Arch dam-valley safety monitoring and early warning method and system based on residual correction

Abstract

The application discloses an arch dam-valley safety monitoring and early warning method and system based on residual correction, and the method comprises the following steps: constructing an arch dam-valley finite element simulation prediction model, combining related monitoring data to inverse load parameters and boundary conditions of the model; simulating and predicting the working behavior of the arch dam-valley system, and constructing a residual time sequence of simulation results and measured data; training and predicting the residual time sequence by using a data-driven method to obtain a residual time sequence prediction model; analyzing and determining arch dam-valley early warning indexes and a graded early warning threshold system; combining the arch dam-valley finite element simulation prediction model, the residual time sequence prediction model and the graded early warning threshold system to construct an arch dam-valley system safety monitoring and early warning model. The application provides key technical support for improving the disaster prevention and reduction capacity of the arch dam-valley system and guaranteeing long-term safety of the project.

Description

Technical Field

[0001] This invention relates to the field of water conservancy and hydropower engineering technology, and in particular to a method and system for monitoring and early warning of the safety of arch dams and valleys based on residual correction. Background Technology

[0002] As a core hydraulic structure in water conservancy and hydropower projects, the structural safety and long-term stable operation of arch dams directly affect the overall benefits of the project, the allocation of water resources in the basin, and the safety of life and property in the surrounding areas. The structural behavior evolution of arch dams is directly affected by external environmental loads such as reservoir water level and temperature. Due to the structural stress characteristics of arch dams, there is a strong mechanical coupling effect between them and the valley foundation. The deformation response of the valley foundation is directly transmitted to the dam body and changes its stress and deformation distribution. Conversely, the stress and deformation of the dam body also have a feedback effect on the mechanical state of the valley foundation. The two constitute an inseparable whole system.

[0003] However, existing technologies and methods for monitoring the safety of arch dams generally treat the arch dam and the valley foundation as independent research objects, without fully considering the dynamic coupling between them. They only monitor the local behavior parameters of the dam body or the valley, which cannot fully reflect the overall working state of the arch dam-valley system. This leads to deviations in the judgment of the true working state of the dam body, making it difficult to accurately identify the stress and deformation correlation problems within the system. Consequently, it can cause incomplete safety assessments, delayed early warnings, and even false alarms or omissions.

[0004] With the development of monitoring and numerical analysis technologies, data-driven models and physical mechanism-driven methods have become the two mainstream technical paths for arch dam safety monitoring. Data-driven models rely on massive amounts of monitoring data to mine statistical patterns and predict parameters such as deformation, requiring no complex physical modeling, and are widely used in engineering. However, these single models lack the constraints and explanations of physical mechanisms, are highly dependent on monitoring data, have weak generalization ability, and are prone to prediction failure under abnormal engineering conditions. Furthermore, they do not adequately handle the non-stationary characteristics in monitoring data, making it difficult to guarantee the stability of prediction accuracy. Physical mechanism-driven methods, with numerical simulation at their core, analyze the structural behavior of arch dams by constructing mechanical models and solving governing equations, providing clear physical interpretations of the analysis results. This is a classic method for arch dam mechanical analysis. However, single numerical simulation methods often suffer from insufficient computational accuracy in practical engineering applications due to simplification of material parameters and boundary conditions, while three-dimensional refined finite element models suffer from low computational efficiency, failing to meet the needs of real-time monitoring in engineering sites and being difficult to adapt to complex actual engineering conditions.

[0005] In summary, the field of safety monitoring and early warning for arch dam-valley systems currently faces dual core technological challenges. On the one hand, traditional monitoring methods neglect the coupling effect between the arch dam and the valley and lack a systematic monitoring and analysis approach, resulting in insufficient systematicness and comprehensiveness in safety assessment. On the other hand, both single data-driven models and single physical mechanism-driven methods have insurmountable technical defects and cannot simultaneously meet the multiple requirements of real-time performance, high precision, and strong interpretability for the safety monitoring of long-term service arch dams. Summary of the Invention

[0006] To address the aforementioned issues, this invention proposes a method and system for monitoring and early warning of arch dam-valley safety based on residual correction. This method comprehensively considers the interaction between the arch dam and the valley, and integrates the advantages of physical mechanism-driven and data-driven approaches. It provides key technical support for enhancing the disaster prevention and mitigation capabilities of the arch dam-valley system and ensuring the long-term safety of the project.

[0007] The technical solution adopted in this invention is as follows: A method for monitoring and early warning of arch dam-valley safety based on residual correction includes: A finite element simulation prediction model of the arch dam-valley was constructed, and the load parameters and boundary conditions of the model were inverted by combining relevant monitoring data. Simulations were used to predict the working behavior of the arch dam-valley system, and residual time series of simulation results and measured data were constructed. A data-driven method was used to train and predict the residual time series to obtain a residual time series prediction model. The early warning indicators and hierarchical early warning threshold system of the arch dam-valley system were analyzed and determined. Based on the finite element simulation prediction model, residual time series prediction model and hierarchical early warning threshold system of the arch dam-valley system, a safety monitoring and early warning model of the arch dam-valley system was constructed.

[0008] Furthermore, the construction of the arch dam-valley finite element simulation prediction model, combined with relevant monitoring data to invert the load parameters and boundary conditions of the model, includes: Finite element simulation prediction model construction: Construct a physical mechanism-driven finite element simulation prediction model of arch dam-valley, and set the time-dependent component of valley deformation as displacement boundary condition; Monitoring data collection: The monthly average temperature distribution along the elevation of the upstream thermometers of the dam section over many years is statistically analyzed, and combined with the multi-year average air temperature of the dam site area, it serves as the boundary conditions for the upstream water temperature and the dam body air temperature. Inversion of key thermodynamic parameters of the dam: A model of the later temperature rise of the dam concrete is constructed by regression analysis, and the linear expansion coefficient of the dam body is inverted using the stress-free gauge and corresponding thermometer observation data in the monitoring data. Inversion of the elastic modulus of the dam body: Based on the deformation increment during the rapid rise of water level, considering only the hydrostatic pressure, the deformation increment of the dam body under different elastic moduli of concrete is calculated; with the goal of minimizing the error between the calculated value and the monitored value, the optimal elastic modulus of the dam body concrete is determined. Inversion of the application pattern of the time-dependent component of valley deformation: The time-dependent component is separated from the valley deformation monitoring data using statistical regression methods. Based on the monitoring data of valley deformation and dam deformation, the construction, impoundment and operation of the arch dam are simulated and inverted to minimize the difference between the calculated value and the monitored value, and obtain the optimal distribution of the time-dependent component of valley deformation.

[0009] Furthermore, during the inversion of the key thermodynamic parameters of the dam, a model for the later-stage temperature rise of the dam concrete is constructed through regression analysis, including:

[0010] in, Any time after cooling water is shut off The regression results of the adiabatic temperature rise at that time For the subsequent adiabatic temperature rise, It is a natural constant. , These are coefficients to be determined; The method of using stress-free gauge and corresponding thermometer observation data from the monitoring data to invert the linear expansion coefficient of the dam body includes:

[0011] in, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

[0012] in, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

[0013] Furthermore, the simulation predicts the working behavior of the arch dam-valley system and constructs a residual time series between the simulation results and measured data, including: Based on the inversion results, a full-process simulation calculation of the arch dam-valley system is performed. The basis for the simulation calculation also includes the inverted material parameters and the application method of valley aging deformation. A residual time series between simulation results and measured data is constructed. The residual time series is formed by calculating and constructing the difference between numerical simulation results and measured data sequentially along the time dimension.

[0014] Furthermore, the method of training and predicting the residual time series using a data-driven approach to obtain a residual time series prediction model includes: The residual time series is used to construct features, which includes constructing basic features such as water level, temperature and time, as well as derived features obtained by nonlinear transformation of the time-related features. The residual time series after feature construction is decomposed, and the decomposed sequence components are clustered according to the complexity index. The complexity index of the sequence components is obtained by calculating the sample entropy of each decomposed sequence component. For each component after different clustering, a corresponding prediction model is selected based on its frequency characteristics. The prediction results of each prediction model are then fused to obtain the final prediction result of the residual time series.

[0015] Furthermore, the analysis determines the arch dam-valley early warning indicators and a tiered early warning threshold system, including: The specific impact of valley deformation on the working performance of arch dams was analyzed. The analysis process combined the correlation monitoring data of valley deformation and dam deformation throughout the construction, impoundment and operation of the arch dam. Then, based on the simulation calculation of the damage evolution process of the arch dam under different valley deformations, the graded early warning thresholds of valley deformation were determined. The selection of early warning indicators for arch dams fully considers the coupling effect of valley deformation and conventional loads. The graded early warning thresholds are determined after conducting arch dam safety analysis through finite element simulation of overload or stress reduction.

[0016] Furthermore, the method of constructing a safety monitoring and early warning model for the arch dam-valley system by combining the finite element simulation prediction model, the residual time series prediction model, and the hierarchical early warning threshold system includes: By integrating a finite element simulation prediction model driven by physical mechanisms, a data-driven residual time series prediction model, and a hierarchical early warning threshold system for valley deformation and arch dam early warning indicators, a safety monitoring and early warning model for the arch dam-valley system is constructed, thereby achieving synchronous monitoring and early warning of the arch dam-valley system.

[0017] A residual-corrected arch dam-valley safety monitoring and early warning system includes: The finite element simulation prediction model building module is configured to build an arch dam-valley finite element simulation prediction model and invert the load parameters and boundary conditions of the model by combining relevant monitoring data. The residual time series prediction model building module is configured to simulate and predict the working behavior of the arch dam-valley system and construct the residual time series of simulation results and measured data; the residual time series is trained and predicted using a data-driven method to obtain the residual time series prediction model. The monitoring and early warning model construction module is configured to analyze and determine the early warning indicators and the hierarchical early warning threshold system of the arch dam-valley system; and to construct a safety monitoring and early warning model of the arch dam-valley system by combining the finite element simulation prediction model, the residual time series prediction model and the hierarchical early warning threshold system of the arch dam-valley system.

[0018] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method for monitoring and early warning of arch dam-valley safety based on residual correction.

[0019] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method for monitoring and early warning of arch dam-valley safety based on residual correction.

[0020] The beneficial effects of this invention are as follows: This invention fully considers the coupling characteristics of the arch dam-valley system, deeply integrates the advantages of physical mechanism-driven and data-driven technologies, and solves the problems of unsystematic monitoring, low model accuracy, and weak interpretability in traditional technologies through refined model construction, parameter inversion, physical-data dual-driven joint prediction and early warning threshold design. Compared with existing technologies, it has achieved multi-dimensional technical improvements, as detailed below.

[0021] 1. A systematic and integrated safety monitoring and early warning system for the arch dam-valley system has been achieved. This invention treats the arch dam and valley as a whole, constructs an overall finite element model of the arch dam-valley system, and incorporates the time-dependent component of valley deformation as a displacement boundary condition into the model analysis. Simultaneously, it analyzes the impact of valley deformation on the working performance of the arch dam, determines dual-level early warning thresholds for valley deformation and arch dam early warning indicators, and achieves synchronous monitoring and early warning of the arch dam structure safety under the influence of valley deformation. This accurately captures the dynamic coupling effect between the two, completely solving the problems of one-sided assessment and delayed early warning caused by the traditional method of studying the two separately, and significantly improving the comprehensiveness and foresight of safety monitoring.

[0022] 2. This invention achieves a deep integration of physical mechanism-driven and data-driven approaches, balancing high accuracy and strong physical interpretability in monitoring and prediction. Based on physical mechanism-driven approaches, this invention uses a refined parameter inversion method to accurately invert the dam's thermodynamic parameters, elastic modulus, valley deformation time-dependent component application mode, model load parameters, and boundary conditions. Combined with the inversion results, it conducts full-process simulation calculations of the arch-dam-valley system, providing solid physical theoretical support for system behavior prediction and ensuring the physical interpretability of the results. Simultaneously, a data-driven method compensates for errors in the physical model. Specialized processing and accurate prediction are performed on the residual time series between simulation results and measured data. Residual correction optimizes the prediction results of the physical simulation, ensuring that the final system behavior prediction conforms to both mechanical laws and actual engineering conditions, achieving a dual guarantee of accuracy and interpretability.

[0023] 3. Significantly improves the prediction accuracy and robustness of non-stationary time series. This invention designs a multi-step, refined data-driven processing flow for residual time series. By constructing basic features and nonlinear derived features to enrich the data dimensions, it clusters the decomposed sequence components using sample entropy as a complexity index, selects a suitable prediction model based on component frequency features, and completes result fusion. This forms a non-stationary time series prediction system of "feature construction - decomposition and clustering - frequency-based modeling - result fusion," which significantly improves the accuracy and stability of prediction. Furthermore, by fusing with numerical simulation prediction results, the overall prediction results of the arch dam-valley system behavior are made more consistent with engineering reality.

[0024] 4. A scientific and practically aligned graded early warning threshold system was constructed, improving the accuracy and reliability of safety early warnings. When determining the graded early warning thresholds for valley deformation, this invention combines monitoring data from the entire construction, impoundment, and operation phases of the arch dam, and determines the thresholds based on the influence of different valley deformation effects on the arch dam's behavior. When selecting early warning indicators for the arch dam, the coupling effect of valley deformation and conventional loads is fully considered. The graded early warning thresholds are determined after conducting finite element simulations of overload / reduction to analyze the arch dam's safety. The determination of the arch dam-valley graded early warning threshold system relies on both actual engineering monitoring data and solid mechanical analysis, effectively avoiding the subjectivity of early warning threshold setting and significantly reducing the probability of false alarms and missed alarms.

[0025] 5. This invention achieves refined inversion of finite element model parameters and boundary conditions, providing reliable support for physical simulation calculations. It designs a multi-dimensional parameter inversion process, constructing a concrete late-stage temperature rise model using quantitative formulas, inverting the dam body's linear expansion coefficient, and inverting the application mode of the dam body's elastic modulus and valley deformation time-dependent components based on the principle of minimizing errors. Simultaneously, it combines multiple types of monitoring data to determine the model's water and air temperature boundary conditions, making the finite element model parameters and boundary conditions more closely match engineering realities. This significantly reduces model errors caused by parameter simplification, laying a solid foundation for the accuracy of subsequent full-process simulation calculations of the arch dam-valley system. Attached Figure Description

[0026] Figure 1 This is a flowchart of a method for monitoring and early warning of arch dam-valley safety based on residual correction, according to Embodiment 2 of the present invention.

[0027] Figure 2 This is a schematic diagram of an overall finite element model of an arch dam-valley according to Embodiment 2 of the present invention.

[0028] Figure 3 This is a schematic diagram of the prediction results and residual time series of the physical mechanism driven model in Embodiment 2 of the present invention.

[0029] Figure 4 This is a detailed flowchart of step three in Embodiment 2 of the present invention.

[0030] Figure 5 This is a schematic diagram of the arch dam-valley early warning indicators and graded early warning threshold system in Embodiment 2 of the present invention. Detailed Implementation

[0031] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments are now described. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention; that is, the described embodiments are only a part of the embodiments of the invention, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0032] Example 1 This embodiment provides a method for monitoring and early warning of arch dam-valley safety based on residual correction, including: A finite element simulation prediction model of the arch dam-valley was constructed, and the load parameters and boundary conditions of the model were inverted by combining relevant monitoring data. Simulations were used to predict the working behavior of the arch dam-valley system, and residual time series of simulation results and measured data were constructed. A data-driven method was used to train and predict the residual time series to obtain a residual time series prediction model. The early warning indicators and hierarchical early warning threshold system of the arch dam-valley system were analyzed and determined. Based on the finite element simulation prediction model, residual time series prediction model and hierarchical early warning threshold system of the arch dam-valley system, a safety monitoring and early warning model of the arch dam-valley system was constructed.

[0033] Preferably, the process of constructing a finite element simulation prediction model of the arch dam-valley and inverting load parameters and boundary conditions based on relevant monitoring data includes five steps: finite element simulation prediction model construction, monitoring data acquisition, inversion of key thermodynamic parameters of the dam, inversion of the dam's elastic modulus, and inversion of the time-dependent component application mode of valley deformation. Specifically, it includes the following steps: Finite element simulation prediction model construction: Construct a physical mechanism-driven finite element simulation prediction model of arch dam-valley, and set the time-dependent component of valley deformation as displacement boundary condition; Monitoring data collection: The monthly average temperature distribution along the elevation of the upstream thermometers of the dam section over many years is statistically analyzed, and combined with the multi-year average air temperature of the dam site area, it serves as the boundary conditions for the upstream water temperature and the dam body air temperature. Inversion of key thermodynamic parameters of the dam: A model of the later temperature rise of the dam concrete is constructed by regression analysis, and the linear expansion coefficient of the dam body is inverted using the stress-free gauge and corresponding thermometer observation data in the monitoring data. Inversion of the elastic modulus of the dam body: Based on the deformation increment during the rapid rise of water level, considering only the hydrostatic pressure, the deformation increment of the dam body under different elastic moduli of concrete is calculated; with the goal of minimizing the error between the calculated value and the monitored value, the optimal elastic modulus of the dam body concrete is determined. Inversion of the application pattern of the time-dependent component of valley deformation: The time-dependent component is separated from the valley deformation monitoring data using statistical regression methods. Based on the monitoring data of valley deformation and dam deformation, the construction, impoundment and operation of the arch dam are simulated and inverted to minimize the difference between the calculated value and the monitored value, and obtain the optimal distribution of the time-dependent component of valley deformation.

[0034] It should be noted that this method enables refined inversion of finite element model parameters and boundary conditions, making the model more closely match the actual engineering conditions of the arch dam-valley system, improving the accuracy of subsequent simulation calculations, and clarifying the application mode of the valley deformation time-effect component, fully demonstrating the coupling effect between the arch dam and the valley.

[0035] More preferably, when inverting key thermodynamic parameters of the dam, a model for the later-stage temperature rise of the dam concrete is constructed through regression analysis, including:

[0036] in, Any time after cooling water is shut off The regression results of the adiabatic temperature rise at that time For the subsequent adiabatic temperature rise, It is a natural constant. , These are coefficients to be determined; The method of using stress-free gauge and corresponding thermometer observation data from the monitoring data to invert the linear expansion coefficient of the dam body includes:

[0037] in, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

[0038] in, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

[0039] It should be noted that this method achieves accurate inversion of key thermodynamic parameters of the dam through quantitative formulas. Compared with traditional empirical methods, it improves the scientificity and accuracy of parameter inversion, provides reliable thermodynamic parameter support for the accurate construction of finite element models, and reduces the impact of parameter errors on subsequent simulation calculations.

[0040] Preferably, when simulating and predicting the working behavior of the arch dam-valley system and constructing the residual time series, the whole process simulation calculation is carried out in combination with the inverted material parameters and the valley deformation application method, and the residual time series is constructed by calculating the difference between the simulation and measured data according to the time dimension.

[0041] Specifically, by integrating the load parameters and boundary conditions obtained from the inversion, and combining the inverted material parameters and the determined valley time-dependent deformation application method, a full-process simulation calculation of the arch dam-valley system from construction and water impoundment to operation is carried out to obtain the predicted results of the working behavior of the system at each stage; field measured data of the corresponding time dimension are collected, and the difference between the numerical simulation results and the measured data is calculated at each time step. All differences are arranged in chronological order to construct a residual time series.

[0042] It should be noted that the full-process simulation calculation fully incorporates various parameters after inversion, making the predicted results of the system's working behavior more consistent with engineering reality. The residual time series constructed according to the time dimension can accurately reflect the deviation between the simulation model and the actual project, providing a quantitative and continuous basis for subsequent residual correction.

[0043] Preferably, when using a data-driven method to train and predict residual time series, the operations of feature construction, sequence decomposition and clustering, frequency-based modeling, and prediction result fusion are completed sequentially, with sample entropy used as the complexity index of sequence components.

[0044] Specifically, the basic features of water level, temperature, and time are first constructed for the residual time series. At the same time, the time-related features are nonlinearly transformed to obtain corresponding derived features, thus completing the feature construction. The residual time series after feature construction is decomposed, and the sample entropy of each decomposed sequence component is calculated and used as a complexity index. The sequence components are then clustered based on this index. According to the frequency characteristics of each clustered component, suitable prediction models are selected for components of different frequencies. After each model completes the prediction of the corresponding component, the prediction results of all models are fused to obtain the final prediction result of the residual time series.

[0045] It should be noted that feature construction enriches the feature dimensions of the residual sequence, sample entropy as a complexity indicator enables accurate clustering of sequence components, frequency-based selection of the appropriate model improves the targeting and accuracy of residual prediction, and prediction result fusion further ensures the reliability of residual time series prediction, providing accurate residual data support for subsequent model correction.

[0046] Preferably, when determining the graded early warning threshold system for valley deformation and arch dam early warning indicators, the influence law of valley deformation is analyzed by combining the monitoring data of the arch dam throughout the entire stage and its threshold is determined. The early warning indicators for arch dam are selected considering the coupling effect of valley deformation and conventional load, and their thresholds are determined by finite element simulation of overload or stress reduction.

[0047] Specifically, monitoring data on the correlation between valley deformation and dam deformation were collected throughout the entire construction, impoundment, and operation phases of the arch dam. Data analysis was used to uncover the specific impact of valley deformation on the dam's operational performance. The damage evolution process of the arch dam under different valley deformation conditions was simulated and calculated, and warning thresholds for valley deformation were determined by level. Considering the coupling effect of valley deformation with conventional loads such as reservoir water level and temperature, key indicators that reflect the safety status of the arch dam were selected as early warning indicators. Safety analysis of the arch dam was conducted using finite element simulation methods for overload or stress reduction, and warning thresholds for the arch dam warning indicators were determined by level based on the analysis results.

[0048] It should be noted that the valley deformation threshold determined by combining full-stage monitoring data and simulation calculations is more in line with the actual operation of the system. The arch dam early warning indicators selected considering the coupling effect are more targeted. The finite element simulation of overload or stress reduction provides a mechanical theoretical basis for the determination of the arch dam early warning threshold, making the arch dam-valley graded early warning threshold system more scientific and reasonable, and improving the accuracy and rationality of subsequent early warning judgments.

[0049] Preferably, when constructing the safety monitoring and early warning model for the arch dam-valley system, a finite element simulation prediction model driven by physical mechanisms, a data-driven residual time series prediction model, and a hierarchical early warning threshold system for valley deformation and arch dam early warning indicators are integrated to achieve synchronous monitoring and early warning of the arch dam-valley system.

[0050] Specifically, the simulation prediction model of the working state of the arch dam-valley system driven by physical mechanisms is integrated with the data-driven residual time series prediction model to complete the corrected prediction of the working state of the system. The physical-data driven fusion prediction model is combined with the defined arch dam-valley graded early warning threshold system to build an integrated arch dam-valley system safety monitoring and early warning model. This model can simultaneously monitor the deformation state of the valley and the safety state of the arch dam, and issue early warnings in a timely manner when relevant indicators reach the threshold.

[0051] It should be noted that this approach achieves a deep integration of physical mechanism-driven and data-driven approaches. Residual correction makes the prediction results of the system's working state more accurate, and the integrated early warning model enables synchronous monitoring and early warning of the valley and arch dam. This solves the problem of traditional monitoring methods studying the two separately and improves the systematicness and comprehensiveness of system monitoring.

[0052] Accordingly, this embodiment also provides an arch dam-valley safety monitoring and early warning system based on residual correction, including: The finite element simulation prediction model building module is configured to build an arch dam-valley finite element simulation prediction model and invert the load parameters and boundary conditions of the model by combining relevant monitoring data. The residual time series prediction model building module is configured to simulate and predict the working behavior of the arch dam-valley system and construct the residual time series of simulation results and measured data; the residual time series is trained and predicted using a data-driven method to obtain the residual time series prediction model. The monitoring and early warning model construction module is configured to analyze and determine the early warning indicators and the hierarchical early warning threshold system of the arch dam-valley system; and to construct a safety monitoring and early warning model of the arch dam-valley system by combining the finite element simulation prediction model, the residual time series prediction model and the hierarchical early warning threshold system of the arch dam-valley system.

[0053] It should be noted that the modular design described above enables the division of labor in each stage of the safety monitoring and early warning of the arch dam-valley system. The collaborative work of each module realizes the full automation of the process from model building to early warning model construction, providing stable system support for real-time monitoring at the engineering site and reducing errors from manual operation.

[0054] Example 2 This embodiment provides a method for monitoring and early warning of arch dam-valley safety based on residual correction, such as... Figure 1 As shown, it includes the following steps: Step 1: Construct a finite element simulation prediction model of the arch dam-valley, such as... Figure 2As shown, the time-dependent component of valley deformation is set as the displacement boundary condition, and combined with monitoring data such as air temperature, reservoir water level, valley deformation and groundwater level, the load parameters and boundary conditions in the finite element simulation calculation are accurately inverted.

[0055] Step Two: Based on the boundary conditions, material parameters, and valley deformation application methods obtained from the inversion, conduct full-process simulation calculations to predict the working behavior of the arch dam-valley system, and construct the residual time series of the numerical simulation results and measured data. The numerical simulation prediction results and residual time series of the valley deformation prediction are shown below. Figure 3 As shown.

[0056] Step 3: Establish a prediction model based on CEEMD decomposition and mixed trend modeling, and use data-driven methods to accurately predict the residual time series constructed in Step 2, such as... Figure 4 As shown.

[0057] Step 4: Conduct an analysis of the impact of valley deformation on the working performance of the arch dam, continuously increasing the valley deformation, and determine the graded early warning thresholds for valley deformation based on the evolution process of the arch dam's performance. Figure 5 As shown.

[0058] Step 5: Based on the coupling effect of valley deformation and conventional loads, this embodiment selects the radial deformation of the arch dam as the early warning indicator. Considering unfavorable load combinations, the safety analysis of the arch dam is carried out through finite element simulation of overload / reduction, and the graded early warning thresholds for the radial deformation of the arch dam are determined as follows: Figure 5 As shown.

[0059] Step Six: Integrate the data-driven model, the physical mechanism-driven model, and the arch dam-valley early warning indicator and threshold system to construct a safety monitoring and early warning model for the arch dam-valley system based on data-driven and physical mechanism-driven approaches.

[0060] Preferably, in step one, the specific method for calculating load parameters and boundary conditions using inversion finite element simulation is as follows: S11: The distribution pattern of the monthly average temperature of upstream thermometers along the elevation over many years is statistically analyzed, and combined with the multi-year average air temperature of the dam site area, it serves as the boundary conditions for upstream water temperature and dam body air temperature.

[0061] S12: Inverse calculation of key thermodynamic parameters of the dam. Specifically, a model of the later-stage temperature rise of the dam concrete is constructed through regression analysis:

[0062] In the above formula, Any time after cooling water is shut off The regression results of the adiabatic temperature rise at that time For the subsequent adiabatic temperature rise, It is a natural constant. , These are coefficients to be determined.

[0063] Specifically, based on the temperature monitoring results of the transverse joint gauge inside the dam body, the undetermined coefficient of C40 concrete is... , The coefficients for C35 concrete are -0.0064 and 0.69, respectively. , The coefficients for C30 concrete are -0.0058 and 0.66, respectively. , The values ​​are -0.006 and 0.68, respectively.

[0064] Using the stress-free gauge and corresponding thermometer observation data from the monitoring data, the linear expansion coefficient of the dam body is inverted using the following formula:

[0065] In the above formula, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

[0066] Specifically, based on monitoring data from 30 stress-free gauges on the dam body, temperature-microstrain relationship curves were plotted at each measuring point and linearly fitted. The average value of the linear expansion coefficient at all measuring points was taken, yielding an inversion value of 7.01 × 10⁻⁶ for the dam concrete linear expansion coefficient. -6 / ℃.

[0067] Inverse calculation of the elastic modulus of the dam body was carried out. Based on the deformation increment during the rapid rise of water level, and considering only the effect of hydrostatic pressure, the deformation increment of the dam body under different elastic moduli of concrete was calculated. The optimal elastic modulus of the dam body concrete was determined with the goal of minimizing the error between the calculated value and the monitored value.

[0068] Specifically, based on the design elastic modulus, five candidate values ​​for the elastic modulus—44 GPa, 46 GPa, 48 GPa, 50 GPa, and 52 GPa—were selected. Inversion calculations were then performed, taking into account the measured increments of radial deformation of the dam body during the rapid rise in water level. Using the vertical line of the arched beam dam section as the analysis object, the average error between the calculated and measured values ​​of deformation under different elastic moduli was compared to obtain the optimal relationship curve between the elastic modulus and the average error. It was determined that when the elastic modulus was 47.8 GPa, the average error between the calculated and measured values ​​of radial deformation at each measuring point was minimized. Therefore, the inversion value for the elastic modulus of the dam body concrete was chosen as 47.8 GPa, and the inversion value for the elastic modulus of the bedrock was 22 GPa.

[0069] S13: Inversion of the time-dependent components of valley deformation. The time-dependent components were separated from the valley deformation monitoring data using statistical regression. Based on the monitoring data of valley deformation and dam deformation, simulation inversion of the arch dam construction, impoundment, and operation processes was performed. The difference between the calculated and monitored values ​​was minimized, and the optimal distribution of the time-dependent components of valley deformation was obtained as shown below: .

[0070] Preferably, such as Figure 4 As shown, the specific method for constructing the residual prediction model of valley deformation based on CEEMD decomposition and hybrid trend modeling in step three is as follows: S31: Construct eigenvalues, using feature engineering methods to build key features affecting valley deformation, such as H (water level), T (temperature), θ (time), lnθ, etc., and then using nonlinear transformation methods to enhance time-dependent features, such as θ. 2 (lnθ) 2 , 1 / (θ+1), etc.

[0071] S32: Use the CEEMDAN decomposition method to decompose the residual sequence into several IMF components, and calculate the sample entropy of each component as a complexity index.

[0072] S33: Cluster IMF components with similar complexity, and use the linear superposition method to sum the clustering results point by point to form a combined component Co-IMF. Further VMD secondary decomposition is performed on the high-frequency Co-IMF components to improve the non-stationary characteristics.

[0073] S34: Filter the relevant features of each IMF component. Calculate the correlation strength (absolute value) between each IMF and each feature using the Pearson correlation coefficient method. Select the most relevant features using the Top-N selection method, and then merge the features of IMFs within the same Co-IMF into a dedicated feature set for that Co-IMF.

[0074] S35: Prediction and Fusion of Different Frequency Components. A multi-layer GRU (Gated Cyclic Unit) prediction model is established to predict the high-frequency and mid-frequency Co-IMFs separately. A hybrid modeling method combining polynomial trend fitting and GRU residual prediction is used to predict the low-frequency Co-IMF. Finally, the prediction results of each component are added together and fused, and R is output. 2 Related evaluation indicators such as RMSE, MAE, and MAPE.

[0075] Example 3 This embodiment is based on embodiment 1: This embodiment provides a computer device, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the arch dam-valley safety monitoring and early warning method based on residual correction described in Embodiment 1. The computer program can be in the form of source code, object code, executable file, or some intermediate form.

[0076] Example 4 This embodiment is based on embodiment 1: This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the arch dam-valley safety monitoring and early warning method based on residual correction described in Embodiment 1. The computer program can be in the form of source code, object code, executable file, or some intermediate form. The storage medium includes any entity or device capable of carrying computer program code, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0077] The above description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the forms disclosed herein and should not be construed as excluding other embodiments. It can be used in various other combinations, modifications, and environments, and can be altered within the scope of the concept described herein through the above teachings or related technologies or knowledge. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention should be within the protection scope of the appended claims.

[0078] It should be noted that, for the sake of simplicity, the foregoing method embodiments are described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

Claims

1. A method for monitoring and early warning of arch dam-valley safety based on residual correction, characterized in that, include: A finite element simulation prediction model of the arch dam-valley was constructed, and the load parameters and boundary conditions of the model were inverted by combining relevant monitoring data. Simulations were used to predict the working behavior of the arch dam-valley system, and residual time series of simulation results and measured data were constructed. A data-driven method was used to train and predict the residual time series to obtain a residual time series prediction model. The early warning indicators and hierarchical early warning threshold system of the arch dam-valley system were analyzed and determined. Based on the finite element simulation prediction model, residual time series prediction model and hierarchical early warning threshold system of the arch dam-valley system, a safety monitoring and early warning model of the arch dam-valley system was constructed.

2. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 1, characterized in that, The construction of the arch dam-valley finite element simulation prediction model, combined with relevant monitoring data, inverts the load parameters and boundary conditions of the model, including: Finite element simulation prediction model construction: Construct a physical mechanism-driven finite element simulation prediction model of arch dam-valley, and set the time-dependent component of valley deformation as displacement boundary condition; Monitoring data collection: The monthly average temperature distribution along the elevation of the upstream thermometers of the dam section over many years is statistically analyzed, and combined with the multi-year average air temperature of the dam site area, it serves as the boundary conditions for the upstream water temperature and the dam body air temperature. Inversion of key thermodynamic parameters of the dam: A model of the later temperature rise of the dam concrete is constructed by regression analysis, and the linear expansion coefficient of the dam body is inverted using the stress-free gauge and corresponding thermometer observation data in the monitoring data. Inversion of the elastic modulus of the dam body: Based on the deformation increment during the rapid rise of water level, considering only the hydrostatic pressure, the deformation increment of the dam body under different elastic moduli of concrete is calculated; with the goal of minimizing the error between the calculated value and the monitored value, the optimal elastic modulus of the dam body concrete is determined. Inversion of the application pattern of the time-dependent component of valley deformation: The time-dependent component is separated from the valley deformation monitoring data using statistical regression methods. Based on the monitoring data of valley deformation and dam deformation, the construction, impoundment and operation of the arch dam are simulated and inverted to minimize the difference between the calculated value and the monitored value, and obtain the optimal distribution of the time-dependent component of valley deformation.

3. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 2, characterized in that, When inverting the key thermodynamic parameters of the dam, a model for the later-stage temperature rise of the dam concrete is constructed through regression analysis, including: in, Any time after cooling water is shut off The regression results of the adiabatic temperature rise at that time For the subsequent adiabatic temperature rise, It is a natural constant. , These are coefficients to be determined; The method of using stress-free gauge and corresponding thermometer observation data from the monitoring data to invert the linear expansion coefficient of the dam body includes: in, The volume deformation observation results are at time i. Let be the self-generated volume deformation at time i. Let be the temperature at time i.

4. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 1, characterized in that, The simulation predicts the working behavior of the arch dam-valley system and constructs a residual time series between the simulation results and measured data, including: Based on the inversion results, a full-process simulation calculation of the arch dam-valley system is performed. The basis for the simulation calculation also includes the inverted material parameters and the application method of valley aging deformation. A residual time series between simulation results and measured data is constructed. The residual time series is formed by calculating and constructing the difference between numerical simulation results and measured data sequentially along the time dimension.

5. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 1, characterized in that, The method of training and predicting residual time series using a data-driven approach to obtain a residual time series prediction model includes: The residual time series is used to construct features, which includes constructing basic features such as water level, temperature and time, as well as derived features obtained by nonlinear transformation of the time-related features. The residual time series after feature construction is decomposed, and the decomposed sequence components are clustered according to the complexity index. The complexity index of the sequence components is obtained by calculating the sample entropy of each decomposed sequence component. For each component after different clustering, a corresponding prediction model is selected based on its frequency characteristics. The prediction results of each prediction model are then fused to obtain the final prediction result of the residual time series.

6. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 1, characterized in that, The analysis determined the early warning indicators and tiered early warning threshold system for arch dams and valleys, including: The specific impact of valley deformation on the working performance of arch dams was analyzed. The analysis process combined the correlation monitoring data of valley deformation and dam deformation throughout the construction, impoundment and operation of the arch dam. Then, based on the simulation calculation of the damage evolution process of the arch dam under different valley deformations, the graded early warning thresholds of valley deformation were determined. The selection of early warning indicators for arch dams fully considers the coupling effect of valley deformation and conventional loads. The graded early warning thresholds are determined after conducting arch dam safety analysis through finite element simulation of overload or stress reduction.

7. The method for monitoring and early warning of arch dam-valley safety based on residual correction according to claim 1, characterized in that, The aforementioned method combines the finite element simulation prediction model of the arch dam-valley system, the residual time series prediction model, and the hierarchical early warning threshold system to construct a safety monitoring and early warning model for the arch dam-valley system, including: By integrating a finite element simulation prediction model driven by physical mechanisms, a data-driven residual time series prediction model, and a hierarchical early warning threshold system for valley deformation and arch dam early warning indicators, a safety monitoring and early warning model for the arch dam-valley system is constructed, thereby achieving synchronous monitoring and early warning of the arch dam-valley system.

8. A safety monitoring and early warning system for arch dams and valleys based on residual correction, characterized in that, include: The finite element simulation prediction model building module is configured to build an arch dam-valley finite element simulation prediction model and invert the load parameters and boundary conditions of the model by combining relevant monitoring data. The residual time series prediction model building module is configured to simulate and predict the working behavior of the arch dam-valley system and construct the residual time series of simulation results and measured data; the residual time series is trained and predicted using a data-driven method to obtain the residual time series prediction model. The monitoring and early warning model construction module is configured to analyze and determine the early warning indicators and hierarchical early warning threshold system for arch dams and valleys; By combining the finite element simulation prediction model of the arch dam-valley system, the residual time series prediction model, and the hierarchical early warning threshold system, a safety monitoring and early warning model for the arch dam-valley system is constructed.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the arch dam-valley safety monitoring and early warning method based on residual correction as described in any one of claims 1-7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the arch dam-valley safety monitoring and early warning method based on residual correction as described in any one of claims 1-7.

Images