Dam safety monitoring system and method based on digital twinning

Abstract

The invention discloses a dam safety monitoring system and method based on digital twinning, and particularly relates to the technical field of digital twinning. The method comprises the following steps: acquiring stress, deformation, displacement, inclination and environmental change data of a dam structure in real time, regularly carrying out high-precision calibration on a sensor, recording a time drift error, analyzing nonlinear sensitivity of a digital twin model under an extreme working condition, simulating the influence of the sensor error and the model sensitivity on dam evaluation, predicting a deviation degree, and calculating the evaluation accuracy of the dam. Based on a prediction result, a dynamic drift compensation algorithm is adopted to correct sensor data in real time, the sensitivity of the model is dynamically adjusted, the optimized data and model are integrated, the structural state and potential risk of the dam under normal and extreme working conditions are accurately evaluated, accurate early warning information and decision scheme are generated, and the accuracy of early warning is improved. The accuracy and reliability of dam safety monitoring and risk assessment are remarkably improved, decision delay or resource waste caused by misjudgment or missed judgment is reduced, and the dam safety guarantee capacity is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of digital twins, and in particular to a dam safety monitoring system and method based on digital twins. Background Art

[0002] Dam safety monitoring based on digital twins refers to the use of digital twin technology to build a virtual digital model of the dam, and to improve the safety management efficiency of the dam through real-time monitoring, data analysis and predictive simulation. This method combines IoT sensors, artificial intelligence and big data analysis to synchronize the real-time data of the physical dam (such as displacement, seepage, water pressure, etc.) with the digital model, helping managers to quickly understand the status of the dam, predict potential risks and develop response plans. For example, the monitoring system of the Three Gorges Dam collects real-time data on the stress, deformation and surrounding environment of the dam structure through sensors installed on the dam, and transmits it to the central control platform. In the digital twin model, these data are used for simulation analysis and prediction, such as simulating the stability of the dam under earthquake or extreme rainfall conditions, thereby improving early warning capabilities. This technology can significantly reduce the limitations of manual inspections and enhance the ability to respond quickly to potential risks.

[0003] The prior art has the following deficiencies: The digital twin model integrates a variety of complex algorithms and modules. Any slight error in a single algorithm (such as incorrect parameter settings or over-simplification of mathematical models) may be magnified in the model, ultimately leading to deviations in the assessment of the entire dam status. In addition, error propagation can cause inaccurate risk assessment and early warning systems in dam safety monitoring, which may misjudge potential hazards (such as failure to detect real risks or false reporting of non-actual threats), thereby delaying key decisions or triggering unnecessary resource deployment, ultimately increasing the possibility of dam instability or accidents. Summary of the invention

[0004] The purpose of the present invention is to provide a dam safety monitoring system and method based on digital twins to address the deficiencies in the background technology.

[0005] In order to achieve the above object, the present invention provides the following technical solution: a dam safety monitoring method based on digital twin, comprising the following steps: S1: Use various types of sensors deployed at key locations on the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; S2: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, and analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions to extract the nonlinear sensitivity of the digital twin model; S3: Simulate the impact of sensor time drift error and model sensitivity on dam condition assessment, and predict the degree of deviation of the digital twin model in dam condition assessment; S4: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; S5: The calibrated monitoring data and optimized digital twin model are used to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and corresponding early warning information and decision-making plans are generated based on the evaluation results.

[0006] Preferably, in S2, the error situation of the time reference of the sensor data acquisition gradually deviating from the true value is analyzed to generate the sensor time drift deviation index, and the method for obtaining the sensor time drift deviation index is: Assuming that the output data of the sensor has time drift as time changes, it can be described by a linear model: ; Where: y(t) is the output of the sensor at time t, p is the slope of the time drift, b is the offset, and ϵ(t) is the error term; Get a set of real time reference data Output data corresponding to the sensor , q is the total number of data points, and the objective function J(p,b) is set as: ; Use J(p,b) to find the partial derivatives of p and b and set them to zero to get the optimal solution. Derivative and solve the equation separately: Derivative with respect to p: ; Derivative with respect to b: ; By solving the equation, we can get the optimal values ​​of p and b and calculate the sensor time drift deviation index, which is expressed as: ; Where SE is the sensor time drift deviation index.

[0007] Preferably, the nonlinear sensitivity of the extracted digital twin model is analyzed to generate a nonlinear sensitivity anomaly index, and the method for obtaining the nonlinear sensitivity anomaly index is: The input data set of the digital twin model is set to , where: R represents a real number set, n is the number of data points, and m is the dimension of the input feature; the data is standardized so that the mean of each feature is 0 and the standard deviation is 1. The original data set is set to ,in is the nth data point; the standardized data matrix is ​​Z, where: ; is the mean of the jth feature, is the standard deviation of the jth feature, is the jth data point. The standardized data matrix Z is used to calculate the covariance matrix Σ. The calculation formula of the covariance matrix Σ is: ; Where T is the matrix transpose, is the covariance matrix of the data; solve the eigenvalues ​​and eigenvectors of the covariance matrix Σ: Σv=λv; v is the eigenvector, λ is the eigenvalue, which represents the variance of the feature; Arrange all eigenvalues ​​in descending order, select the eigenvectors of the first k eigenvalues ​​as principal components, and set the first k principal components to be selected. The corresponding eigenvectors are , then the principal component matrix is: ; Project the standardized data into the selected principal component space to obtain the principal component score: T=ZP; where, is the projected score matrix, which indicates the position of each data point in the principal component space and calculates the standard deviation of each principal component score. , the expression is: ;in, is the score of the i-th sample on the j-th principal component, is the average score of the jth principal component, and defines the anomaly score The degree of abnormality of each sample in the principal component space is obtained by calculating the deviation between the score of each sample and its mean and standardizing it: ; Represents the standard deviation of the jth principal component score, and calculates the nonlinear sensitivity anomaly index. The expression is: ; In the formula, is the nonlinear sensitivity anomaly index.

[0008] Preferably, in S3, simulating the influence of sensor time drift error and model sensitivity on dam state assessment and predicting the degree of deviation of the digital twin model on dam state assessment specifically includes: The sensor time drift deviation index and nonlinear sensitivity anomaly index are normalized, and the accuracy coefficient of the digital twin model for dam status assessment is calculated based on the normalized sensor time drift deviation index and nonlinear sensitivity anomaly index.

[0009] Preferably, the acquired accuracy coefficient of the digital twin model for dam status assessment is compared with the accuracy coefficient reference threshold value pre-set based on historical data. If the accuracy coefficient of the digital twin model for dam status assessment is greater than or equal to the accuracy coefficient reference threshold value pre-set, it means that the digital twin model has high accuracy in dam status assessment and small error, and the prediction result of the digital twin model is reliable; if the accuracy coefficient of the digital twin model for dam status assessment is less than the accuracy coefficient reference threshold value pre-set, it means that the digital twin model has low accuracy in dam status assessment and large error, and the prediction result of the digital twin model is unreliable, and the model needs to be adjusted.

[0010] Preferably, in S4, according to the prediction result, the sensor data is corrected in real time using a dynamic drift compensation algorithm, specifically: assuming that the output value of the sensor is S(t), and the actual measurement value is M(t), then the drift error Δ(t) of the sensor is: Δ(t)=S(t)−M(t); the output data of the sensor is adjusted by the drift compensation factor γ(t), and the compensation factor γ(t)= , is the accuracy coefficient of the digital twin model for dam status assessment. According to the relationship between the predicted drift error and the drift compensation factor, the real-time correction formula is: ;in: is the corrected sensor data after drift compensation.

[0011] Preferably, the nonlinear sensitivity of the digital twin model is dynamically adjusted, specifically: assuming that the input vector of the digital twin model is , the output of the model is Y(t). According to the definition of model sensitivity, the elements of the sensitivity matrix J(t) are Represents the influence of the ith input on the jth output. Nonlinear sensitivity adjustment is to adjust the input variables according to the change of model sensitivity to improve the accuracy of the model. The dynamic adjustment formula is expressed as: ; Where: ΔY(t) is the change in model output due to changes in input variables, J(t) is the sensitivity matrix of the model, and ΔX(t) is the change in input variables; The sensitivity matrix J(t) is based on the nonlinear sensitivity anomaly index The adjustment formula of the sensitivity matrix is ​​dynamically adjusted with historical data: ;in: is the adjusted sensitivity matrix, J(t) is the original sensitivity matrix, is the normalized value of the nonlinear sensitivity anomaly index.

[0012] The present invention also provides a dam safety monitoring system based on digital twins, including a data acquisition module, a data analysis module, a simulation prediction module, an adjustment module and a dam assessment module; Data acquisition module: using various types of sensors deployed at key locations of the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; Data analysis module: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions, and extract the nonlinear sensitivity of the digital twin model; Simulation prediction module: simulates the impact of sensor time drift error and model sensitivity on dam status assessment, and predicts the degree of deviation of the digital twin model in dam status assessment; Adjustment module: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; Dam assessment module: Comprehensively uses the corrected monitoring data and optimized digital twin model to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and generates corresponding early warning information and decision-making plans based on the assessment results.

[0013] In the above technical solution, the technical effects and advantages provided by the present invention are: 1. The present invention deploys a variety of sensors at key locations of the dam to collect data such as stress, deformation, and displacement of the dam in real time, and calibrates the sensors regularly to accurately capture time drift errors. At the same time, combined with nonlinear sensitivity analysis, it can identify potential error sources in the model. By comprehensively analyzing sensor errors and model sensitivity, the present invention can predict the degree of deviation in the digital twin model's assessment of the dam state under extreme working conditions, and then propose a targeted dynamic compensation solution to improve the model's prediction accuracy. This method effectively reduces the risk of missed or misjudgment caused by error accumulation or failure to correct in a timely manner, and greatly improves the reliability of dam safety monitoring.

[0014] 2. The present invention ensures the accuracy of the dam's status assessment under various working conditions by accurately correcting sensor data and dynamically adjusting the nonlinear sensitivity of the digital twin model. The comprehensively optimized monitoring data and model can accurately assess the structural health status and potential risks of the dam under normal and extreme working conditions. The early warning information and decision-making plans generated based on this can not only detect potential hidden dangers in advance and issue early warnings in time, but also assist decision makers in accurate resource allocation and risk prevention, thereby effectively ensuring the long-term safe operation of the dam and reducing the risk of safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0016] Figure 1 The figure is a flow chart of the method of the present invention.

[0017] Figure 2 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Example 1, please refer to Figure 1 As shown, the dam safety monitoring method based on digital twin described in this embodiment includes the following steps: S1: Use various types of sensors deployed at key locations on the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; S2: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, and analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions to extract the nonlinear sensitivity of the digital twin model; S3: Simulate the impact of sensor time drift error and model sensitivity on dam condition assessment, and predict the degree of deviation of the digital twin model in dam condition assessment; S4: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; S5: The calibrated monitoring data and optimized digital twin model are used to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and corresponding early warning information and decision-making plans are generated based on the evaluation results.

[0020] In S1, various types of sensors are deployed at key locations of the dam, including the dam body, dam foundation, spillway and dam crest, including: Stress sensor: used to monitor the stress changes of the dam structure in real time and evaluate the structural stress conditions of the dam under the influence of water flow, external loads, etc.

[0021] Strain sensor: measures the deformation, expansion or contraction of the dam body, reflecting the health status of the dam material.

[0022] Displacement sensor: monitors the displacement of the dam body, especially when the dam body or its surrounding area moves, and provides real-time displacement amount and direction.

[0023] Tilt sensor: monitors the tilt angle of the dam, especially whether the dam body is tilted or has uneven settlement.

[0024] Environmental sensors: including temperature, humidity, rainfall, wind speed, earthquake monitoring, etc., to monitor changes in the environment around the dam to assess its impact on dam safety.

[0025] These sensors transmit the real-time collected data to the central control platform through wireless communication, optical fiber transmission or Internet of Things technology, realizing centralized management and real-time monitoring of data. The system can automatically record and process various environmental and structural data to ensure efficient and stable collection of sensor data.

[0026] The sensor collection frequency settings are optimized according to the needs of different types of data: Structural stress and strain data: higher sampling frequency to ensure real-time reflection of the force changes of the dam body. Displacement and tilt data: real-time collection, while adjusting the frequency in combination with different working conditions (such as water level, rainfall). Environmental data: collected once every hour or every day to ensure sufficient time resolution, especially in extreme weather or environmental changes, increase the collection frequency.

[0027] The system has built-in data verification and cleaning modules to ensure the accuracy of collected data. It compares historical data, detects outliers, and uses algorithms to remove noise to avoid data deviations caused by sensor failure or external interference.

[0028] S2: Use high-precision calibration equipment to regularly recalibrate all sensors and record the time drift error of each sensor, including: Regular deployment of high-precision calibration equipment To ensure the accuracy of sensor data in the dam safety monitoring system, high-precision calibration equipment is regularly used to recalibrate various sensors. The measurement accuracy of the sensor is verified by precision equipment, and the drift phenomenon that may occur during its long-term use is corrected, especially the decrease in sensor accuracy caused by factors such as temperature, humidity, and long-term work.

[0029] During long-term operation, the sensor's measured value may gradually deviate from the true value due to environmental factors (such as temperature changes, power supply fluctuations, etc.). This deviation usually manifests as a gradual drift, which may not be easy to detect in the early stage, but accumulates in long-term monitoring, resulting in inaccurate data.

[0030] The calibration results and time drift errors of each sensor will be recorded and archived. Based on historical data analysis, a drift error model is established and the collected data is corrected regularly to ensure the accuracy and consistency of the data. Dynamic error compensation can be performed in combination with algorithms such as Kalman filtering.

[0031] Dynamic correction combines the collected real-time data with the known calibration error model. The system can automatically adjust the measurement results during operation and correct the data deviation caused by drift error, thereby improving data quality and ensuring that the input data of the digital twin model is more reliable.

[0032] During the long-term operation of the sensor, the error of the time reference of the sensor data collection gradually deviates from the true value, and then the sensor time drift deviation index is generated. The method for obtaining the sensor time drift deviation index is as follows: Assume that the output data of the sensor has a certain time drift as time changes, and use a linear model to describe the change of this drift: ; Where: y(t) is the output of the sensor at time t (sensor reading). p is the slope of the time drift (representing the rate of sensor drift, that is, the change in error per unit time). b is the offset (representing the error at the start of the sensor). ϵ(t) is the error term (representing the noise or unmodeled part of the sensor output data). Get a set of real time reference data Output data corresponding to the sensor , q is the total number of data points. These data can come from the output of long-term monitoring sensors, and the objective function J(p,b) is set as: ; Use J(p,b) to find the partial derivatives of p and b and set them to zero to get the optimal solution. Derivative and solve the equation separately: Derivative with respect to p: ; Derivative with respect to b: ; By solving these two equations, we can get the optimal values ​​of p and b and calculate the sensor time drift deviation index, which is expressed as: ; Where SE is the sensor time drift deviation index.

[0033] The sensitivity analysis of digital twin models focuses on the response intensity of the model when facing changes in input parameters. Especially under extreme conditions, such as natural disasters such as floods and earthquakes, the nonlinear response of the model to input parameters (such as dam stress, environmental changes, etc.) may be more complex. Traditional linear models assume that the relationship between input and output is direct, while in digital twin models, especially under extreme conditions, the relationship between input parameters and output results is often nonlinear. Small input changes may cause the system to show a large response.

[0034] By using methods such as Taylor series expansion, partial differentiation or Lagrange interpolation, the influence of each input parameter in the digital twin model on the model output is calculated. The larger the sensitivity coefficient, the stronger the influence of the input parameter on the model result, and vice versa.

[0035] In extreme environments (such as extreme rainfall, strong earthquakes, dam aging, etc.), the digital twin model needs to simulate multiple factors. By introducing historical extreme event data, the impact of these extreme conditions on the stress, deformation and other changes of the dam structure is simulated, revealing the nonlinear behavior of the model under these conditions.

[0036] For nonlinear sensitivity analysis in dam monitoring, it is possible to identify which input parameters (such as initial stress of the dam body, external load, temperature change, etc.) have a greater impact on dam safety under extreme conditions, thus providing a basis for model optimization.

[0037] By extracting the nonlinear sensitivity coefficients of the model, it is possible to predict the amplification effect of errors under different parameter and input changes when extreme events occur. For example, slight errors in certain parameters may be exponentially amplified in the model, leading to over-prediction or underestimation of dam stability.

[0038] Based on the results of sensitivity analysis, the system can identify the amplification effect of errors on key parameters and take corresponding compensation measures, such as measuring sensitive parameters with higher accuracy or adding error correction factors during the simulation process.

[0039] For the identified highly sensitive parameters, adaptive optimization algorithms (such as genetic algorithms and particle swarm optimization) are used to tune the model to ensure that it can accurately predict the state of the dam under extreme working conditions. By adjusting the parameter weights of the model, the error propagation and amplification effects are reduced, thereby improving the stability and robustness of the digital twin model.

[0040] After analyzing the nonlinear sensitivity of the extracted digital twin model, a nonlinear sensitivity anomaly index is generated. The method for obtaining the nonlinear sensitivity anomaly index is as follows: The input data set of the digital twin model is set to , where: R represents a set of real numbers, n is the number of data points (i.e., the number of observation time points or data samples collected by sensors), and m is the dimension of the input features (i.e., multiple sensor measurement data of the dam model, environmental factors, etc.).

[0041] The data is standardized to remove the influence of different dimensions so that the mean of each feature is 0 and the standard deviation is 1. The original data set is set to ,in is the nth data point; the standardized data matrix is ​​Z, where: ; is the mean of the jth feature, is the standard deviation of the jth feature, is the jth data point. The standardized data matrix Z is used to calculate the covariance matrix Σ, which describes the correlation between each feature. The formula for calculating the covariance matrix Σ is: ; Where T is the matrix transpose, is the covariance matrix of the data; solve the eigenvalues ​​and eigenvectors of the covariance matrix Σ: Σv=λv; v is the eigenvector, λ is the eigenvalue, which represents the variance of the feature; Arrange all eigenvalues ​​in descending order, and select the first k eigenvectors with larger eigenvalues ​​as principal components. The selection of the number of principal components usually depends on the cumulative variance explanation, that is, the selected principal components can explain 80%~95% of the total variance of the data.

[0042] Set the first k principal components to be selected, and the corresponding eigenvector is , then the principal component matrix is: ; Project the standardized data into the selected principal component space to obtain the principal component score: T=ZP; where, is the projected score matrix, which indicates the position of each data point in the principal component space and calculates the standard deviation of each principal component score. , the expression is: ;in, is the score of the i-th sample on the j-th principal component, is the average score of the jth principal component, and defines the anomaly score The degree of abnormality of each sample in the principal component space is obtained by calculating the deviation between the score of each sample and its mean and standardizing it: ; represents the standard deviation of the jth principal component score, the anomaly score A higher value indicates that the sample is outlier in the principal component space, which may be a manifestation of abnormal sensitivity. The nonlinear sensitivity abnormality index is calculated, and the expression is: ; In the formula, is the nonlinear sensitivity anomaly index.

[0043] By calculating the nonlinear sensitivity anomaly index for each data point, we can identify which data points or samples show unusual sensitivity in the model. These abnormal samples may be due to nonlinear effects caused by changes in input parameters in the model, or potential abnormal behavior in the system. Ultimately, these anomalies can serve as a basis for further correction and optimization of the model.

[0044] S3: Simulate the impact of sensor time drift error and model sensitivity on dam condition assessment, and predict the degree of deviation of the digital twin model in dam condition assessment, including: The sensor time drift deviation index and nonlinear sensitivity anomaly index are normalized, and the accuracy coefficient of the digital twin model for dam status assessment is calculated based on the normalized sensor time drift deviation index and nonlinear sensitivity anomaly index.

[0045] For example, the present invention can use the following formula to calculate the accuracy coefficient of the digital twin model for dam status assessment, and the calculation expression is: ; In the formula, is the accuracy coefficient of the digital twin model for dam status assessment, SE is the sensor time drift deviation index, is the nonlinear sensitivity anomaly index, is the proportional coefficient of the sensor time drift deviation index and nonlinear sensitivity anomaly index, and .

[0046] The obtained accuracy coefficient of the digital twin model for dam status assessment is compared with the accuracy coefficient reference threshold value pre-set based on historical data. If the accuracy coefficient of the digital twin model for dam status assessment is greater than or equal to the pre-set accuracy coefficient reference threshold value, it means that the digital twin model has high accuracy in dam status assessment and small error, and the prediction result of the digital twin model is reliable; if the accuracy coefficient of the digital twin model for dam status assessment is less than the pre-set accuracy coefficient reference threshold value, it means that the digital twin model has low accuracy in dam status assessment and large error, and the prediction result of the digital twin model is unreliable, and the model needs to be adjusted.

[0047] S4: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model, including: During long-term operation, sensors will generate time drift errors due to environmental changes or hardware aging. This error may lead to deviations in dam status assessment in the digital twin model. The dynamic drift compensation algorithm reduces this drift error by correcting sensor data in real time, thereby improving the accuracy of model prediction.

[0048] Assume that the output value of the sensor is S(t), and the actual measured value is M(t), then the drift error Δ(t) of the sensor is: Δ(t)=S(t)−M(t); In order to perform real-time compensation, the output data of the sensor is adjusted by the drift compensation factor γ(t) to make it closer to the true value. This compensation factor can be dynamically calculated through the historical data of the sensor and the drift prediction results, that is, let the compensation factor γ(t)= , is the accuracy coefficient of the digital twin model for dam status assessment. According to the relationship between the predicted drift error and the drift compensation factor, the real-time correction formula is: ;in: It is the corrected sensor data after drift compensation. Through the dynamic compensation method, real-time adjustments can be made at each moment according to the drift error and sensitivity anomaly of the sensor, reducing the impact of the drift error of the sensor data on the dam status assessment.

[0049] Nonlinear sensitivity in digital twin models refers to how small changes in input variables are amplified and affect the predicted results of the output (dam status). For digital twin models, the adjustment of nonlinear sensitivity is very important because it will change with changes in the environment and system status. Therefore, the sensitivity of the model needs to be adjusted dynamically to improve the accuracy of the assessment.

[0050] Assume that the input vector of the digital twin model is , the output of the model is Y(t), then the sensitivity of the model can be calculated by linearly approximating the input change, assuming that there is a nonlinear relationship between the output and the input.

[0051] According to the definition of model sensitivity, the elements of the sensitivity matrix J(t) Represents the degree of influence of the ith input on the jth output. The core goal of nonlinear sensitivity adjustment is to adjust the input variables according to the changes in model sensitivity and improve the accuracy of the model. The dynamic adjustment formula is expressed as: ; Where: ΔY(t) is the change in model output due to changes in input variables, J(t) is the sensitivity matrix of the model, which describes the nonlinear effects of each input variable on the output, and ΔX(t) is the change in the input variable.

[0052] The sensitivity matrix J(t) is based on the nonlinear sensitivity anomaly index Dynamically adjusted with historical data. The adjustment formula of the sensitivity matrix is: ;in: is the adjusted sensitivity matrix, J(t) is the original sensitivity matrix, is the normalized value of the nonlinear sensitivity anomaly index.

[0053] Through the above formula, in actual operation, the sensitivity matrix will be dynamically adjusted according to the real-time nonlinear sensitivity anomaly index, thereby optimizing the prediction accuracy of the dam status.

[0054] S5: In the dam safety monitoring system, sensor data is usually affected by various factors, such as sensor aging, environmental changes or external interference. These factors may cause data deviation, thus affecting the accuracy of dam safety assessment. Through the aforementioned dynamic drift compensation algorithm and nonlinear sensitivity dynamic adjustment, sensor data can be corrected in real time and the sensitivity of the digital twin model can be adjusted, thereby improving the accuracy of dam safety monitoring.

[0055] The sensor data corrected by the dynamic drift compensation algorithm has removed the drift error, ensuring that the data is closer to the true value. The sensor data includes multiple dimensions such as stress, displacement, deformation, temperature, and inclination of the dam. These data can reflect the structural health status of the dam in real time.

[0056] The digital twin model integrates real-time sensor data and takes into account changes in environmental factors through accurate modeling of the dam structure. In the aforementioned optimization process, by dynamically adjusting the nonlinear sensitivity anomaly index and the time drift deviation index, the digital twin model can more accurately simulate the performance of the dam under different working conditions, especially the strain performance during extreme events (such as earthquakes, extreme rainfall or water level rise).

[0057] Based on the corrected monitoring data and the optimized digital twin model, two types of evaluations can be performed: evaluation under normal conditions and evaluation under extreme conditions.

[0058] Evaluation under normal working conditions: Under normal working conditions, the system will conduct status evaluation based on the current working environment of the dam, historical data, real-time data collected by sensors, and the optimized digital twin model. These evaluations may include: Stress distribution analysis: evaluate whether the stress of each part of the dam is within the design range and whether there is a risk of local overload. Deformation and displacement monitoring: monitor the structural deformation and displacement of the dam to ensure that there are no abnormalities during normal operation. Environmental factor evaluation: analyze the impact of environmental factors such as water level changes and temperature changes on the dam structure.

[0059] Evaluation under extreme conditions includes: The evaluation under extreme conditions focuses on the performance of the dam under special circumstances (such as earthquakes, extreme rainfall, landslides, destructive water level rise, etc.). In this case, the digital twin model will evaluate the bearing capacity and stability of the dam structure by simulating the operation under extreme conditions. For example: Earthquake simulation: Use sensor data to simulate the strain response of the dam in an earthquake and analyze whether there will be risks such as cracks and displacement. Extreme rainfall simulation: Simulate the changes in the water level of the dam and the pressure of the surrounding environment under extreme rainfall to evaluate whether catastrophic accidents such as dam failure may occur. Stress / deformation amplification effect: Under extreme conditions, the model will take into account the nonlinear amplification effect of the input data to ensure that potential risks are fully predicted.

[0060] Based on the evaluation results, the system can generate corresponding warning information according to different risk thresholds and propose decision-making plans. This process usually includes the following steps: According to the evaluation results, when the dam structure status or potential risks exceed the predetermined safety range, the system will generate early warning information in time. Early warning information can be defined in the following aspects: Status change alarm: If the local stress of the dam is too large or the deformation exceeds the safety standard, the system will trigger an alarm.

[0061] Risk level classification: Based on the risk assessment results, the system classifies the risk level faced by the dam (such as low, medium, and high risk) and issues warnings based on the risk level.

[0062] Response to extreme events: For example, in extreme conditions such as simulated earthquakes or heavy rains, if risk assessments show that the dam may face major damage, the system will issue a high-level warning and require relevant personnel to take emergency measures.

[0063] Based on the early warning information, decision makers can quickly formulate corresponding emergency response and follow-up treatment plans to ensure the safety of the dam. The decision-making plans include: Repair and maintenance plans: If the assessment finds local damage or potential risks, repair and reinforcement suggestions can be made through real-time monitoring and data analysis.

[0064] Resource allocation plan: In extreme risk situations, the system will provide resource allocation plans, such as personnel evacuation, emergency material supply, and enhanced monitoring, based on the predicted scope and impact of the disaster.

[0065] Preventive measures recommendations: For possible safety hazards, the system can provide preventive measures such as strengthening structural monitoring, reducing external loads, and optimizing water flow control to reduce potential risks.

[0066] In practical applications, the dam safety monitoring system needs to provide decision support based on accuracy coefficients, risk assessment models and early warning systems. For example: Combined with the corrected sensor data and the optimized digital twin model, the reliability of the model assessment is determined by calculating the accuracy coefficient of the digital twin model for the dam status assessment. Risk assessment model: By combining historical data and real-time monitoring data, the assessment model is dynamically adjusted to determine the stability and potential risks of the dam. Early warning and decision support system: Once the risk assessment exceeds the preset threshold, an early warning report is automatically generated to support decision makers in making effective decisions.

[0067] In this application, through the comprehensive correction of sensor data and the optimized digital twin model, the safety status and potential risks of the dam can be fully evaluated under normal and extreme conditions. This technology not only improves the monitoring accuracy of the dam's health status, but also can generate early warning information and provide decision support in an emergency, helping relevant personnel to respond quickly, ensuring the safe operation of the dam and reducing the possibility of accidents.

[0068] In this embodiment, a variety of sensors are deployed at key locations of the dam to collect data on structural stress, deformation, displacement, tilt, and environmental changes in real time. On this basis, high-precision calibration equipment is used to regularly recalibrate the sensors, record time drift errors, and analyze the input parameter changes and nonlinear sensitivity of the digital twin model under extreme conditions. By simulating the impact of sensor drift errors and model sensitivity on dam status assessment, the degree of assessment deviation is predicted. Based on the prediction results, a dynamic drift compensation algorithm is used to correct sensor data in real time, and the nonlinear sensitivity of the digital twin model is dynamically adjusted. Finally, the structural status and potential risks of the dam under normal and extreme conditions are evaluated by combining the corrected monitoring data and the optimized digital twin model, and corresponding early warning information and decision-making plans are generated to provide a scientific basis for the safety management of the dam.

[0069] Example 2, please refer to Figure 2 As shown, the dam safety monitoring system based on digital twin described in this embodiment includes a data acquisition module, a data analysis module, a simulation prediction module, an adjustment module and a dam assessment module; Data acquisition module: using various types of sensors deployed at key locations of the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; Data analysis module: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions, and extract the nonlinear sensitivity of the digital twin model; Simulation prediction module: simulates the impact of sensor time drift error and model sensitivity on dam status assessment, and predicts the degree of deviation of the digital twin model in dam status assessment; Adjustment module: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; Dam assessment module: Comprehensively uses the corrected monitoring data and optimized digital twin model to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and generates corresponding early warning information and decision-making plans based on the assessment results.

[0070] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A dam safety monitoring method based on digital twins, characterized by: The following steps are involved: S1: Use various types of sensors deployed at key locations on the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; S2: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, and analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions to extract the nonlinear sensitivity of the digital twin model; S3: Simulate the impact of sensor time drift error and model sensitivity on dam condition assessment, and predict the degree of deviation of the digital twin model in dam condition assessment; S4: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; S5: The calibrated monitoring data and optimized digital twin model are used to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and corresponding early warning information and decision-making plans are generated based on the evaluation results.

2. A dam safety monitoring method based on digital twins according to claim 1, characterized in that: In S2, the error of the time base of sensor data acquisition gradually deviating from the true value is analyzed to generate the sensor time drift deviation index. The method for obtaining the sensor time drift deviation index is: Assuming that the output data of the sensor has time drift as time changes, it can be described by a linear model: ; Where: y(t) is the output of the sensor at time t, p is the slope of the time drift, b is the offset, and ϵ(t) is the error term; Get a set of real time reference data Output data corresponding to the sensor , q is the total number of data points, and the objective function J(p,b) is set as: ; Use J(p,b) to find the partial derivatives of p and b and set them to zero to get the optimal solution. Derivative and solve the equation separately: Derivative with respect to p: ; Derivative with respect to b: ; By solving the equation, we can get the optimal values ​​of p and b and calculate the sensor time drift deviation index, which is expressed as: ; Where SE is the sensor time drift deviation index.

3. A dam safety monitoring method based on digital twins according to claim 2, characterized in that: After analyzing the nonlinear sensitivity of the extracted digital twin model, a nonlinear sensitivity anomaly index is generated. The method for obtaining the nonlinear sensitivity anomaly index is as follows: The input data set of the digital twin model is set to , where: R represents a real number set, n is the number of data points, and m is the dimension of the input feature; the data is standardized so that the mean of each feature is 0 and the standard deviation is 1. The original data set is set to ,in is the nth data point; the standardized data matrix is ​​Z, where: ; is the mean of the jth feature, is the standard deviation of the jth feature, is the jth data point. The standardized data matrix Z is used to calculate the covariance matrix Σ. The calculation formula of the covariance matrix Σ is: ; Where T is the matrix transpose, is the covariance matrix of the data; solve the eigenvalues ​​and eigenvectors of the covariance matrix Σ: Σv=λv; v is the eigenvector, λ is the eigenvalue, which represents the variance of the feature; Arrange all eigenvalues ​​in descending order, select the eigenvectors of the first k eigenvalues ​​as principal components, and set the first k principal components to be selected. The corresponding eigenvectors are , then the principal component matrix is: ; Project the standardized data into the selected principal component space to obtain the principal component score: T=ZP; where, is the projected score matrix, which indicates the position of each data point in the principal component space and calculates the standard deviation of each principal component score. , the expression is: ;in, is the score of the i-th sample on the j-th principal component, is the average score of the jth principal component, and defines the anomaly score The degree of abnormality of each sample in the principal component space is obtained by calculating the deviation between the score of each sample and its mean and standardizing it: ; Represents the standard deviation of the jth principal component score, and calculates the nonlinear sensitivity anomaly index. The expression is: ; In the formula, is the nonlinear sensitivity anomaly index.

4. A dam safety monitoring method based on digital twins according to claim 3, characterized in that: In S3, the influence of sensor time drift error and model sensitivity on dam status assessment is simulated, and the degree of deviation of the digital twin model in dam status assessment is predicted, including: The sensor time drift deviation index and nonlinear sensitivity anomaly index are normalized, and the accuracy coefficient of the digital twin model for dam status assessment is calculated based on the normalized sensor time drift deviation index and nonlinear sensitivity anomaly index.

5. A dam safety monitoring method based on digital twins according to claim 4, characterized in that: The obtained accuracy coefficient of the digital twin model for dam status assessment is compared with the accuracy coefficient reference threshold value pre-set based on historical data. If the accuracy coefficient of the digital twin model for dam status assessment is greater than or equal to the pre-set accuracy coefficient reference threshold value, it means that the digital twin model has high accuracy in dam status assessment and small error, and the prediction result of the digital twin model is reliable; if the accuracy coefficient of the digital twin model for dam status assessment is less than the pre-set accuracy coefficient reference threshold value, it means that the digital twin model has low accuracy in dam status assessment and large error, and the prediction result of the digital twin model is unreliable, and the model needs to be adjusted.

6. A dam safety monitoring method based on digital twins according to claim 5, characterized in that: In S4, according to the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time. Specifically, let the output value of the sensor be S(t) and the actual measured value be M(t), then the drift error Δ(t) of the sensor is: Δ(t)=S(t)−M(t); adjust the output data of the sensor by the drift compensation factor γ(t), and let the compensation factor γ(t)= , is the accuracy coefficient of the digital twin model for dam status assessment. According to the relationship between the predicted drift error and the drift compensation factor, the real-time correction formula is: ;in: is the corrected sensor data after drift compensation.

7. A dam safety monitoring method based on digital twins according to claim 6, characterized in that: Dynamically adjust the nonlinear sensitivity of the digital twin model. Specifically, assume that the input vector of the digital twin model is , the output of the model is Y(t). According to the definition of model sensitivity, the elements of the sensitivity matrix J(t) are Represents the influence of the ith input on the jth output. Nonlinear sensitivity adjustment is to adjust the input variables according to the change of model sensitivity to improve the accuracy of the model. The dynamic adjustment formula is expressed as: ; Where: ΔY(t) is the change in model output due to changes in input variables, J(t) is the sensitivity matrix of the model, and ΔX(t) is the change in input variables; The sensitivity matrix J(t) is based on the nonlinear sensitivity anomaly index The adjustment formula of the sensitivity matrix is ​​dynamically adjusted with historical data: ;in: is the adjusted sensitivity matrix, J(t) is the original sensitivity matrix, is the normalized value of the nonlinear sensitivity anomaly index.

8. A dam safety monitoring system based on digital twins, used to implement a dam safety monitoring method based on digital twins as described in any one of claims 1 to 7, characterized in that: It includes data acquisition module, data analysis module, simulation prediction module, adjustment module and dam assessment module; Data acquisition module: using various types of sensors deployed at key locations of the dam to collect real-time data on dam structure stress, deformation, displacement, tilt and environmental changes; Data analysis module: Use high-precision calibration equipment to regularly recalibrate all sensors, record the time drift error of each sensor, analyze the amplification effect of the digital twin model on input parameter changes under extreme working conditions, and extract the nonlinear sensitivity of the digital twin model; Simulation prediction module: simulates the impact of sensor time drift error and model sensitivity on dam status assessment, and predicts the degree of deviation of the digital twin model in dam status assessment; Adjustment module: Based on the prediction results, the dynamic drift compensation algorithm is used to correct the sensor data in real time and dynamically adjust the nonlinear sensitivity of the digital twin model; Dam assessment module: Comprehensively uses the corrected monitoring data and optimized digital twin model to evaluate the structural status and potential risks of the dam under normal and extreme conditions, and generates corresponding early warning information and decision-making plans based on the assessment results.

Images