Dam safety studying and judging method based on monitoring data multi-physical field simulation

Abstract

The invention relates to the technical field of hydraulic engineering safety monitoring, and discloses a dam safety studying and judging method based on monitoring data multi-physics field simulation, which comprises the following steps: S1, multi-source data acquisition and time-space alignment; s2, dynamically updating the numerical model; s3, safety evaluation and early warning decision making; s4, performing multi-source risk coupling analysis; and S5, issuing the early warning information in a multi-mode manner. According to the method, time-space reference unification of multi-source monitoring data is achieved through feature point matching and sliding window cross-correlation analysis, a self-adaptive Kalman filtering algorithm is adopted to dynamically invert permeability coefficients and elastic modulus parameters, and boundary conditions of a finite element model are adjusted in combination with real-time water level changes; the dynamic simulation precision of a seepage field-displacement field-stress field coupling model is improved, the static evaluation limitation of a fixed threshold value method is broken through through a three-dimensional time-varying safety envelope surface and a Bayesian network grading early warning decision tree, and the risk prediction capability in the flood routing process is enhanced through multi-parameter joint probabilistic reasoning.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project safety monitoring technology, and more specifically, to a method for assessing dam safety based on multi-physics field simulation of monitoring data. Background Technology

[0002] Currently, seepage stability analysis of dams is a key technology for safety monitoring of water conservancy projects. Its core lies in achieving effective coordination between monitoring data and numerical models. Existing technologies mainly rely on piezometer networks to obtain seepage pressure data, and then perform stability assessments by matching the static parameters of the finite element model. However, existing methods have the following significant drawbacks in data processing: the lack of a unified spatiotemporal reference for multi-source monitoring data leads to deviations in the correlation analysis between the displacement field and the seepage field; the boundary condition updates of the numerical model lag behind actual operating condition changes; and the safety threshold setting relies on empirical coefficient methods, making it difficult to capture the nonlinear evolution of seepage parameters. This directly results in insufficient dynamic coupling between seepage prediction results and measured data, especially during flood rise and fall phases, which easily leads to systematic errors and severely restricts the timeliness of early warnings and the reliability of decision-making.

[0003] The limitations of the current technology system are mainly reflected in three dimensions: First, the heterogeneity of displacement, seepage, and stress monitoring data makes multi-physics coupling analysis difficult, and the conventional weighted average method cannot eliminate spatial scale bias caused by differences in sensor deployment. Second, the material parameter update mechanism of the numerical model is rigid, and the offline correction mode that relies on manual intervention is difficult to meet the needs of real-time assessment. Finally, the fixed threshold early warning method lacks the ability to respond to the dynamic changes in hydraulic gradient during flood evolution, causing the safety assessment results to lag behind the actual risk evolution, which seriously threatens the safety of dam operation. Therefore, a dynamic assessment method for dam safety based on collaborative simulation of multi-source monitoring data is needed to achieve safe operation and maintenance of dams.

[0004] A search revealed that among existing technologies related to dam safety assessment, patent publication JP2000178951A discloses a LEVEE method. However, this document lacks a unified spatiotemporal reference calibration mechanism, making it difficult to achieve accurate spatiotemporal alignment of multiphysics monitoring data, including seepage, displacement, and stress. Patent publication JP2012117353A discloses a construction method for control structures in permeable sea areas; however, when sensor fault detection lags, this document lacks a real-time anomaly factor analysis mechanism, leading to fault data contaminating the model input. Technical issues: Existing patent JP2015078488A discloses a repair method and safety protection measures for management departments, but the above document lacks an online evaluation and automatic reconstruction mechanism for model confidence, and cannot quickly switch to alternative models when a single constitutive model fails, resulting in calculation results deviating from the actual physical process; Patent CN118859336A discloses a dam safety monitoring method and system, but the above document relies too much on manual rules and does not implement multi-event conditional probability reasoning through Bayesian networks, resulting in a delay in response to sudden working conditions.

[0005] To address this issue, this application provides a method for assessing dam safety based on multiphysics simulation of monitoring data, in order to solve the problems existing in the prior art. Summary of the Invention

[0006] The purpose of this invention is to solve the technical problems mentioned in the background section and to provide a method for assessing dam safety based on multi-physics simulation of monitoring data.

[0007] The above-mentioned objective of the present invention is achieved as follows: The present invention provides a method for assessing dam safety based on multiphysics simulation of monitoring data, comprising the following steps: S1. Multi-source data acquisition and spatiotemporal alignment to generate a monitoring dataset with unified spatiotemporal reference; S2. The numerical model is dynamically updated to establish a multi-physics coupling model of seepage field, displacement field and stress field. S3. Security assessment and early warning decision-making: Based on the hierarchical early warning decision tree of Bayesian network, joint probabilistic reasoning is performed on the degree and duration of parameter deviation from the baseline value to trigger hierarchical early warning and response plan; S4. Multi-source risk coupling analysis enables the calculation of dam safety reserve coefficient under extreme working conditions; S5. Multimodal release of early warning information: Based on the BIM+GIS digital twin decision-making platform, it performs three-level linkage of three-dimensional visualization early warning, hierarchical SMS push, and automatic loading of emergency plans. It achieves data interoperability with gate control system and drainage pumping station through OPC UA protocol, and builds a closed-loop management system of "monitoring-assessment-early warning-response".

[0008] Further, step S1 specifically involves: deploying piezometers, displacement gauges, and stress gauges at key sections of the dam to form a three-dimensional monitoring network; performing spatial coordinate registration of multi-source monitoring data using a feature point matching algorithm; and using sliding window cross-correlation analysis to achieve time series alignment, thereby generating a monitoring dataset with unified spatiotemporal reference. Step S2 is as follows: Based on real-time monitoring data, the dynamic parameters of the dam permeability coefficient k and elastic modulus E are inverted using the adaptive Kalman filter algorithm. Combined with the upstream water level change, the seepage boundary conditions and mechanical constraints of the finite element model are dynamically adjusted to establish a multi-physics field coupled model of seepage field-displacement field-stress field. Step S3 specifically involves: fusing the seepage hydraulic gradient J, displacement rate v, and stress parameters to construct a three-dimensional time-varying safety envelope that automatically scales with the evolution of the flood; and using a Bayesian network-based hierarchical early warning decision tree to perform joint probabilistic inference on the degree and duration of parameter deviation from the baseline value, thereby triggering hierarchical early warning and response plans. Step S4 specifically involves: establishing a multi-factor coupled analysis module for rainfall infiltration, seismic load, and human activities; using an equivalent linearization method to process random seismic loads; simulating the rainfall infiltration process using the Green-Ampt model; constructing a composite risk scenario library containing typical working conditions; and calculating the dam safety reserve coefficient under extreme working conditions.

[0009] Furthermore, the spatial coordinate registration in step S1 specifically includes: establishing a three-dimensional coordinate mapping model through the topological relationship of the sensor deployment locations, using a feature point matching algorithm to correct local coordinate deviations caused by installation errors, and achieving spatial scale unification of displacement field and seepage field monitoring data; Step S1 also includes a data quality control module, which identifies outlier points through wavelet packet decomposition, detects sensor faults using the local anomaly factor algorithm, and uses Kriging interpolation to perform spatiotemporal completion of missing data to ensure that the monitoring data integrity rate is not lower than the set threshold.

[0010] Furthermore, the parameter inversion in step S2 specifically involves using pore water pressure data and deformation data as observation vectors, iteratively updating the posterior estimates of the permeability coefficient k and elastic modulus E through an adaptive Kalman filter, and simultaneously correcting the material parameter matrix of the finite element model.

[0011] Furthermore, the dynamic adjustment of seepage boundary conditions in step S2 specifically includes: calculating the position of the seepage line using the unsteady flow equation based on real-time upstream water level data, and inputting the calculation results as the time-varying boundary conditions for transient seepage analysis into the finite element model.

[0012] Furthermore, the construction of the three-dimensional time-varying safety envelope in step S3 includes: dividing the flood evolution process into multiple working condition stages, determining the dynamic safety threshold boundary in each stage through the extreme value distribution of the seepage hydraulic gradient J, and generating the scaling factor of the three-dimensional envelope by combining the cumulative change of the displacement rate v.

[0013] Furthermore, the hierarchical early warning decision tree in step S3 is implemented based on a Bayesian network, specifically including: establishing a conditional probability table for three types of events: seepage anomaly, displacement mutation, and stress exceeding limit; calculating the joint posterior probability based on real-time monitoring data; and triggering an early warning signal and generating a response plan when the probability value exceeds a preset threshold.

[0014] Furthermore, step S2 also includes a model confidence assessment module. When the parameter inversion error exceeds a preset threshold or the monitoring data residuals show a trend deviation, the model reconstruction process is automatically triggered, and alternative constitutive models are loaded through parallel computing for comparative verification.

[0015] Furthermore, the multi-source risk coupling analysis in step S4 specifically includes: establishing a transfer function model of environmental factors and dam response, generating random samples through Monte Carlo simulation, extracting key risk factors using principal component analysis, and constructing a three-dimensional risk surface that includes safety factor, reliability index, and failure probability.

[0016] Furthermore, step S3 also includes visualization processing: overlaying and displaying seepage equipotential lines, displacement vector fields, and stress cloud maps in the three-dimensional dam model, marking the over-limit areas with dynamic color levels, and simultaneously generating a safety assessment report and a risk heat map.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention achieves high-precision spatiotemporal benchmark unification of multi-source monitoring data through feature point matching and sliding window cross-correlation analysis; further, it adopts an adaptive Kalman filter algorithm to dynamically invert the permeability coefficient and elastic modulus parameters, and combines real-time water level changes to dynamically adjust the boundary conditions of the finite element model, which significantly improves the dynamic simulation accuracy of the seepage field-displacement field-stress field coupled model; by introducing a three-dimensional time-varying safety envelope surface and a Bayesian network hierarchical early warning decision tree, it breaks through the static evaluation limitations of the traditional fixed threshold method, and significantly enhances the risk identification and prediction capabilities during flood evolution by using multi-parameter joint probabilistic inference. 2. This invention establishes a topological relationship and three-dimensional coordinate mapping model of sensor deployment locations, and combines a feature point matching algorithm to effectively correct local coordinate deviations caused by installation errors, thereby achieving precise unification of multi-physics monitoring data in spatial scale. The data quality control module uses wavelet packet decomposition technology to identify outliers in the monitoring data and integrates a local anomaly factor algorithm to detect sensor fault status in real time. At the same time, it uses Kriging interpolation to perform multi-dimensional completion of spatiotemporally missing data, constructing a high-completeness monitoring dataset, effectively eliminating the impact of noise interference and data missingness on model accuracy, and ensuring that the data integrity rate continuously meets the requirements of dynamic analysis. 3. This invention uses pore water pressure and deformation data as observation vectors, and employs an adaptive Kalman filter to dynamically iteratively update the posterior estimates of the permeability coefficient k and elastic modulus E, while simultaneously correcting the finite element model parameter matrix, significantly improving the real-time performance and accuracy of parameter inversion. Combined with real-time upstream water level data, the invention uses the unsteady flow equation to dynamically calculate the wetting line position as the transient seepage boundary condition, enabling the model boundary to accurately match the unsteady seepage characteristics during flood rise and fall, significantly enhancing the spatiotemporal adaptability of the seepage field simulation. The model confidence assessment module automatically triggers the model reconstruction mechanism by tracking the inversion error and data residual trend in real time, and uses parallel computing to quickly load alternative constitutive models for comparison and verification, effectively avoiding the risk of single model failure. This forms a closed-loop optimization system of "dynamic correction - boundary synchronization - fault-tolerant reconstruction," ensuring that the multi-physics coupled model possesses excellent computational robustness and reliability under all operating conditions. 4. Based on the extreme value distribution of the seepage hydraulic gradient J and the cumulative change of displacement rate v, this invention dynamically constructs a three-dimensional time-varying safety envelope surface, realizing intelligent scaling of the safety boundary with water level changes, significantly improving the sensitivity of risk assessment under complex working conditions; a hierarchical early warning decision tree is constructed using a Bayesian network, and the conditional probability table and joint posterior probability calculation of three types of events—seepage anomaly, displacement mutation, and stress over-limit—achieves accurate quantification of multi-parameter coupled risks, enhancing the system's autonomous reasoning and early warning capabilities for sudden working conditions, effectively reducing the false alarm rate and significantly shortening the early warning response time. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow in this invention; Figure 2 This is a schematic diagram of the multi-source data acquisition and spatiotemporal alignment process in this invention; Figure 3 This is a schematic diagram of the dynamic update process of the numerical model in this invention; Figure 4 This is a schematic diagram of the safety assessment and early warning decision-making process in this invention; Figure 5 This is a schematic diagram of the multi-source risk coupling analysis process in this invention; Figure 6 This is a schematic diagram of the multimodal release process of early warning information in this invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0020] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0021] Reference Figures 1-6 The image shows a preferred embodiment of the present invention.

[0022] Example 1: Please refer to Figures 1-6 As shown in the figure, this embodiment provides a method for assessing dam safety based on multiphysics simulation of monitoring data. The method includes the following steps: Step S1: Multi-source data acquisition and spatiotemporal alignment: Piezometers, displacement gauges and stress gauges are deployed at key sections of the dam to form a three-dimensional monitoring network. Spatial coordinate registration of multi-source monitoring data is performed through feature point matching algorithm, and time series alignment is achieved by sliding window cross-correlation analysis to generate a monitoring dataset with unified spatiotemporal reference.

[0023] Step S2, Dynamic Update of Numerical Model: Based on real-time monitoring data, the dynamic parameters of the dam permeability coefficient k and elastic modulus E are inverted using the adaptive Kalman filter algorithm. Combined with the upstream water level changes, the seepage boundary conditions and mechanical constraints of the finite element model are dynamically adjusted to establish a multi-physics coupled model of seepage field-displacement field-stress field.

[0024] Step S3, Safety Assessment and Early Warning Decision: By integrating the seepage hydraulic gradient J, displacement rate v, and stress parameters, a three-dimensional time-varying safety envelope surface that automatically scales with the evolution of the flood is constructed. Based on a Bayesian network-based hierarchical early warning decision tree, joint probabilistic inference is performed on the degree and duration of parameter deviation from the baseline value to trigger hierarchical early warning and response plans.

[0025] Step S4, Multi-source Risk Coupling Analysis: Establish a multi-factor coupling analysis module for rainfall infiltration, seismic load, and human activities. Use the equivalent linearization method to process random seismic loads. Simulate the rainfall infiltration process using the Green-Ampt model. Construct a composite risk scenario library containing typical working conditions to calculate the dam safety reserve coefficient under extreme working conditions.

[0026] Step S5, Multimodal Release of Early Warning Information: Develop a digital twin decision-making platform based on BIM+GIS to achieve three-level linkage of 3D visualization early warning, hierarchical SMS push, and automatic loading of emergency plans. Through the OPC UA protocol, data exchange is achieved with the gate control system and drainage pumping station to build a closed-loop management system of "monitoring-assessment-early warning-response".

[0027] In this embodiment, the spatiotemporal benchmark of multi-source monitoring data is unified through feature point matching and sliding window cross-correlation analysis, effectively solving the data fusion deviation problem caused by sensor deployment differences in traditional methods. The adaptive Kalman filter algorithm is used to dynamically invert the permeability coefficient and elastic modulus parameters, and the boundary conditions of the finite element model are adjusted in combination with real-time water level changes, which significantly improves the dynamic simulation accuracy of the coupled seepage field-displacement field-stress field model. The innovative three-dimensional time-varying safety envelope surface and Bayesian network hierarchical early warning decision tree break through the limitations of static evaluation of the fixed threshold method. The risk prediction capability in the process of flood evolution is enhanced through multi-parameter joint probabilistic inference. By establishing a multi-factor coupled analysis module of rainfall-earthquake-human activity, the equivalent linearization method and Green-Ampt model are used to simulate extreme working conditions, and the accurate calculation of the dam safety reserve coefficient in complex environments is realized. The multimodal early warning release system built on the BIM+GIS digital twin platform realizes closed-loop management of monitoring-assessment-disposal through three-dimensional visualization early warning and OPC UA protocol, which greatly improves the efficiency of emergency response and system response speed, and provides dynamic and intelligent decision support for dam safety operation and maintenance.

[0028] Example 2: Please refer to Figure 2 As shown, this embodiment provides a specific implementation of spatial coordinate registration, including: establishing a three-dimensional coordinate mapping model through the topological relationship of the sensor deployment location, using a feature point matching algorithm to correct local coordinate deviations caused by installation errors, and realizing the spatial scale unification of displacement field and seepage field monitoring data.

[0029] Step S1 also includes a data quality control module, which identifies outlier points through wavelet packet decomposition, detects sensor faults using the local anomaly factor algorithm, and uses Kriging interpolation to perform spatiotemporal completion of missing data to ensure that the monitoring data integrity rate is not lower than the set threshold.

[0030] In this embodiment, a three-dimensional coordinate mapping model is established based on the topological relationship of the sensor deployment locations. Combined with a feature point matching algorithm, the local coordinate deviation caused by sensor installation errors is effectively corrected. This overcomes the bottleneck of correlation analysis between displacement field and seepage field monitoring data caused by spatial benchmark differences in traditional methods, achieving precise and unified spatial scale of multi-physics field data. At the same time, the data quality control module uses wavelet packet decomposition technology to identify outliers in the monitoring data, combines a local anomaly factor algorithm to detect sensor fault status in real time, and uses Kriging interpolation to perform multi-dimensional completion of spatiotemporally missing data, forming a highly complete monitoring dataset. This eliminates the impact of noise interference and data missingness on model accuracy, ensuring that the data integrity rate always meets the needs of dynamic analysis, and providing a reliable data foundation for subsequent multi-physics field coupling modeling and safety assessment.

[0031] Example 3: Please refer to Figure 3 As shown, the solution in this embodiment provides a specific method for parameter inversion, namely: using pore water pressure data and deformation data as observation vectors, iteratively updating the posterior estimates of permeability coefficient k and elastic modulus E through an adaptive Kalman filter, and simultaneously correcting the material parameter matrix of the finite element model.

[0032] The dynamic adjustment of seepage boundary conditions specifically includes: calculating the position of the seepage line using unsteady flow equations based on real-time upstream water level data, and inputting the calculation results as time-varying boundary conditions for transient seepage analysis into the finite element model.

[0033] Step S2 also includes a model confidence assessment module. When the parameter inversion error exceeds a preset threshold or the monitoring data residuals show a trend deviation, the model reconstruction process is automatically triggered, and alternative constitutive models are loaded through parallel computing for comparative verification.

[0034] In this embodiment, by using pore water pressure and deformation data as observation vectors, an adaptive Kalman filter is used to dynamically iteratively update the posterior estimates of the permeability coefficient k and elastic modulus E, and the finite element model parameter matrix is ​​corrected simultaneously. This improves the real-time performance and accuracy of parameter inversion. Combined with real-time upstream water level data, the position of the wetting line is dynamically calculated using the unsteady flow equation as the transient seepage boundary condition, enabling the model boundary condition to accurately match the unsteady seepage characteristics during flood rise and fall, thus enhancing the spatiotemporal adaptability of the seepage field simulation. The model confidence assessment module automatically triggers the model reconstruction mechanism by monitoring the inversion error and data residual trend in real time. It uses parallel computing to quickly load alternative constitutive models for comparison and verification, effectively avoiding the risk of single model failure. This forms a closed-loop optimization system of "dynamic correction - boundary synchronization - fault-tolerant reconstruction," ensuring the computational robustness and reliability of the multiphysics coupled model under all working conditions.

[0035] Example 4: Please refer to Figure 4As shown, the solution in this embodiment provides a method for constructing a three-dimensional time-varying safety envelope, including: dividing the flood evolution process into multiple working condition stages, determining the dynamic safety threshold boundary in each stage by the extreme value distribution of the seepage hydraulic gradient J, and generating the scaling factor of the three-dimensional envelope by combining the cumulative change of the displacement rate v.

[0036] In step S3, the hierarchical early warning decision tree is implemented based on a Bayesian network. Specifically, it includes: establishing conditional probability tables for three types of events: seepage anomaly, displacement mutation, and stress exceeding limits; calculating joint posterior probabilities based on real-time monitoring data; and triggering an early warning signal and generating a response plan when the probability value exceeds a preset threshold. Step S3 also includes visualization processing: overlaying seepage equipotential lines, displacement vector fields and stress cloud maps on the three-dimensional dam model, marking the over-limit areas with dynamic color levels, and simultaneously generating a safety assessment report and risk heat map.

[0037] In this embodiment, a three-dimensional time-varying safety envelope is dynamically constructed by the extreme value distribution of the seepage hydraulic gradient J and the cumulative change of displacement rate v. This breaks through the static limitations of the traditional fixed threshold method and realizes the intelligent scaling of the safety boundary with the rise and fall of water level, significantly improving the sensitivity of risk assessment under complex working conditions. A hierarchical early warning decision tree is constructed using a Bayesian network. By using conditional probability tables and joint posterior probability calculations for three types of events—seepage anomaly, displacement mutation, and stress over-limit—multi-parameter coupled risks are quantified, enhancing the autonomous reasoning ability for sudden working conditions, effectively reducing the false alarm rate and shortening the early warning response time. Combined with the visualization processing of the three-dimensional dam model, seepage equipotential lines, displacement vector fields, and stress cloud maps are dynamically superimposed. The over-limit areas are identified by dynamic color levels, and a risk heat map is generated, transforming multi-dimensional data into an intuitive spatiotemporal evolution map. This assists decision-makers in accurately locating potential hazard areas. The synchronously output safety assessment report and disposal plan form a closed-loop feedback of "data-model-decision," comprehensively improving the scientific nature of dam safety management and the efficiency of emergency response.

[0038] Example 5: Please refer to Figure 5 As shown, this embodiment provides a specific implementation of multi-source risk coupling analysis, including: establishing a transfer function model of environmental factors and dam response, generating random samples through Monte Carlo simulation, extracting key risk factors using principal component analysis, and constructing a three-dimensional risk surface that includes safety factor, reliability index, and failure probability.

[0039] In this embodiment, by establishing a transfer function model of environmental factors and dam response, the coupling mechanism of multi-source risks is quantified, overcoming the limitations of traditional single-factor analysis. Combined with Monte Carlo simulation to generate random samples, it covers low-probability, high-hazard events under extreme conditions, improving the completeness and reliability of risk assessment. Principal component analysis is used to extract key risk factors, focusing on dominant threat elements and reducing multidimensional data redundancy. Finally, a three-dimensional risk surface is constructed, integrating deterministic analysis and probabilistic assessment methods, intuitively revealing the multidimensional evolution of dam safety status under complex conditions. This provides a quantitative decision-making basis with both theoretical depth and engineering applicability for flood control scheduling and engineering reinforcement, significantly enhancing the dam's resilience in extreme complex disaster scenarios.

[0040] In summary, through the above embodiments of the present invention, the method of the present invention achieves spatiotemporal benchmark unification of multi-source monitoring data through feature point matching and sliding window cross-correlation analysis, effectively solving the data fusion deviation problem caused by sensor deployment differences in traditional methods. It employs an adaptive Kalman filter algorithm to dynamically invert permeability coefficient and elastic modulus parameters, and adjusts the boundary conditions of the finite element model in conjunction with real-time water level changes, significantly improving the dynamic simulation accuracy of the seepage field-displacement field-stress field coupled model. The innovatively constructed three-dimensional time-varying safety envelope surface and Bayesian network hierarchical early warning decision tree overcome the limitations of static evaluation using the fixed threshold method. It enhances the risk prediction capability during flood evolution through multi-parameter joint probabilistic inference. By establishing a multi-factor coupled analysis module for rainfall, earthquakes, and human activities, and using the equivalent linearization method and Green-Ampt model to simulate extreme conditions, it achieves accurate calculation of dam safety reserve coefficients in complex environments. The multi-modal early warning release system built based on the BIM+GIS digital twin platform, through three-dimensional visualization early warning and OPC... The UA protocol enables closed-loop management of monitoring, assessment, and response, significantly improving emergency response efficiency and system response speed. It provides dynamic and intelligent decision support for dam safety operation and maintenance. A three-dimensional coordinate mapping model is established based on the topological relationship of sensor deployment locations. Combined with a feature point matching algorithm, it effectively corrects local coordinate deviations caused by sensor installation errors. This overcomes the bottleneck in correlation analysis between displacement and seepage field monitoring data caused by spatial benchmark differences in traditional methods, achieving precise spatial scale unification of multi-physics data. Simultaneously, the data quality control module uses wavelet packet decomposition technology to identify outliers in the monitoring data, combines a local anomaly factor algorithm to detect sensor fault states in real time, and uses Kriging interpolation to perform multi-dimensional completion of spatiotemporally missing data, forming a highly complete monitoring dataset. This eliminates the impact of noise interference and data gaps on model accuracy, ensuring that data integrity consistently meets the needs of dynamic analysis, thus laying the foundation for subsequent multi-physics coupling modeling and... The safety assessment provides a reliable data foundation. By using pore water pressure and deformation data as observation vectors, an adaptive Kalman filter is employed to dynamically and iteratively update the posterior estimates of the permeability coefficient k and elastic modulus E, while simultaneously correcting the finite element model parameter matrix. This improves the real-time performance and accuracy of parameter inversion. Combined with real-time upstream water level data, the wetting line position is dynamically calculated using the unsteady flow equation as the transient seepage boundary condition. This allows the model boundary conditions to accurately match the unsteady seepage characteristics during flood rise and fall, enhancing the spatiotemporal adaptability of the seepage field simulation. The model confidence assessment module automatically triggers the model reconstruction mechanism by monitoring the inversion error and data residual trends in real time. Parallel computing is used to quickly load alternative constitutive models for comparison and verification, effectively avoiding the risk of single model failure. This forms a closed-loop optimization system of "dynamic correction - boundary synchronization - fault-tolerant reconstruction," ensuring the computational robustness and reliability of the multiphysics coupled model under all operating conditions.By dynamically constructing a three-dimensional time-varying safety envelope surface using the extreme value distribution of the seepage hydraulic gradient J and the cumulative change of displacement rate v, this method overcomes the static limitations of the traditional fixed threshold method, enabling intelligent scaling of the safety boundary with water level fluctuations. This significantly improves the sensitivity of risk assessment under complex conditions. A hierarchical early warning decision tree is constructed using a Bayesian network. Conditional probability tables and joint posterior probability calculations for three types of events—seepage anomalies, displacement mutations, and stress exceedances—quantify multi-parameter coupled risks, enhancing the autonomous reasoning ability for sudden conditions, effectively reducing false alarm rates and shortening early warning response time. Combined with the visualization of a three-dimensional dam model, seepage equipotential lines, displacement vector fields, and stress cloud maps are dynamically overlaid. Dynamic color-coded areas of exceedance are identified, and a risk heatmap is generated, transforming multi-dimensional data into an intuitive spatiotemporal evolution map. This assists decision-makers in accurately locating potential hazard areas, simultaneously generating a safety assessment report and a response plan. A closed-loop feedback mechanism of "data-model-decision" comprehensively improves the scientific nature of dam safety management and the efficiency of emergency response. By establishing a transfer function model of environmental factors and dam response, the mechanism of multi-source risk coupling is quantified, breaking through the limitations of traditional single-factor analysis. Monte Carlo simulation is used to generate random samples, covering low-probability, high-hazard events under extreme conditions, thus improving the completeness and reliability of risk assessment. Principal component analysis is used to extract key risk factors, focusing on dominant threat elements and reducing the redundancy of multidimensional data. Finally, a three-dimensional risk surface of safety factor, reliability index, and failure probability is constructed, integrating deterministic analysis and probabilistic assessment methods. This intuitively reveals the multidimensional evolution of dam safety status under complex conditions, providing a quantitative decision-making basis with both theoretical depth and engineering applicability for flood control scheduling and engineering reinforcement, significantly enhancing the dam's resilience in extreme and complex disaster scenarios.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for dam safety research and judgment based on multi-physical field simulation of monitoring data, characterized in that, The method comprises the following steps: S1, multi-source data acquisition and space-time alignment, generating monitoring data set unified in space-time reference; S2, dynamic updating of numerical model, establishing a multi-physical field coupling model of seepage field-displacement field-stress field; S3, safety evaluation and early warning decision, the hierarchical early warning decision tree based on Bayesian network carries out joint probability reasoning on the degree and duration of parameter deviation from the reference value, and triggers hierarchical early warning and response scheme; S4, multi-source risk coupling analysis, realizing the calculation of dam safety reserve coefficient under extreme working conditions; S5, multi-modal release of early warning information, based on BIM+GIS digital twin decision platform, three-dimensional visualization early warning, short message hierarchical push, emergency plan automatic loading three-level linkage, through OPC UA protocol and gate control system, drainage pump station realizes data intercommunication, constructs "monitoring-evaluation-early warning-disposal" closed-loop management system.

2. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 1, characterized in that, Step S1 is: arranging seepage pressure gauge, displacement gauge and stress gauge on the key section of the dam to form a three-dimensional monitoring network, matching the spatial coordinates of multi-source monitoring data through a feature point matching algorithm, and using a sliding window cross-correlation analysis to realize time series alignment, generating a monitoring data set unified in space-time reference; Step S2 is: based on real-time monitoring data, using adaptive Kalman filter algorithm to inverse dynamic parameters of dam permeability coefficient k and elastic modulus E, combining upstream water level change to dynamically adjust the seepage boundary condition and mechanical constraint of finite element model, establishing a multi-physical field coupling model of seepage field-displacement field-stress field; Step S3 is: combining seepage hydraulic gradient J, displacement rate v and stress parameters, constructing a three-dimensional time-varying safety envelope that automatically scales with flood evolution, based on Bayesian network hierarchical early warning decision tree, carrying out joint probability reasoning on the degree and duration of parameter deviation from the reference value, and triggering hierarchical early warning and response scheme; Step S4 is: establishing a multi-factor coupling analysis module of rainfall infiltration-seismic dynamic load-human activity, using equivalent linearization method to process random seismic dynamic load, simulating rainfall infiltration process through Green-Ampt model, constructing a complex risk scenario library containing typical working conditions, realizing the calculation of dam safety reserve coefficient under extreme working conditions.

3. The method of dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The spatial coordinate registration in step S1 specifically includes: establishing a three-dimensional coordinate mapping model through the topological relationship of sensor arrangement positions, correcting local coordinate deviations caused by installation errors using a feature point matching algorithm, and realizing the spatial scale unification of displacement field and seepage field monitoring data; Step S1 also includes a data quality control module, which identifies outlier points through wavelet packet decomposition, detects sensor faults using a local anomaly factor algorithm, and uses Kriging interpolation method to complete the time and space of missing data, ensuring that the monitoring data integrity rate is not less than the set threshold.

4. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The parameter inversion in step S2 is: taking pore water pressure data and deformation data as observation vectors, updating the posteriori estimation value of permeability coefficient k and elastic modulus E through adaptive Kalman filter, and synchronously correcting the material parameter matrix of finite element model.

5. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The dynamic adjustment of the seepage boundary condition in step S2 specifically comprises: according to the real-time data of the upstream water level, the position of the phreatic line is calculated by using the unsteady flow equation, and the calculation result is input into the finite element model as the time-varying boundary condition of the transient seepage analysis.

6. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The construction of the three-dimensional time-varying safety envelope surface in step S3 comprises: according to the flood evolution process, a plurality of working condition stages are divided, the dynamic safety threshold boundary is determined by the extreme value distribution of the seepage hydraulic gradient J in each stage, and the scaling coefficient of the three-dimensional envelope surface is generated by combining the cumulative change of the displacement rate v.

7. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The hierarchical early warning decision tree in step S3 is realized based on a Bayesian network, and specifically comprises: a conditional probability table of three types of events of seepage anomaly, displacement mutation and stress overrun is established, the joint posterior probability is calculated according to the real-time monitoring data, and when the probability value exceeds a preset threshold, an early warning signal is triggered and a disposal scheme is generated.

8. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, Step S2 further comprises a model confidence evaluation module, when the parameter inversion error exceeds a preset threshold or the monitoring data residual deviates from the trend, the model reconstruction process is automatically triggered, and the comparison and verification are performed by loading the alternative constitutive model through parallel computing. 9.The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, The multi-source risk coupling analysis in step S4 specifically comprises: a transfer function model of environmental factors and dam response is established, random samples are generated by Monte Carlo simulation, key risk factors are extracted by principal component analysis, and a three-dimensional risk surface including safety factor-reliability index-failure probability is constructed.

10. The method for dam safety research and judgment based on multi-physical field simulation of monitoring data according to claim 2, characterized in that, Step S3 further comprises a visual processing: the seepage equipotential lines, displacement vector field and stress nephogram are superimposed and displayed in the three-dimensional dam model, the overrun area is marked with dynamic color scale, and a safety evaluation report and a risk heat map are generated synchronously.

Images