High arch dam risk prediction method based on digital twinborn model

Abstract

The invention discloses a high arch dam risk prediction method based on a digital twinborn model, and relates to the field of high arch dam risk prediction, and the method comprises the steps: constructing a measurement point monitoring model; constructing a measuring point mapping model, and performing model training by using a dynamic monitoring loss function; constructing a digital twinborn model; constructing a deformation-stress prediction model; and constructing a damage risk analysis model. According to the method, the problems of large global state speculation error, insufficient security evaluation space consistency constraint, insufficient damage evaluation function and the like in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of high arch dam risk prediction, and in particular to a high arch dam risk prediction method based on a digital twin model. Background Art

[0002] High arch dams have been widely used in water resources projects around the world due to their excellent overload capacity, seismic resistance and light and tough dam structure. In recent years, China has built a number of super-high arch dams, such as Jinping I, Xiaowan, Baihetan and Wudongde, whose safety is crucial to water resources management in the basin. However, the complex structure, time-varying characteristics and spatial variability of high arch dams make structural health monitoring and risk assessment extremely challenging. At present, statistical models based on HST theory and shallow machine learning methods are mainstream data-driven means, but they are difficult to capture complex nonlinear relationships and have limited generalization capabilities. Deep learning (such as LSTM, GRU and GCN) improves deformation prediction capabilities, but it relies on limited measurement point data and is difficult to globally analyze the state of the dam. The mechanism-driven method is based on finite elements and can evaluate the safety of the dam from the perspective of physical mechanisms, but it has problems such as complex calculations and poor real-time performance. In addition, the impact of engineering uncertainty factors has not been fully quantified. Digital twin technology in the era of Industry 4.0 is expected to integrate the advantages of data-driven and mechanism-driven to achieve intelligent and precise structural health monitoring of high arch dams. However, the current technology is still in the exploratory stage, and there are challenges such as modeling accuracy, data fusion and real-time interaction. Summary of the invention

[0003] In view of the above-mentioned deficiencies in the prior art, the present invention provides a high arch dam risk prediction method based on a digital twin model, which solves the problems of the prior art such as large errors in global state speculation, insufficient spatial consistency constraints in safety assessment, and insufficient damage assessment function.

[0004] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is: a high arch dam risk prediction method based on a digital twin model, comprising the following steps:

[0005] S1: Construct a monitoring model for measuring points, use the environmental impact factors extracted based on the HST model as the input of the monitoring model for measuring points, and obtain the predicted value of the deformation process of the high arch dam;

[0006] S2: Build a measurement point mapping model and use the dynamic monitoring loss function to train the model;

[0007] S3: Build a digital twin model, input the predicted value of the high arch dam deformation process into the trained measurement point mapping model through the measurement point monitoring model, and obtain the three-dimensional global deformation situation;

[0008] S4: Construct a deformation-stress prediction model, use the principal component analysis method to reduce the dimension of the three-dimensional global deformation, and obtain the predicted values ​​of the major principal stress, intermediate principal stress and minor principal stress of the three-dimensional global deformation through the deformation-stress prediction model;

[0009] S5: Construct a damage risk analysis model, use the Monte Carlo Dropout method, simulate the prediction distribution of the model under different weight configurations, and perform random Dropout sampling to obtain the high arch dam failure risk prediction value, and complete the high arch dam risk prediction based on the digital twin model.

[0010] Furthermore, the monitoring model of the measuring point in S1 is:

[0011]

[0012] Among them, H 0 is the environmental feature input to the graph convolutional feature extractor, and is the combination of hydraulic load influence factor, temperature influence factor and time-effect influence factor at time t, W 0 is the trainable weight matrix, H (·) is the node's influence factor feature matrix, l is the number of node layers, σ is the nonlinear activation function, is the normalized degree matrix, A~ is the adjacency matrix with the influence factor of self-connection, W (·) is a trainable weight matrix.

[0013] Furthermore, the measurement point mapping model in S2 is:

[0014]

[0015] θ=(θ1,θ2)

[0016] in, is the predicted value of the measurement point mapping model, W is the learnable weight, b is the learnable bias, is the LSTM output state, is the input point of the sensor measurement value, f LSTM (·) is the LSTM network, θ1 is the parameter of the LSTM network, θ2 is the parameter of W and b, and θ is all the parameters of the measurement point mapping model.

[0017] Furthermore, the dynamic monitoring loss function L in S2 Digital twin Including the accuracy loss function L accuracy and the spatial variability loss function L space , the formula is:

[0018] L Digitaltwin =λ1L accuracy +λ2L space

[0019]

[0020] Among them, λ1 and λ2 are weight parameters, N is the total number of global calculation points, i is the i-th global calculation point, l is the loss function calculation method, and is the spatial gradient between the predicted value of the measurement point mapping model and the numerical simulation result, α is the loss balance parameter, is the loss of the finite element node corresponding to the sensor measurement point, is the loss of the sensor measuring point, is the loss of all finite element nodes.

[0021] Furthermore, the digital twin model in S3 is:

[0022]

[0023] in, is the predicted value of the digital twin at time t, f Synapse (·) is the measurement point mapping model, f Cell body (·) is the monitoring model of the measuring point, θ Digital twin are the pre-trained parameters of the measurement point mapping model, and It is a collection of hydraulic load influencing factors, temperature influencing factors and time influencing factors.

[0024] Furthermore, the deformation-stress prediction model in S4 is:

[0025]

[0026]

[0027] in, is the stress prediction value, f σ (·) is the deformation-stress prediction function, ω σ is the parameter of the deformation-stress prediction model, L σ is the deformation-stress prediction model loss, σ1, σ2 and σ3 are the calculated values ​​of major principal stress, intermediate principal stress and minor principal stress, δ Digital twin is the digital twin prediction value, ω is the parameter of the deformation-stress prediction model, is the deformed data after dimensionality reduction, PCA(·) is the dimensionality reduction operation, h t is the hidden state, o tis the memory state, r is a t The Bernoulli random variable of the same dimension has a probability of 1 of p, LSTM(·) is the LSTM framework, and h t-1 is the hidden state for the next moment, o t-1 is the memory state of the next moment, ⊙ represents element-by-element multiplication, Bernoulli(·) is Bernoulli randomness, and are the predicted values ​​of major principal stress, intermediate principal stress and minor principal stress.

[0028] Furthermore, the damage risk analysis model in S5 is:

[0029]

[0030] in, To destroy the i-th sampling result of the risk analysis model at time t, and are the major principal stress, intermediate principal stress and minor principal stress at time t, Perform random Dropout sampling for the deformation-stress prediction function, T is the number of random Dropout sampling, is the high arch dam failure risk prediction value, P(·) is the indicator function, is the value of the Drucker-Prager yield function of the entire domain, and n is the number of connections of the plastic unit.

[0031] The beneficial effects of the present invention are:

[0032] (1) The present invention constructs a digital twin model of a high arch dam based on the fusion of neural network and finite element method, realizes the mapping from point monitoring to global deformation, and further combines deep learning with finite element simulation to predict the stress distribution of the dam body. This method improves the prediction accuracy while avoiding the dependence of traditional finite element methods on a large number of physical parameters, reducing the consumption of computing resources caused by complex iterative calculations, and thus significantly improving the real-time performance and computing efficiency of the model.

[0033] (2) Compared with the existing digital twin methods, the present invention introduces spatial variability constraints and jump connection strategies, and verifies their effectiveness through ablation experiments. This optimization strategy can more accurately capture the spatial distribution characteristics of dam deformation, optimize the data transmission path, and avoid information attenuation in the deep network, thereby improving the stability and generalization ability of the prediction. In addition, the deformation-stress prediction model constructed based on dimensionality reduction technology realizes the accurate deduction of the global stress state of the dam. At the same time, the method supports the dynamic update of the model and can perform extrapolated predictions for unknown extreme conditions, which makes up for the lack of generalization ability of traditional deep learning methods when dealing with conditions beyond the distribution of the training set, thereby enhancing the adaptability of structural safety monitoring.

[0034] (3) The global plastic failure risk assessment method proposed in the present invention combines uncertainty quantification analysis, effectively addresses the multi-source uncertainty problem existing in the monitoring process of high arch dam structures, and realizes real-time assessment of the safety of the dam structure. 800,000 random sampling tests based on this method show that under different dropout rates, the probability of plastic failure of the dam body always remains within the safety threshold range, verifying the robustness of the present invention under complex working conditions.

[0035] (4) This invention breaks through the limitations of the traditional "point monitoring" mode, realizes the real-time perception, prediction and early warning of the deformation of the entire high arch dam, and can combine the mechanism methods such as plastic failure analysis to conduct real-time structural safety assessment. Compared with the existing numerical simulation and statistical monitoring methods, this method not only improves the prediction accuracy, computational efficiency and generalization ability, but also optimizes data security and real-time monitoring capabilities, providing an economical and efficient solution for the intelligent and refined safety management of high arch dams, and has broad engineering application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a high arch dam risk prediction method based on a digital twin model.

[0037] Figure 2 Construct a flow chart for the digital twin model of high arch dams.

[0038] Figure 3 This is the layout diagram of the deformation monitoring points of DG Dam.

[0039] Figure 4 A process line diagram for monitoring the measured values ​​at the measuring point.

[0040] Figure 5 Comparison chart of mesh division and partitioning of finite element model.

[0041] Figure 6 This is a comparison chart of the accuracy of the proposed model, the finite element model and the existing model.

[0042] Figure 7 Comparison chart of the accuracy of the proposed model and the ablation model.

[0043] Figure 8 This is a comparison chart of the spatial variation rules of the three-dimensional model slices of the proposed model, the finite element model, and the existing model.

[0044] Fig. 9 Comparison diagram of the spatial variation of slices between the proposed model and the ablation model.

[0045] Fig.10 This is a comparison chart of the prediction accuracy between the measurement point monitoring model and the baseline model.

[0046] Fig.11 This is the analysis result diagram of the failure risk analysis model. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0048] like Figure 1 As shown, a high arch dam risk prediction method based on a digital twin model includes the following steps:

[0049] S1: Construct a monitoring model for measuring points, use the environmental impact factors extracted based on the HST model as the input of the monitoring model for measuring points, and obtain the predicted value of the deformation process of the high arch dam;

[0050] S2: Build a measurement point mapping model and use the dynamic monitoring loss function to train the model;

[0051] S3: Build a digital twin model, input the predicted value of the high arch dam deformation process into the trained measurement point mapping model through the measurement point monitoring model, and obtain the three-dimensional global deformation situation;

[0052] S4: Construct a deformation-stress prediction model, use the principal component analysis method to reduce the dimension of the three-dimensional global deformation, and obtain the predicted values ​​of the major principal stress, intermediate principal stress and minor principal stress of the three-dimensional global deformation through the deformation-stress prediction model;

[0053] S5: Construct a damage risk analysis model, use the Monte Carlo Dropout method, simulate the prediction distribution of the model under different weight configurations, and perform random Dropout sampling to obtain the high arch dam failure risk prediction value, and complete the high arch dam risk prediction based on the digital twin model.

[0054] The monitoring model of the measuring point in S1 is:

[0055]

[0056] Among them, H 0 is the environmental feature input to the graph convolutional feature extractor, and is the combination of hydraulic load influence factor, temperature influence factor and time-effect influence factor at time t, W 0 is the trainable weight matrix, H (·) is the node's influence factor feature matrix, l is the number of node layers, σ is the nonlinear activation function, is the normalized degree matrix, A~ is the adjacency matrix with the influence factor of self-connection, W (·) is a trainable weight matrix.

[0057] The above-mentioned measurement point monitoring model first extracts environmental impact factors as input variables based on the HST model and adopts a feature extractor-decoder architecture based on GCN.

[0058] The essence of this formula is to normalize and aggregate the information of the adjacent nodes of the influencing factors so that the characteristics of the influencing factor nodes at each layer not only depend on themselves, but also on the information of their neighboring influencing factor nodes.

[0059] The measurement point mapping model in S2 is:

[0060]

[0061] θ=(θ1,θ2)

[0062] in, is the predicted value of the measurement point mapping model, W is the learnable weight, b is the learnable bias, is the LSTM output state, is the input point of the sensor measurement value, f LSTM (·) is the LSTM network, θ1 is the parameter of the LSTM network, θ2 is the parameter of W and b, and θ is all the parameters of the measurement point mapping model.

[0063] The loss function L is dynamically monitored in S2 Digital twin Including the accuracy loss function L accuracy and the spatial variability loss function L space , the formula is:

[0064] L Digital twin =λ1L accuracy +λ2L space

[0065]

[0066] Among them, λ1 and λ2 are weight parameters, N is the total number of global calculation points, i is the i-th global calculation point, l is the loss function calculation method, and is the spatial gradient between the predicted value of the measurement point mapping model and the numerical simulation result, α is the loss balance parameter, is the loss of the finite element node corresponding to the sensor measurement point, is the loss of the sensor measuring point, is the loss of all finite element nodes.

[0067] The digital twin model in S3 is:

[0068]

[0069] in, is the predicted value of the digital twin at time t, f Synapse (·) is the measurement point mapping model, f Cell body (·) is the monitoring model of the measuring point, θ Digital twin are the pre-trained parameters of the measurement point mapping model, and It is a collection of hydraulic load influencing factors, temperature influencing factors and time influencing factors.

[0070] The deformation-stress prediction model in S4 is:

[0071]

[0072] in, is the stress prediction value, f σ (·) is the deformation-stress prediction function, ω σ is the parameter of the deformation-stress prediction model, L σ is the deformation-stress prediction model loss, σ1, σ2 and σ3 are the calculated values ​​of major principal stress, intermediate principal stress and minor principal stress, δ Digital twin is the digital twin prediction value, ω is the parameter of the deformation-stress prediction model, is the deformed data after dimensionality reduction, PCA(·) is the dimensionality reduction operation, h t is the hidden state, o t is the memory state, r is a t The Bernoulli random variable of the same dimension has a probability of 1 of p, LSTM(·) is the LSTM framework, and h t-1 is the hidden state for the next moment, o t-1 is the memory state of the next moment, ⊙ represents element-by-element multiplication, Bernoulli(·) is Bernoulli randomness, and are the predicted values ​​of major principal stress, intermediate principal stress and minor principal stress.

[0073] The deformation-stress prediction model uses principal component analysis to reduce the dimensionality of the dam deformation twin data, constructs an LSTM-based prediction framework, and captures the spatiotemporal evolution characteristics of the dam stress through time series modeling.

[0074] The damage risk analysis model in S5 is:

[0075]

[0076] in, To destroy the i-th sampling result of the risk analysis model at time t, and are the major principal stress, intermediate principal stress and minor principal stress at time t, Perform random Dropout sampling for the deformation-stress prediction function, T is the number of random Dropout sampling, is the high arch dam failure risk prediction value, P(·) is the indicator function, is the value of the Drucker-Prager yield function of the entire domain, and n is the number of connections of the plastic unit.

[0077] In one embodiment of the present invention, the main task of the DG hydropower station is to generate electricity. The dam site controls an area of ​​62,727 square kilometers, accounting for 81% of the total area of ​​the Dadu River Basin. The main buildings of the power station include a river-blocking concrete hyperbolic arch dam, a spillway energy dissipation structure, a water diversion and power generation structure, etc. The maximum dam height is 210.00 meters. The normal water level of the reservoir is 1130.00 meters, and the dead water level is 1120.00 meters. The reservoir is about 32.1 kilometers long, with a storage capacity of about 742 million cubic meters below the normal water level and a regulating storage capacity of 117 million cubic meters, which can be adjusted daily. Figure 3 As shown in the figure, the horizontal displacement monitoring of DG Dam includes the vertical line method, which is used to monitor the 6th, 10th, 14th dam sections (crown beams), 19th, 24th dam sections and two abutment grouting galleries. A total of 7 vertical lines are set to monitor the horizontal displacement of the dam body. Figure 4 As shown, this study uses horizontal deformation monitoring data from January 1, 2016 to September 1, 2022 and numerical simulation data of the finite element method as data sets. The training data, validation data, and test data are divided in proportion. The training, validation, and test data division ratios are shown in Table 1. In order to evaluate the proposed method, the finite element model and the following ablation models were selected for comparison: general neural network (LSTM framework only), Synapse model, ablation model I (eliminating spatial variability loss), and ablation model II (eliminating jump connections). The parameters of all models remain consistent, with 192 hidden layers.

[0078] Table 1 Dataset

[0079]

[0080] The digital twin model of DG is established by the present invention, and the specific steps are as follows:

[0081] (1) According to the structural characteristics, topography and geological conditions of the DG project, a three-dimensional finite element model of the arch dam and foundation was established. In the direction of water flow, the simulation range extends outward from the heel and toe of the dam to 2.5 times the dam height. In the cross-flow direction, it extends outward from the left and right dam shoulders to 2 times the dam height. In the vertical direction, it extends downward from the foundation surface to 2 times the dam height. The arch dam and foundation are divided into 22,019 8-node three-dimensional solid elements, including 4,902 dam body elements, 16,414 bedrock elements, and 703 weak structure interface elements. Through parameter sensitivity analysis, it was found that the elastic moduli of the three dam sections and the deformation moduli of the three bedrocks are sensitive parameters of the arch dam displacement. Therefore, in this case, these six parameters are selected as variables for time-varying inversion. The simulated structural interface and the three-dimensional finite element model mesh are as follows: Figure 5 According to the project design report and core sample tests, the inversion results of the physical and mechanical parameters of the dam concrete and foundation rock used in this case study are shown in Table 2.

[0082] Table 2 Key material parameters of DG dam

[0083]

[0084]

[0085] (2) Establishment of the monitoring model for measuring points. According to step S1, the monitoring model for measuring points is established, and the input is the hydraulic load factor (H, H) constructed based on the HST model. 2 ,H 3 ,H 4 ), temperature factor and the aging factor (θ,ln(θ),e -θ ), output the monitoring effect quantity prediction values ​​of downstream and cross-river deformation of 18 measuring points including PL6-1-PL24-3. The experiment was carried out using Pycharm IDE and Pytorch2.5.1 based on Python3.10 in Windows 10 system environment. The server configuration is as follows: CPU is Intel Core i7-13700, GPU is Nvidia RTX 4070SSUPER. In the experiment, the learning rate is set to 0.001, the batch size is 64, the optimizer selects Adam optimizer to update the network parameters, and the number of iterations is set to 500. The model has 64 hidden layers, 2 graph convolution layers, and 2 multi-layer perceptron layers.

[0086] (3) Establishment of the cross-section model of the measuring point mapping. According to step S2, the cross-section model of the measuring point mapping was constructed, and the finite element calculation values ​​of the downstream and transverse deformations of 18 measuring points such as PL6-1-PL24-3 were output as the predicted values ​​of all finite element nodes. During the training process, the loss function was divided into two parts: spatial variability loss and precision loss (the loss of the prediction of the actual monitoring quantity, the loss of the prediction of the node calculation value corresponding to the arranged measuring points, and the loss of the prediction of the calculation value of all finite element nodes) to dynamically adjust the comprehensive learning of the actual monitoring status and the finite element calculation status. Pycharm IDE and Pytorch2.5.1 based on Python3.10 were used in the Windows 10 system environment. The server configuration is as follows: the CPU is Intel Core i7-13700 and the GPU is Nvidia RTX 4070SSUPER. In the experiment, the learning rate was set to 0.001, the batch size was 64, the optimizer selected the Adam optimizer to update the network parameters, the number of iterations was set to 500, and the model hidden layer was 192.

[0087] (4) Establishment of digital twin model. According to step S3, the predicted values ​​of the monitoring effect of the downstream and cross-river deformation of 18 measuring points such as measuring points PL6-1-PL24-3 are input into the pre-trained measuring point mapping section model, and the deformation of the three-dimensional whole domain of the high arch dam is output. Figure 6 , Figure 7 , Figure 8 , Fig. 9 As shown, Figure 8 (a) is 1135 meters, (b) is 1080 meters, (c) is 1030 meters, and (d) is 980 meters; Fig. 9 (a) is May 1, 2021, (b) is August 1, 2021, (c) is May 1, 2022, (d) is August 1, 2022, (e) is 1135m, (f) is 1080m, (j) is 1030m, and (h) is 980m.

[0088] (5) Establishment of deformation-stress prediction model. According to step S4, the three-dimensional deformation data calculated by the finite element model of the high arch dam is used as input after PCA dimensionality reduction, and the output is the major principal stress, intermediate principal stress and minor principal stress of the three-dimensional whole domain. Pycharm IDE and Pytorch2.5.1 based on Python3.10 are used in the Windows 10 system environment. The server configuration is as follows: CPU is Intel Core i7-13700, GPU is Nvidia RTX 4070SSUPER. In the experiment, the learning rate is set to 0.001, the batch size is 64, the optimizer selects Adam optimizer to update the network parameters, the number of iterations is set to 500, and the model hidden layer is 192. The accuracy analysis is as follows Fig.10 shown.

[0089] (6) Damage risk analysis model. According to step S4 and step S5, the three-dimensional deformation data calculated by the digital twin model of the high arch dam based on the pre-trained parameters is used as input, and the output is the major principal stress, intermediate principal stress and minor principal stress of the entire three-dimensional domain. Through Monte Carlo Dropout sampling 800,000 times, the plastic failure risk of the entire high arch dam is analyzed in real time. Pycharm IDE and Pytorch2.5.1 based on Python3.10 are used in the Windows 10 system environment. The server configuration is as follows: CPU is Intel Core i7-13700, GPU is Nvidia RTX 4070SSUPER. In the experiment, the learning rate is set to 0.001, the batch size is 64, the optimizer selects the Adam optimizer to update the network parameters, the number of iterations is set to 500, and the model hidden layer is 192. The results are analyzed as follows. Fig.11 As shown, (a) represents the average value of node 25319 under different Dropout parameters, (b) represents the standard deviation of node 25319 under different Dropout parameters, (c) represents the MC sampling statistics of node 25319 under different Dropout parameters on May 1, 2021, and (d) represents the real-time risk process line of global plastic failure of the digital twin model under different Dropout parameters.

[0090] This embodiment applies the digital twin model of high arch dam to high arch dam projects. Compared with the traditional numerical simulation method, the digital twin model not only improves the prediction accuracy, but also more accurately captures the structural response characteristics of the dam body. Its prediction accuracy is 47.09% higher than that of the numerical model based on parameter inversion. The ablation experiment verifies the spatial variability loss function and jump connection strategy of the measurement point mapping section model, which can effectively optimize the spatial distribution of deformation and improve the prediction accuracy. In addition, the deformation-stress prediction model successfully realizes the mapping from the global deformation state of the dam body to the global stress state. The predicted stress distribution is highly consistent with the finite element calculation results. The mean absolute error (MAE) of the three-way principal stress is 0.00509MPa, 0.00679MPa and 0.01243MPa, respectively, indicating that it has a high prediction accuracy. Based on the proposed digital twin framework, real-time assessment of the global plastic failure risk of high arch dams is also achieved. In 800,000 random sampling tests, as the Dropout rate increases, the distribution of three-way principal stresses tends to be normally distributed. At all tested Dropout rates, from May 1, 2021 to September 1, 2022, the global plastic failure probability of the high arch dam is always 0, indicating that its safety margin is high, which is consistent with actual observations.

[0091] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the invention.

Claims

1. A high arch dam risk prediction method based on a digital twin model, characterized in that: The following steps are involved: S1: Construct a monitoring model for measuring points, use the environmental impact factors extracted based on the HST model as the input of the monitoring model for measuring points, and obtain the predicted value of the deformation process of the high arch dam; S2: Build a measurement point mapping model and use the dynamic monitoring loss function to train the model; S3: Build a digital twin model, input the predicted value of the high arch dam deformation process into the trained measurement point mapping model through the measurement point monitoring model, and obtain the three-dimensional global deformation situation; S4: Construct a deformation-stress prediction model, use the principal component analysis method to reduce the dimension of the three-dimensional global deformation, and obtain the predicted values ​​of the major principal stress, intermediate principal stress and minor principal stress of the three-dimensional global deformation through the deformation-stress prediction model; S5: Construct a damage risk analysis model, use the Monte Carlo Dropout method, simulate the prediction distribution of the model under different weight configurations, and perform random Dropout sampling to obtain the high arch dam failure risk prediction value, and complete the high arch dam risk prediction based on the digital twin model.

2. According to claim 1, a high arch dam risk prediction method based on a digital twin model is characterized in that: The monitoring model of the measuring point in S1 is: Among them, H 0 is the environmental feature input to the graph convolutional feature extractor, and is the combination of hydraulic load influence factor, temperature influence factor and time-effect influence factor at time t, W 0 is the trainable weight matrix, H (·) is the node's influence factor feature matrix, l is the number of node layers, σ is the nonlinear activation function, is the normalized degree matrix, is the adjacency matrix with the impact factor of self-connection added, W (·) is a trainable weight matrix.

3. The high arch dam risk prediction method based on digital twin model according to claim 2 is characterized in that: The measurement point mapping model in S2 is: θ=(θ1,θ2) in, is the predicted value of the measurement point mapping model, W is the learnable weight, b is the learnable bias, is the LSTM output state, is the input point of the sensor measurement value, f LSTM (·) is the LSTM network, θ1 is the parameter of the LSTM network, θ2 is the parameter of W and b, and θ is all the parameters of the measurement point mapping model.

4. The high arch dam risk prediction method based on digital twin model according to claim 3 is characterized in that: The loss function L is dynamically monitored in S2 Digital twin Including the accuracy loss function L accuracy and the spatial variability loss function L space , the formula is: L Digital twin =λ1L accuracy +λ2L space Among them, λ1 and λ2 are weight parameters, N is the total number of global calculation points, i is the i-th global calculation point, l is the loss function calculation method, and is the spatial gradient between the predicted value of the measurement point mapping model and the numerical simulation result, α is the loss balance parameter, is the loss of the finite element node corresponding to the sensor measurement point, is the loss of the sensor measuring point, is the loss of all finite element nodes.

5. The high arch dam risk prediction method based on digital twin model according to claim 4 is characterized in that: The digital twin model in S3 is: in, is the predicted value of the digital twin at time t, f Synapse (·) is the measurement point mapping model, f Cell body (·) is the monitoring model of the measuring point, θ Digital twin are the pre-trained parameters of the measurement point mapping model, and It is a collection of hydraulic load influencing factors, temperature influencing factors and time influencing factors.

6. The high arch dam risk prediction method based on digital twin model according to claim 5 is characterized in that: The deformation-stress prediction model in S4 is: in, is the stress prediction value, f σ (·) is the deformation-stress prediction function, ω σ is the parameter of the deformation-stress prediction model, L σ is the deformation-stress prediction model loss, σ1, σ2 and σ3 are the calculated values ​​of major principal stress, intermediate principal stress and minor principal stress, δ Digital twin is the digital twin prediction value, ω is the parameter of the deformation-stress prediction model, is the deformed data after dimensionality reduction, PCA(·) is the dimensionality reduction operation, h t is the hidden state, o t is the memory state, r is a t The Bernoulli random variable of the same dimension has a probability of 1 of p, LSTM(·) is the LSTM framework, and h t-1 is the hidden state for the next moment, o t-1 is the memory state of the next moment, ⊙ represents element-by-element multiplication, Bernoulli(·) is Bernoulli randomness, and are the predicted values ​​of major principal stress, intermediate principal stress and minor principal stress.

7. The high arch dam risk prediction method based on digital twin model according to claim 6 is characterized in that: The damage risk analysis model in S5 is: in, To destroy the i-th sampling result of the risk analysis model at time t, and are the major principal stress, intermediate principal stress and minor principal stress at time t, Perform random Dropout sampling for the deformation-stress prediction function, T is the number of random Dropout sampling, is the high arch dam failure risk prediction value, P(·) is the indicator function, is the value of the Drucker-Prager yield function of the entire domain, and n is the number of connections of the plastic unit.

Images