in As the initial inertia weight, To terminate the inertia weight, This represents the maximum number of iterations.

[0066] If all items in the model error index corresponding to the current global optimal position are less than or equal to the corresponding preset threshold, or if the number of iterations reaches the preset maximum number of iterations, then the current global optimal position gbest will be set to gbest. , and This is the optimal combination of parameters.

[0067] In this embodiment, the particle swarm optimization algorithm uses the equivalent elastic support stiffness of the surrounding rock. External load correction factor and the elastic modulus of lining concrete As parameters to be optimized, the position vector of each particle is represented as follows: In each iteration, the parameter combination corresponding to the particle is substituted into the digital twin model, and the unidirectional fluid-structure interaction simulation is re-executed. The fitness function value is calculated based on the displacement error, stress error, and strain error at three typical monitoring sections T1, T2, and T3. The smaller the fitness function value, the closer the simulation result is to the measured result under the parameter combination. The implementation parameters of the particle swarm optimization algorithm include: a particle swarm size of 30 particles, a maximum number of iterations of 50, an inertia weight w that decreases linearly from 0.9 to 0.4 to balance global search and local exploitation, a learning factor c1=c2=2.0, and random numbers r1 and r2 that are regenerated in each step. After multiple iterations and optimizations, the algorithm converged and obtained a set of optimal parameter combinations, as shown in Table 4. Table 4

[0068] Under this optimal parameter combination, the digital twin model was rerun and its accuracy evaluation index was calculated. The results are shown in Table 5. Table 5

[0069] Compared to the initial parameter model, the optimized model shows significant reductions in MAE and RMSE for displacement, stress, and strain, with BIAS closer to zero and R² further improved and approaching 1. This indicates that the PSO optimization algorithm can effectively improve the model's sensitivity to key parameters, enabling the digital twin model to have higher prediction accuracy and reliability.

[0070] In this embodiment, S5, the initial digital twin model is updated based on the optimal parameter combination, and the structural health assessment of the water diversion tunnel is performed based on the updated digital twin model.

[0071] Based on the updated digital twin model, the displacement, stress, and strain values ​​of each part of the tunnel lining can be calculated point by point and compared with preset allowable thresholds or historical health benchmarks. When the calculated value of a certain area exceeds the preset safety limit, it is determined to be an abnormal area. The system automatically marks the location and issues an early warning to guide on-site inspection and maintenance. At the same time, by changing the load conditions in the model, such as water level rise and changes in surrounding rock pressure, the structural response of the tunnel under different operating conditions in the future can be predicted, and the estimated values ​​of displacement, stress, and strain can be obtained, providing a quantitative basis for taking reinforcement measures or adjusting the operation plan in advance.

[0072] In addition, in order to intuitively demonstrate the prediction accuracy of the initial digital twin model and the model after calibration by the particle swarm optimization algorithm, this embodiment compares the simulation results and measured data of the arch positions of the starting section T1, the middle section T2, and the ending section T3. Furthermore, such as Figure 8 , Figure 9 , Figure 10 As shown, the comparison between twin data and measured data of the arch positions of three sections T1, T2, and T3 under the initial parameter model is presented. It can be seen that the overall trend of displacement, stress, and strain curves is basically consistent with the measured data. Sections T2 and T3 show good fit, with peak values ​​and variation trends highly consistent. However, there is a certain deviation in the displacement of section T1, indicating that the initial model's representation accuracy in local areas needs to be improved. Nevertheless, the overall trend of the three sections fits well, indicating that the model can capture the main mechanical response characteristics of the tunnel structure and has the basis for further optimization. Furthermore, such as Figure 11 , Figure 12 , Figure 13As shown, the results of comparing the twin data and measured data of the same three cross sections after calibration by the particle swarm optimization algorithm are presented. The optimized displacement, stress and strain curves are highly consistent with the measured data in terms of overall trend and local peak values. The displacement deviation of section T1 is significantly corrected, and the responses of sections T2 and T3 are more stable. The curve fitting results show that the parameter-optimized digital twin model can accurately reflect the variation law between cross sections, providing a high-precision digital basis for reliable assessment and forward-looking prediction of the structural health status of water diversion tunnels.

[0073] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0074] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for structural health assessment of water diversion tunnels based on digital twin models, characterized in that, Includes the following steps: S1. Obtain the geometric and material parameters of the water diversion tunnel and construct an initial digital twin model; S2. Based on the initial digital twin model, perform numerical simulations including fluid-structure interaction effects to obtain simulated physical response data of the tunnel structure under preset load conditions; including: The pressure distribution on the tunnel inner wall obtained from the steady-state flow field sub-model in the initial digital twin model is used as a surface load and is unidirectionally transferred to the lining inner wall of the static structural sub-model to form a fluid pressure equivalent nodal load vector. The displacement, stress and strain distribution of the lining structure under the preset load conditions are obtained by solving the dynamic equilibrium equation of the structural domain. S3. Compare the simulated physical response data with the actual monitoring data to calculate the model error index; S4. When the model error index exceeds the preset threshold, a fitness function is constructed with the goal of minimizing the model error index, and the particle swarm optimization algorithm is used to iteratively optimize the key parameters of the model to obtain the optimal parameter combination. S5. Update the initial digital twin model based on the optimal parameter combination, and conduct a structural health assessment of the water diversion tunnel based on the updated digital twin model.

2. The method for structural health assessment of water diversion tunnels based on digital twin models according to claim 1, characterized in that, S1 includes: A three-dimensional solid geometric model of a representative section of the water diversion tunnel is constructed based on the aforementioned geometric parameters; The three-dimensional solid geometric model is lightweighted by removing local redundant geometric details and regularizing the solid topology to obtain a computable three-dimensional geometric model. Based on the computable three-dimensional geometric model and the material parameters, a property model including a static structural sub-model and a steady-state flow field sub-model is constructed to obtain an initial digital twin model.

3. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 2, characterized in that, In the static structural sub-model, the lining structure of the water diversion tunnel is taken as the analysis object. The lining material parameters are set, and constant earth pressure, hydrostatic distribution pressure and equivalent elastic support are applied as boundary conditions. The equilibrium equations of the static structural sub-model are: in, The overall stiffness matrix of the structure. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. Let be the nodal displacement vector. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

4. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 3, characterized in that, The constant earth pressure equivalent nodal load vector Equivalent nodal load vector of hydrostatic pressure They are represented as follows: in, For the unit shape function matrix transpose, and These are the surfaces acting under earth pressure and hydrostatic pressure, respectively. The constant earth pressure acting on the outer wall of the lining, It is the hydrostatic pressure.

5. The method for structural health assessment of water diversion tunnels based on digital twin models according to claim 2, characterized in that, In the steady-state flow field sub-model, the lightweight tunnel cavity geometry is used as the computational domain, the inlet adopts the mass flow inlet boundary, the outlet adopts the mass flow outlet boundary, and the tunnel inner wall is set as a non-slip wall boundary. Solving the mass conservation equation for incompressible fluids and momentum conservation equation The pressure field distribution of the water flow inside the cave was obtained; among which, For Hamiltonian operators, For fluid velocity vector, For fluid density, For fluid pressure, For fluid dynamic viscosity, This is the gravitational acceleration vector.

6. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 1, characterized in that, In S2, the displacement, stress, and strain distribution results under the preset load conditions are expressed as follows: Where M is the mass matrix and C is the damping matrix. Let be the nodal displacement vector. and These are the acceleration and velocity vectors of the node, respectively. The stiffness matrix contributing to the equivalent elastic support of the surrounding rock. This is the equivalent nodal load vector of constant earth pressure. This is the equivalent nodal load vector for hydrostatic pressure.

7. The method for structural health assessment of water diversion tunnels based on digital twin models according to claim 1, characterized in that, S3 includes: At the arch top of multiple preset monitoring sections in the water diversion tunnel, actual monitoring data is collected by sensors. The actual monitoring data includes measured displacement values, measured stress values, and measured strain values. From the simulated physical response data, extract the simulated displacement value, simulated stress value, and simulated strain value at the same location as multiple preset monitoring sections; For each monitoring section, the displacement error, stress error, and strain error are calculated separately. Based on the displacement error, stress error, and strain error of all monitoring sections, a comprehensive model error index is calculated.

8. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 1, characterized in that, In S4, the fitness function is expressed as: in, To monitor the total number of cross sections, , , Let be the absolute error of displacement, absolute error of stress, and absolute error of strain at the j-th cross-section, respectively. , , These are the corresponding weighting coefficients.

9. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 1, characterized in that, In step S4, the iterative optimization of key model parameters using the particle swarm optimization algorithm includes: Initialize the particle swarm, and the position vector of the i-th particle at the t-th iteration is represented as: in, The equivalent elastic support stiffness of the surrounding rock. This is the correction factor for external loads. The elastic modulus of the lining concrete; The position vector of each particle is used as the current parameter combination and substituted into the initial digital twin model. Steps S2 and S3 are executed to obtain the corresponding simulation physical response data and model error index, and the fitness function value of each particle is calculated. Update the historical best position of each particle based on the fitness function value. And the global optimal position of the entire particle swarm, gbest; The velocity and position of each particle are updated using the velocity and position update formulas of the particle swarm optimization algorithm until the termination condition is met, thus obtaining the optimal parameter combination.

10. The method for structural health assessment of a water diversion tunnel based on a digital twin model according to claim 9, characterized in that, The velocity and position update formula is expressed as follows: in, and Let be the velocity and position of the i-th particle in generation t, respectively, and w be the inertial weight. and As a learning factor, and A random number in the range [0,1]. This represents the historical best position of the i-th particle. This represents the historical global optimal position of the entire particle swarm.

Images