A method for identifying potential hazards in dikes based on multi-source data fusion

Abstract

This invention relates to the field of levee management technology, and more particularly to a method for identifying levee hazards based on multi-source data fusion. The method includes dividing the levee structure into multiple monitoring and control sections according to hydrogeological units, and deploying multi-level sensor arrays within each control section; real-time acquisition of spatiotemporal sequence data from each sensor array to construct a field parameter functional including displacement, strain, seepage pressure, and temperature fields; extracting the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition to reflect the disorder of energy distribution in the levee body; calculating the coupling degree index based on the stress-seepage coupling constitutive equation to characterize the soil-water interaction intensity; and inputting the strain energy spectrum entropy value, the coupling degree index, and a pre-determined hydrodynamic load correction factor into the levee structure analysis model to obtain the levee safety and reliability. This method can improve the comprehensiveness, timeliness, and accuracy of levee hazard identification.

Description

Technical Field

[0001] This invention relates to the technical field of dike management, and in particular to a method for identifying dike hazards based on multi-source data fusion. Background Technology

[0002] As a vital water conservancy project for resisting floods and protecting the lives and property of people and property along the river, the structural stability of dikes is directly related to the reliability of the flood control system. However, during long-term operation, dikes are susceptible to various hidden dangers such as cracks, piping, and landslides due to multiple factors including changes in hydrological and meteorological conditions, geological tectonic movements, and human activities. If these hidden dangers are not identified and dealt with in a timely manner, they may lead to major disasters such as dike breaches.

[0003] Existing methods for identifying potential hazards in dikes primarily rely on monitoring with single-type sensors. Therefore, they can only acquire localized data from a single physical field within the dike. This single-field monitoring data is insufficient to reflect the complex coupling effects of multiple physical fields on the dike, leading to incomplete and delayed hazard identification results. Particularly under extreme conditions such as sudden water level changes and continuous heavy rainfall, existing methods lack sensitivity to early, subtle hazards, easily resulting in false alarms and missed detections, making it difficult to identify dike hazards in a timely and accurate manner. Summary of the Invention

[0004] This invention provides a method for identifying potential hazards in dikes based on multi-source data fusion, which can improve the comprehensiveness, timeliness and accuracy of the identification of potential hazards in dikes and can effectively solve the problems in the background technology.

[0005] To achieve the above objectives, in a first aspect, the present invention provides a method for identifying potential hazards in dikes based on multi-source data fusion, comprising:

[0006] The dike structure is divided into multiple monitoring and control sections according to hydrogeological units, and multi-level sensor arrays are deployed in each control section.

[0007] Real-time acquisition of spatiotemporal sequence data from each sensing array to construct a field parameter functional including displacement field, strain field, osmotic pressure field and temperature field;

[0008] The strain energy spectrum entropy value of the field parameter functional is extracted by spatiotemporal singular value decomposition to reflect the disorder of the energy distribution of the embankment; and the coupling degree index is calculated based on the stress-seepage coupling constitutive equation to characterize the soil-water interaction intensity.

[0009] The strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor are input into the embankment structure analysis model to obtain the embankment safety and reliability.

[0010] Based on the comparison between the safety and reliability of the embankment and the preset health threshold, a graded early warning response is triggered.

[0011] In conjunction with the first aspect, in one possible design, the method of dividing the levee structure into multiple monitoring and control sections according to hydrogeological units includes:

[0012] The dikes are divided into zones based on differences in engineering characteristic parameters along the dike line. These engineering characteristic parameters include the dike height gradient change rate, the dike body compaction distribution coefficient, the dike foundation seepage stability index, and the bank slope scour coefficient.

[0013] An evaluation model is established to weight and score each parameter. When the difference in the comprehensive score between adjacent sections exceeds a preset threshold, the boundary of the monitoring and control section is delineated. The length of a single monitoring and control section is determined based on the uniformity of the embankment structure, and each control section is ensured to contain a complete micro-geomorphic unit.

[0014] In conjunction with the first aspect, in one possible design, the multi-stage sensing array includes a surface crack gauge and a soil moisture monitor deployed on the top of the dike, a distributed fiber optic strain gauge and a matrix piezometer embedded inside the dike body, and a pore water pressure sensor array deployed at a set depth below the groundwater surface.

[0015] In conjunction with the first aspect, one possible design involves preprocessing the spatiotemporal sequence data of each sensing array when constructing the field parameter functional.

[0016] In conjunction with the first aspect, in one possible design, the extraction of the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition includes:

[0017] The spacetime matrix M∈R of the field parameter functional T×S Perform eigenvalue decomposition M = UΣV T , where T is the number of time sampling points, S is the number of spatial monitoring points, U is the left singular matrix, Σ is the diagonal singular value matrix, and V is the right singular matrix;

[0018] The strain energy spectrum entropy value is calculated using the following formula:

[0019]

[0020] in, is the element on the main diagonal of the singular value matrix, and k is the number of singular values ​​whose cumulative contribution rate exceeds a preset value.

[0021] In conjunction with the first aspect, in one possible design, the stress-seepage coupling constitutive equation is:

[0022]

[0023] Where CSC represents the coupling index; δ str δ represents the rate of change of the strain field. pre Represents the osmotic pressure gradient field; ks The saturated permeability coefficient of the soil is represented by t; t represents the time variable. This represents the Hamiltonian operator.

[0024] In conjunction with the first aspect, in one possible design, the hydrodynamic load correction factor is determined by the water level fluctuation and the rainfall intensity index, and the calculation formula for the hydrodynamic load correction factor is:

[0025] Φ(W)=1+λ1·|ΔH river |+λ2·I rain ;

[0026] Where Φ(W) represents the hydrodynamic load correction factor; ΔH river Indicates the water level fluctuation; I rain λ represents the rainstorm intensity index; λ1 represents the water level sensitivity coefficient; λ2 represents the rainfall sensitivity coefficient.

[0027] In conjunction with the first aspect, in one possible design, the calculation formula for the embankment structure analysis model is as follows:

[0028]

[0029] Among them, H SD Φ(W) represents the safety and reliability of the embankment; SES represents the strain energy spectrum entropy value; CSC represents the coupling index; Φ(W) represents the hydrodynamic load correction operator.

[0030] In conjunction with the first aspect, in one possible design, the method for determining the water level sensitivity coefficient λ1 and the rainfall sensitivity coefficient λ2 includes:

[0031] Based on a dataset of historical levee breach cases, a coefficient optimization model was established.

[0032]

[0033] The constraints are λ1 + λ2 = 1, 0 < λ1, λ2 < 1, where N is the number of historical cases, and H... SDm To measure health status, To predict health status.

[0034] Secondly, the present invention also provides a levee hazard identification system based on multi-source data fusion, comprising:

[0035] The monitoring and control section division module is used to divide the dike structure into multiple monitoring and control sections according to hydrogeological units, and to deploy multi-level sensor arrays in each control section.

[0036] The spatiotemporal sequence data acquisition and field parameter construction module is used to acquire spatiotemporal sequence data of each sensing array in real time and construct field parameter functionals including displacement field, strain field, osmotic pressure field and temperature field.

[0037] The feature index calculation module is used to extract the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition; and to calculate the coupling degree index based on the stress-seepage coupling constitutive equation.

[0038] The safety and reliability assessment module is used to input the strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor into the embankment structure analysis model to obtain the safety and reliability of the embankment.

[0039] The graded early warning response triggering module is used to trigger a graded early warning response based on the comparison result between the safety and reliability of the embankment and a preset health threshold.

[0040] The technical solution of this invention can achieve the following technical effects:

[0041] By forming a closed-loop chain of dike zoning, multi-level sensor data acquisition, field parameter functional construction, feature index extraction, and health assessment, single physical field data such as displacement, strain, and seepage pressure are transformed into comprehensive characteristics reflecting the overall structural state of the dike through spatiotemporal coupling analysis. This reduces the problem of misjudgment of hidden dangers caused by data bias. By characterizing energy disorder by strain energy spectrum entropy and quantifying the intensity of soil-water interaction by coupling index, and combined with dynamic correction of hydrodynamic loads, it is possible to capture, dynamically track, and accurately assess initial weak hidden dangers under extreme conditions, thereby improving the flood control system's ability to proactively prevent and control complex disasters. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the method for identifying potential hazards in dikes based on multi-source data fusion in this invention.

[0043] Figure 2 This is a structural block diagram of the dike hazard identification system based on multi-source data fusion in this invention; Detailed Implementation

[0044] This application will now be described with reference to the accompanying drawings.

[0045] like Figure 1 As shown, the present invention provides a method for identifying potential hazards in dikes based on multi-source data fusion, which specifically includes the following steps:

[0046] Step S1: Divide the dike structure into multiple monitoring and control sections according to hydrogeological units, and deploy multi-level sensor arrays in each control section;

[0047] Step S2: Collect spatiotemporal sequence data of each sensing array in real time and construct a field parameter functional including displacement field, strain field, pressure field and temperature field.

[0048] Step S3: Extract the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition to reflect the disorder of the energy distribution of the embankment; and calculate the coupling degree index based on the stress-seepage coupling constitutive equation to characterize the soil-water interaction intensity.

[0049] Step S4: Input the strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor into the embankment structure analysis model to obtain the embankment safety and reliability.

[0050] Step S5: Based on the comparison result between the safety and reliability of the embankment and the preset health threshold, trigger a graded early warning response.

[0051] In this embodiment, by dividing the monitoring and control sections according to hydrogeological units and deploying multi-level sensing arrays, it is possible to simultaneously acquire multi-physical field data such as the displacement field, strain field, seepage pressure field, and temperature field of the embankment. Compared with single-physical field monitoring, multi-source field parametric functionals can comprehensively characterize the multi-field coupling effect of the embankment in complex environments, avoiding misjudgment or omission of hidden dangers due to data bias. By extracting the strain energy spectrum entropy value through spatiotemporal singular value decomposition, it is possible to effectively capture subtle disordered changes in the energy distribution of the embankment, reflecting the structural stability decline caused by initial weak hidden dangers. Based on stress-seepage coupling... The coupling degree index calculated by the constitutive equation can quantify the abnormal trend of soil-water interaction intensity. Combining the two and introducing the hydrodynamic load correction factor can make the safety and reliability assessment of the embankment more in line with actual working conditions and improve the sensitivity to early hidden dangers under extreme conditions. At the same time, the embankment structure analysis model integrates multi-source characteristic parameters, and the output health measurement results can be directly compared with preset thresholds to achieve automated analysis. The graded early warning response mechanism can trigger corresponding disposal strategies according to the health level, avoid the problems of delayed early warning or over-response, and improve the efficiency and pertinence of embankment safety management.

[0052] In some embodiments of the present invention, the engineering characteristics of different locations along the dike vary, making it difficult to accurately identify potential hazards through unified monitoring. Dividing the area into zones based on engineering characteristic parameters and establishing monitoring and control sections allows for more targeted monitoring and analysis, specifically including:

[0053] Step S11: Divide the dike into zones based on the differences in engineering characteristic parameters along the dike. The engineering characteristic parameters include the dike height gradient change rate, the dike body compaction distribution coefficient, the dike foundation seepage stability index, and the bank slope scour coefficient.

[0054] In this step, the levee height gradient change rate reflects the variation in levee height along its length. A large levee height gradient indicates that the area may be subjected to varying water pressures and soil self-weight stresses, increasing the complexity and instability of the levee structure. This parameter needs to be obtained through topographic surveys and data calculations. The levee compaction distribution coefficient reflects the uniformity of soil compaction. Insufficient compaction or uneven distribution will result in differences in the density and strength of the levee soil. Under the influence of external forces such as floods, weak points are prone to deformation or even failure. This parameter also needs to be obtained through... Combining on-site testing and statistical analysis, the embankment foundation permeability stability index is used to assess the embankment foundation's ability to resist seepage damage. When the embankment foundation permeability stability index is low, under high water levels, the embankment foundation is prone to seepage damage phenomena such as piping and soil erosion, posing a threat to the overall stability of the embankment. This parameter needs to be obtained in conjunction with geological surveys and seepage calculations. The bank slope erosion coefficient reflects the degree of bank slope erosion caused by water flow. In areas with severe bank slope erosion, the embankment soil is gradually eroded, the embankment slope stability decreases, and landslides and other hazards may occur. This parameter needs to be obtained in conjunction with hydrological monitoring and topographic comparison.

[0055] Step S12: By establishing an evaluation model, each parameter is weighted and scored. When the difference in the comprehensive score between adjacent sections exceeds a preset threshold, the boundary of the monitoring and control section is delineated. The length of a single monitoring and control section is determined based on the uniformity of the embankment structure, and it is ensured that each control section contains a complete micro-geomorphic unit.

[0056] In this step, the determination of the preset threshold needs to comprehensively consider historical data, engineering experience, and numerical simulation results. The length of a single monitoring control section is determined based on the uniformity of the levee structure. For sections with a relatively uniform levee structure and gradual changes in parameters, the length of the monitoring control section can be appropriately increased. However, for sections with complex levee structures and large parameter variations, the length of the monitoring control section should be shortened accordingly. It is also essential to ensure that each control section contains a complete micro-geomorphic unit, such as a complete river bend or a complete beach, to ensure that the levee condition of the area can be fully reflected in subsequent monitoring. This will ensure that the levee structure and hydrogeological conditions within each monitoring control section are relatively consistent, thereby improving the targeting and accuracy of the monitoring.

[0057] In some embodiments of the present invention, the hidden dangers in different parts of the dike manifest themselves differently, requiring multiple types of sensors to monitor from different angles to obtain comprehensive information. The multi-stage sensing array includes:

[0058] a) Surface crack gauge: Deployed on the top of the dike to monitor changes in cracks on the surface of the dike. Cracks are a direct manifestation of damage to the dike structure, and the development of early micro-cracks may indicate the existence of internal hidden dangers. The surface crack gauge measures the changes in the width, length, and depth of cracks in real time through a high-precision displacement sensor. During installation, both ends of the crack gauge need to be fixed to the dike body on both sides of the crack to ensure that the sensor can accurately detect changes in the opening and closing of the crack. At the same time, in order to comprehensively monitor the distribution of cracks on the top of the dike, the surface crack gauges should be evenly distributed at a certain interval, and the spacing should be determined according to the width of the dike top and the historical occurrence of cracks.

[0059] b) Soil Moisture Monitoring Instrument: Deployed at different elevations on the top of the embankment and the slope, it is used to monitor the soil moisture content and rate of change in moisture in real time. When there is a potential seepage risk inside the embankment, the seepage water will change the moisture content distribution of the surrounding soil, causing abnormal fluctuations in local soil moisture data, such as a sudden increase or sustained high moisture content. The soil moisture monitoring instrument can capture abnormal changes in soil moisture content by collecting parameters such as soil volumetric water content and soil water potential at high frequency. During deployment, monitoring points are set up at preset intervals along the embankment slope. Each monitoring point is buried with a set number of soil moisture sensors at different depths along the direction perpendicular to the embankment slope, forming a three-dimensional monitoring network. The sensors use time-domain reflectometry or frequency-domain reflectometry to ensure stable operation in different soil textures. The monitoring frequency is adjusted according to the season.

[0060] c) Distributed fiber optic strain gauges: Embedded within the embankment, these gauges monitor strain changes within the embankment. During stress loading, the embankment experiences strain; abnormal strain changes may indicate stress concentration or damage to the embankment structure. Distributed fiber optic strain gauges utilize optical time-domain reflectometry (OTDR) to calculate strain values ​​by measuring changes in the scattering characteristics of light within the fiber. During installation, the fiber optic strain gauges must be laid out in layers along the embankment's depth. Each layer should be laid horizontally and evenly distributed across the embankment's cross-section to comprehensively acquire strain information from different locations within the embankment. The embedding depth of the fiber optic strain gauges is determined based on the embankment height.

[0061] d) Matrix piezometer: Embedded inside the embankment, it is used to monitor the distribution of seepage pressure inside the embankment. Seepage pressure is an important factor leading to seepage hazards such as piping and soil erosion in embankments. The matrix piezometer consists of multiple seepage sensors arranged in a matrix. During installation, the matrix piezometer should be buried in key parts of the embankment where seepage may occur, such as the junction of the embankment and the foundation, and the water-facing side of the embankment. The spacing between the sensors is determined according to the size of the embankment and the seepage characteristics.

[0062] e) Pore water pressure sensor array: Deployed at a predetermined depth below the groundwater level to monitor changes in pore water pressure in the foundation and lower part of the embankment below the groundwater level; changes in pore water pressure directly affect the effective stress of the soil, and thus the stability of the embankment; each sensor in the pore water pressure sensor array adopts the vibrating wire or piezoresistive principle, and can measure the pore water pressure value in real time. The determination of the predetermined depth needs to comprehensively consider the range of groundwater level changes, the depth of the foundation, and engineering experience; the sensor array should be evenly distributed on the plane, and the spacing should be determined according to the width of the foundation and geological conditions.

[0063] In some embodiments of the present invention, the original spatiotemporal sequence data collected by each sensing array may have problems such as noise, error or incompleteness, which will affect the subsequent analysis and calculation results. If the unprocessed original data is used directly for analysis, it will lead to distortion of the field parameter functional, which will affect the accuracy of feature index extraction and ultimately cause deviation in the hazard identification results. Therefore, preprocessing is required when constructing the field parameter functional, specifically including:

[0064] Step S21: Remove noise and outliers from the original data through data cleaning. For continuous data acquired by surface crack gauges and distributed fiber strain gauges, wavelet threshold denoising algorithms can be used to filter noise. This algorithm decomposes the data into wavelet coefficients of different frequencies, sets a threshold to suppress high-frequency noise coefficients, and then uses wavelet reconstruction to recover the true signal. For time-series soil moisture data acquired by soil moisture monitoring instruments, such as soil volumetric water content and soil water potential, a moving average filtering algorithm can be used to eliminate random noise and drift errors in the data, while retaining abrupt changes and trend characteristics of abnormal moisture changes. The filtering window length is reasonably set according to the monitoring frequency and the rate of soil moisture change. When handling outliers, a detection method based on statistical criteria is used. When the deviation of a data point from the mean exceeds the normal range, it is judged as an outlier and replaced with interpolated data from adjacent time points to ensure the continuity of the data sequence.

[0065] Step S22: Through spatiotemporal alignment, data from different sensors are synchronized in both time and space dimensions. In the time dimension, using a high-precision clock as a reference, the acquisition timestamps of each sensor array are uniformly calibrated, adjusting the time resolution of all data to a unified standard. Sensor data with acquisition frequencies higher than this resolution are downsampled, while data with frequencies lower than this resolution are supplemented using linear interpolation to ensure consistency of the time series. In the spatial dimension, based on the geographic coordinate system of the monitoring and control section, a spatial index matrix is ​​established, mapping the discrete measurement points of the distributed fiber strain gauge and the array data of the matrix piezometer to a unified three-dimensional spatial grid. Spatial interpolation algorithms are used to fill the data gaps in the grid, achieving spatial coordinate alignment of different physical field data.

[0066] Step S23: Eliminate the dimensional differences between different physical field data by feature normalization; that is, use a standardization method to normalize the processed spatiotemporal sequence data and compress the data into a specific interval.

[0067] Step S24: Repair missing data caused by sensor failure or transmission interruption; when the data loss in a single time period is minor, the sliding window mean method can be used to fill it in, and the window size is set according to the data change frequency; when the loss is moderate, the spatial co-interpolation method can be used to repair it by combining the spatial correlation of adjacent monitoring points; when the loss is severe, a sensor failure warning is triggered, and the trend characteristics of historical data from the same period are called to supplement it, and the validity of the data in this area is verified in the subsequent data collection process.

[0068] Through the above steps, the preprocessed spatiotemporal sequence data can more realistically reflect the physical field distribution characteristics of the embankment; at the same time, normalization processing can eliminate interference caused by the dimensions, and can avoid the problem of feature weight imbalance caused by differences in data magnitude; under extreme working conditions, the preprocessed data can more clearly present the characteristic signals of the initial weak hidden dangers.

[0069] In some embodiments of the present invention, the strain energy spectrum entropy value can reflect the disorder of the energy distribution in the embankment. Extracting this value through spatiotemporal singular value decomposition of the field parameter functional helps in analyzing the state changes of the embankment. Specifically, the extraction of the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition includes:

[0070] The spacetime matrix M∈R of the field parameter functional T×S Perform eigenvalue decomposition M = UΣV T , where T is the number of time sampling points, S is the number of spatial monitoring points, U is the left singular matrix, Σ is the diagonal singular value matrix, and V is the right singular matrix;

[0071] The strain energy spectrum entropy value is calculated using the following formula:

[0072]

[0073] in, is the element on the main diagonal of the singular value matrix, and k is the number of singular values ​​whose cumulative contribution rate exceeds a preset value.

[0074] In this embodiment, the number of time sampling points T is determined based on the sensor's acquisition frequency and monitoring duration. A higher acquisition frequency and a longer monitoring duration can more accurately reflect data changes in the time dimension. The number of spatial monitoring points S depends on the number of sensor arrays deployed within the monitoring and control section and the distribution of monitoring points on each sensor array. The denser the monitoring points, the more comprehensive the spatial coverage. The left singular matrix U reflects the characteristic patterns in the time dimension, and the singular value matrix Σ contains diagonal elements. The importance of the corresponding singular vector is indicated by the right singular matrix V, which reflects the characteristic pattern in the spatial dimension. The preset value of k is usually determined based on engineering experience and data characteristics. Spatiotemporal singular value decomposition can effectively reduce dimensionality and simplify the data structure while preserving the main characteristics of the field parameter functional. The strain energy spectrum entropy value, as a quantitative indicator, can intuitively reflect the disorder of the energy distribution in the embankment. When the embankment shows initial minor hidden dangers, the energy distribution will change abnormally, and the entropy value will also change accordingly, thereby realizing early detection of hidden dangers.

[0075] In some embodiments of the present invention, during the operation of the dike, there is a complex interaction between the soil skeleton inside the dike and the pore water. This soil-water interaction is a core factor affecting the stability of the dike. When the dike is subjected to external loads such as water level changes and rainfall infiltration, the stress field and seepage field will influence each other. The coupling degree index can be calculated by the stress-seepage coupling constitutive equation to characterize the strength of this interaction. The stress-seepage coupling constitutive equation is as follows:

[0076]

[0077] Where CSC represents the coupling index; δ str δ represents the rate of change of the strain field. pre Represents the osmotic pressure gradient field; k s The saturated permeability coefficient of the soil is represented by t; t represents the time variable. This represents the Hamiltonian operator.

[0078] In this embodiment, δ str The deformation rate of the embankment structure under stress can be obtained through time series analysis of strain data collected by distributed fiber optic strain gratings deployed inside the embankment; δ pre The spatial rate of change of seepage pressure within the levee, characterizing the seepage pressure data collected by a matrix piezometer and pore water pressure sensor array, is obtained through spatial gradient algorithm processing; k s It is a key parameter reflecting the permeability of soil. It can be determined based on the geotechnical test data during the embankment survey stage and dynamically corrected in combination with the seepage data monitored on site. Used to describe the convergence or divergence of an osmotic pressure field; It represents the rate of change of the seepage gradient field over time, reflecting the dynamic evolution of the seepage field over time; through the stress-seepage coupling constitutive equation, it can more comprehensively reflect the complex state inside the embankment; when the embankment has initial minor hidden dangers, the soil-water interaction intensity will change abnormally, and the coupling degree index can keenly capture it, improving the sensitivity to initial hidden dangers. At the same time, in scenarios such as sudden changes in water level and continuous heavy rainfall, it can promptly reflect the hidden danger risks caused by the intensification of coupling effect inside the embankment.

[0079] In some embodiments of the present invention, the stability of the dike is affected by hydrodynamic loads, with water level fluctuations and rainfall intensity being two key factors. Significant water level changes alter the water pressure on the dike, affecting stress distribution and seepage within the dike. Continuous heavy rainfall increases the water content of the dike, reducing soil strength, and rainwater infiltration also changes the seepage pressure field. Therefore, a hydrodynamic load correction factor needs to be determined to correct the assessment results of the dike's safety and reliability, thereby improving the accuracy of hazard identification. Specifically, the calculation formula for the hydrodynamic load correction factor is as follows:

[0080] Φ(W)=1+λ1·|ΔH river |+λ2·I rain ;

[0081] Where Φ(W) represents the hydrodynamic load correction factor; ΔH river Indicates the water level fluctuation; I rain λ represents the rainstorm intensity index; λ1 represents the water level sensitivity coefficient; λ2 represents the rainfall sensitivity coefficient.

[0082] In this embodiment, ΔH river The difference in river water level along the dike over a certain period of time can be calculated using data collected in real time by water level monitoring stations deployed near the dike. The sampling frequency is determined based on the water level changes, and can be increased when water level changes are drastic. rain This parameter reflects the intensity of the rainstorm and can be calculated based on rainfall data provided by the meteorological department, combined with the specific topographic features of the area where the levee is located. It is usually represented by the rainfall per unit time. λ1 is used to measure the weight of the influence of water level changes on the safety and reliability of the levee, and λ2 is used to measure the weight of the influence of rainstorm intensity on the safety and reliability of the levee. By weighting the influence of water level fluctuation and rainstorm intensity using the water level sensitivity coefficient and the rainfall sensitivity coefficient, the influence weights of the two can be reasonably allocated according to the actual situation and historical data of different levee projects, so that the corrected safety and reliability of the levee is more in line with the actual levee condition. In terms of extreme conditions, this correction factor can sensitively reflect the drastic changes in hydrodynamic loads in extreme conditions such as sudden changes in water level and continuous heavy rainfall.

[0083] By integrating strain energy spectrum entropy, coupling degree index, and hydrodynamic load correction factor, a structural analysis model for the embankment is established, which can more comprehensively assess the health status of the embankment. Specifically, the calculation formula for the structural analysis model is as follows:

[0084]

[0085] Among them, H SD Φ(W) represents the safety and reliability of the embankment; SES represents the strain energy spectrum entropy value; CSC represents the coupling index; Φ(W) represents the hydrodynamic load correction operator.

[0086] In this embodiment, H SD The value reflects the stability of the embankment structure. The smaller the value, the greater the possibility of hidden dangers in the embankment and the worse its health condition.

[0087] In some embodiments of the present invention, the accuracy of the hydrodynamic load correction factor directly affects the assessment result of the levee's safety and reliability. The water level sensitivity coefficient λ1 and the rainfall sensitivity coefficient λ2 are key parameters for calculating the hydrodynamic load correction factor. Different levee projects are situated in different hydrogeological environments and have varying structural characteristics, resulting in different degrees of impact from water level and heavy rainfall on levee health. Using fixed sensitivity coefficient values ​​cannot adapt to the actual conditions of different levees, leading to inaccurate calculation of the hydrodynamic load correction factor and consequently affecting the accuracy of the levee's safety and reliability assessment. Therefore, it is necessary to determine the water level sensitivity coefficient λ1 and the rainfall sensitivity coefficient λ2 using scientific methods based on historical levee breach case datasets to improve the rationality and applicability of the hydrodynamic load correction factor. The method for determining the water level sensitivity coefficient λ1 and the rainfall sensitivity coefficient λ2 includes:

[0088] Based on a dataset of historical levee breach cases, a coefficient optimization model was established.

[0089]

[0090] The constraints are λ1 + λ2 = 1, 0 < λ1, λ2 < 1, where N is the number of historical cases, and H... SDm To measure health status, To predict health status.

[0091] The historical levee breach case dataset should cover various situations such as different hydrological and meteorological conditions, levee structure types, and geological environments to ensure that the dataset has broad representativeness and rich information content. By utilizing historical experience data, the determined sensitivity coefficients can better reflect actual engineering conditions, improve the accuracy of hydrodynamic load correction factor calculation, and thus enhance the reliability of levee safety reliability assessment results. In the solution process, various optimization algorithms, such as gradient descent and genetic algorithms, can be used to iteratively optimize the sensitivity coefficients until the optimal solution that minimizes the objective function value is found.

[0092] In some embodiments of the present invention, the graded early warning response triggered based on the comparison result between the safety reliability of the embankment and a preset health threshold is as follows:

[0093] The preset health thresholds are divided into three levels: Level 1, Level 2, and Level 3, corresponding to green, yellow, and red warning levels, respectively. The specific warning response mechanism for each level is as follows:

[0094] Green Alert: When the safety and reliability of the levee reaches or exceeds the first-level threshold, the levee is determined to be in a stable state with no significant hidden risks. At this time, routine inspection reminders can be pushed to the levee management department through the monitoring platform. The content includes the health values ​​of each monitoring and control section, the stability trend of the functional of the main field parameters, and recent maintenance suggestions. The management department carries out its work according to the daily inspection plan and updates the monitoring data summary table every 72 hours. There is no need to initiate the emergency response process.

[0095] Yellow Alert: If the health status is within the secondary threshold range, it indicates that there are potential minor hidden dangers in the embankment, abnormal fluctuations in the soil-water interaction intensity, and a slight increase in the disorder of energy distribution. At this time, the yellow alert signal is immediately triggered. In addition to sending a detailed analysis report to the management department, which includes the abrupt change points of strain energy spectrum entropy and abnormal periods of coupling degree index, the multi-level sensor array of the monitoring and control section is scheduled to enter the high-frequency acquisition mode. At the same time, the local embankment management and maintenance unit is notified to conduct targeted investigations, focusing on checking the seepage outlets, embankment slope deformation areas, and the surrounding environment of the sensor equipment deployment points in the control section. The investigation results are uploaded to the monitoring platform in real time. If no obvious hidden dangers are found during the investigation, the health status change trend is reviewed every 48 hours.

[0096] Red Alert: When the health level falls below the Level 3 threshold, there are serious structural hazards in the embankment, and initial signs of local piping, crack expansion, or slope instability may have already appeared. At this time, the red alert response should be activated immediately, sending audible and visual alarm signals and an emergency report containing the coordinates of the hazard location, real-time field parameter data curves, and peak values ​​of the coupling degree index to the management department, emergency command center, and local government. At the same time, the emergency monitoring equipment around the control section should be activated to collect displacement field and seepage pressure field data more frequently, and video surveillance equipment should be activated to continuously film the hazard area. The emergency command center should dispatch a professional technical team to the site within 1 hour to conduct detailed testing such as drilling and sampling and ground-penetrating radar scanning. Based on the test results, temporary measures such as sandbag seepage control and slope reduction should be taken immediately, and a warning area should be demarcated to prohibit unauthorized personnel from entering. The health assessment results should be updated every 15 minutes until the hazard is effectively controlled or the health level rises to above the Level 2 threshold.

[0097] In this embodiment, the determination of the preset health threshold needs to be comprehensively formulated based on the dike engineering level, historical hazard database, and safety redundancy requirements, and personalized adjustments should be made according to the characteristics of the dike engineering. For important dikes such as primary dikes protecting the core urban area, the primary threshold can be raised to increase safety redundancy. For secondary dikes such as farmland protection dikes, the primary threshold can be lowered to balance monitoring costs. Threshold calibration needs to be performed every certain period of time. By comparing the correlation between the measured health value and the probability of historical hazard occurrence during that period, the least squares method can be used to optimize the threshold boundary to ensure the accuracy of threshold division. At the same time, before the arrival of extreme weather such as typhoon season and flood season, the dynamic threshold correction coefficient needs to be temporarily activated to temporarily raise the lower limit of the secondary threshold to enhance the sensitivity to initial hazards.

[0098] like Figure 2 As shown, the present invention also provides a levee hazard identification system based on multi-source data fusion, which specifically includes the following modules;

[0099] The monitoring and control section division module is used to divide the dike structure into multiple monitoring and control sections according to hydrogeological units, and to deploy multi-level sensor arrays in each control section.

[0100] The spatiotemporal sequence data acquisition and field parameter construction module is used to acquire spatiotemporal sequence data of each sensing array in real time and construct field parameter functionals including displacement field, strain field, osmotic pressure field and temperature field.

[0101] The feature index calculation module is used to extract the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition; and to calculate the coupling degree index based on the stress-seepage coupling constitutive equation.

[0102] The safety and reliability assessment module is used to input the strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor into the embankment structure analysis model to obtain the safety and reliability of the embankment.

[0103] The graded early warning response triggering module is used to trigger a graded early warning response based on the comparison result between the safety and reliability of the embankment and a preset health threshold.

[0104] In this embodiment, by dividing the monitoring and control sections according to hydrogeological units and deploying multi-level sensor arrays, comprehensive acquisition of multi-physical field data such as displacement, strain, seepage pressure, and temperature can be achieved, overcoming the limitations of single-sensor monitoring. By extracting the strain energy spectrum entropy value through spatiotemporal singular value decomposition and calculating the coupling degree index by combining the stress-seepage coupling constitutive equation, the characteristics of hidden dangers under multi-field coupling effects can be captured, improving the sensitivity to initial weak hidden dangers. By fusing multiple feature indicators with hydrodynamic load correction factors for safety and reliability assessment and triggering graded early warnings, the problems of false alarms and missed judgments under extreme working conditions can be reduced, enabling timely and accurate identification and graded response of dike hidden dangers, and improving the comprehensiveness, timeliness, and reliability of hidden danger identification.

[0105] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for identifying potential hazards in dikes based on multi-source data fusion, characterized in that, include: The dike structure is divided into multiple monitoring and control sections according to hydrogeological units, and multi-level sensor arrays are deployed in each control section. Real-time acquisition of spatiotemporal sequence data from each sensing array to construct a field parameter functional including displacement field, strain field, osmotic pressure field and temperature field; The strain energy spectrum entropy value of the field parameter functional is extracted by spatiotemporal singular value decomposition to reflect the disorder of the energy distribution of the embankment; and the coupling degree index is calculated based on the stress-seepage coupling constitutive equation to characterize the soil-water interaction intensity. The strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor are input into the embankment structure analysis model to obtain the embankment safety and reliability. Based on the comparison between the safety reliability of the embankment and the preset health threshold, a graded early warning response is triggered; The extraction of the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition includes: The spacetime matrix of the field parameter functional Perform eigenvalue decomposition Where T is the number of time sampling points, S is the number of spatial monitoring points, and U is the left singular matrix. V is a diagonal singular value matrix, and V is a right singular matrix; The strain energy spectrum entropy value is calculated using the following formula: ； in, represents the main diagonal elements of the singular value matrix, and k is the number of singular values ​​whose cumulative contribution rate exceeds a preset value; The stress-seepage coupling constitutive equation is: ； CSC represents the coupling index; δ str Indicates the rate of change of the strain field; δ pre Represents the osmotic pressure gradient field; k s The saturated permeability coefficient of the soil is represented by t; t represents the time variable. Represents the Hamiltonian operator; The hydrodynamic load correction factor is determined by the water level fluctuation and the rainfall intensity index, and the calculation formula for the hydrodynamic load correction factor is as follows: ； Where Φ(W) represents the hydrodynamic load correction factor; ΔH river Indicates the water level fluctuation; I rain λ represents the rainstorm intensity index; λ1 represents the water level sensitivity coefficient; λ2 represents the rainfall sensitivity coefficient.

2. The method for identifying potential hazards in dikes based on multi-source data fusion according to claim 1, characterized in that, The method for dividing the dike structure into multiple monitoring and control sections according to hydrogeological units includes: The dikes are divided into zones based on differences in engineering characteristic parameters along the dike line. These engineering characteristic parameters include the dike height gradient change rate, the dike body compaction distribution coefficient, the dike foundation seepage stability index, and the bank slope scour coefficient. An evaluation model is established to weight and score various engineering characteristic parameters. When the difference in comprehensive scores between adjacent sections exceeds a preset threshold, the boundary of the monitoring and control section is delineated. The length of a single monitoring and control section is determined based on the uniformity of the embankment structure, and each control section is ensured to contain a complete micro-geomorphic unit.

3. The method for identifying potential hazards in dikes based on multi-source data fusion according to claim 1, characterized in that, The multi-stage sensing array includes a surface crack gauge and a soil moisture monitor deployed on the top of the dike, a distributed fiber optic strain gauge and a matrix piezometer embedded inside the dike body, and a pore water pressure sensor array deployed at a set depth below the groundwater surface.

4. The method for identifying potential hazards in dikes based on multi-source data fusion according to claim 1, characterized in that, When constructing the field parameter functional, preprocessing of the spatiotemporal sequence data of each sensing array is also included. The preprocessing methods include: Data cleaning removes noise and outliers from the original data; By aligning data from different sensors in time and space, data from different sensors can be synchronized in both time and space. By normalizing features, the dimensional differences between different physical field data are eliminated; Repair missing data caused by sensor failure or transmission interruption.

5. The method for identifying potential hazards in dikes based on multi-source data fusion according to claim 1, characterized in that, The calculation formula for the embankment structure analysis model is as follows: ； Among them, H SD Φ(W) represents the safety and reliability of the embankment; SES represents the strain energy spectrum entropy value; CSC represents the coupling index; Φ(W) represents the hydrodynamic load correction operator.

6. The method for identifying potential hazards in dikes based on multi-source data fusion according to claim 5, characterized in that, The methods for determining the water level sensitivity coefficient λ1 and the rainfall sensitivity coefficient λ2 include: Based on a dataset of historical levee breach cases, a coefficient optimization model was established. ； The constraints are , Where N is the number of historical cases. To measure health status, To predict health status.

7. A levee hazard identification system based on multi-source data fusion, wherein the system is applied to the levee hazard identification method based on multi-source data fusion as described in claim 1, characterized in that, The system includes: The monitoring and control section division module is used to divide the dike structure into multiple monitoring and control sections according to hydrogeological units, and to deploy multi-level sensor arrays in each control section. The spatiotemporal sequence data acquisition and field parameter construction module is used to acquire spatiotemporal sequence data of each sensing array in real time and construct field parameter functionals including displacement field, strain field, osmotic pressure field and temperature field. The feature index calculation module is used to extract the strain energy spectrum entropy value of the field parameter functional through spatiotemporal singular value decomposition; and to calculate the coupling degree index based on the stress-seepage coupling constitutive equation. The safety and reliability assessment module is used to input the strain energy spectrum entropy value, the coupling degree index, and the pre-determined hydrodynamic load correction factor into the embankment structure analysis model to obtain the safety and reliability of the embankment. The graded early warning response triggering module is used to trigger a graded early warning response based on the comparison result between the safety and reliability of the embankment and a preset health threshold.

Images