Embankment flood control risk dynamic weight and threshold collaborative evaluation and disposal method

Abstract

The invention provides an embankment flood control risk dynamic weight and threshold collaborative evaluation and disposal method, and belongs to the field of flood control construction. Aiming at the problems that static weight and dynamic hydrological environment are incompatible, basic data are scarce and poor in adaptability, a threshold value is fixed, secondary misjudgment is easily caused, and evaluation and practical operation are disjointed in the prior art, real-time data support is provided by constructing a three-layer data system, and a basic weight and dynamic correction dual-module weighting system is adopted to adapt to environment change, so that real-time evaluation and practical operation are realized. In combination with scene adaptation and a special working condition two-factor threshold calibration system, weight and threshold collaboration is realized, a lightweight architecture of index grouping and parallel computing is established to improve real-time performance, and practical operation guidance is converted through risk level, short plate index and disposal template three-dimensional closed-loop output. According to the method, quantification accuracy, emergency real-time performance and basic feasibility are taken into consideration, the high-risk dike misjudgment rate is reduced, the disposal suggestion landing rate is improved, different types of dikes and complex working conditions are adapted, operability is high, and engineering popularization is facilitated.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of flood control construction, in particular to a dike flood control risk dynamic weight and threshold value cooperative evaluation and disposal method. BACKGROUND

[0002] Flood disaster is an important disaster type that restricts urban development and threatens people's life and property safety. As a core component of the flood control engineering system, the scientificity and practicality of dike flood control risk assessment directly affect the effectiveness of flood control work.

[0003] Current dike flood control risk calculation technology mostly uses static analytic hierarchy process (AHP) for weighting. Once the weight is determined, it cannot dynamically respond to changes in hydrological environment, leading to incompatibility between the static weight assignment mechanism of AHP algorithm and the time sequence of dynamic hydrological environment and real-time monitoring data, which seriously affects the accuracy of evaluation. Traditional evaluation methods require the collection of complete multi-dimensional data, which has a high requirement for the data collection capacity of the grassroots level. However, there are problems such as data scarcity and difficulty in collection at the grassroots level. Meanwhile, the threshold value in the prior art is mostly fixed, and it does not form a cooperative adaptation with the dynamically adjusted weight. When the weight is adjusted due to environmental changes, the fixed threshold value cannot reflect the actual risk level, which easily leads to secondary misjudgment and further reduces the evaluation accuracy.

[0004] In the prior art, some schemes try to improve the evaluation accuracy by mixing weighting such as entropy weighting with trapezoidal fuzzy numbers or a wide-dimensional index system, but there are problems such as complex technical logic, deviation from grassroots operation scenarios, and easy data overload. Another scheme focuses only on improving the calculation speed or optimizing the index system, and cannot solve the contradiction between quantitative accuracy, emergency real-time performance and low data dependence of the grassroots level, so it cannot form a complete solution. Therefore, there is an urgent need for an integrated technical solution that can solve the above problems and fill the gap in the prior art. SUMMARY

[0005] The main purpose of the present application is to provide a dike flood control risk dynamic weight and threshold value cooperative evaluation and disposal method, which takes into account the quantitative accuracy of dike flood control risk assessment and the real-time performance of flood season emergency decision-making and the feasibility of grassroots deployment, and solves the problems of high risk dike misjudgment rate, risk assessment lag and insufficient grassroots disposal suggestion landing rate caused by the incompatibility between the static weight assignment mechanism of AHP algorithm and the time sequence of dynamic hydrological environment and real-time monitoring data, and the coupling failure with the engineering environment of data scarcity at the grassroots level.

[0006] To solve the above technical problems, the technical solution adopted by the present application is: a dike flood control risk dynamic weight and threshold value cooperative evaluation and disposal method, which comprises the following steps: S1. Construct a three-tiered data system covering levee risk assessment, including core essential data, important auxiliary data, and real-time hydrological data, to provide real-time data support for subsequent dynamic weight calculation and threshold calibration. S2. Construct a dual-module weighting system of basic weights and dynamic corrections, and dynamically adjust the weights through real-time hydrological data to accurately reflect environmental indicators in real time; S3. Construct a two-factor calibration system that integrates scenario adaptation factors with special working condition factors, so that the threshold can be dynamically adjusted according to dynamic weights and actual working conditions. S4. Construct an architecture that features indicator grouping, pipelined parallelism, and lightweight computing to shorten the evaluation cycle; S5. Construct a three-dimensional closed-loop output system of risk level, weakness indicators, and handling templates to transform quantitative results into practical guidelines that can be directly implemented.

[0007] In the preferred embodiment, in step S1: The core essential data includes structural safety data and levee crest elevation data, among which Structural safety data includes the anti-sliding stability of the flood control wall slope, the structural stability of the flood control wall, and the seepage safety performance, corresponding to the structural bearing capacity and disaster resistance capacity of the embankment itself; The data on the top elevation of the dike includes the design top elevation of the dike and the top elevation of the existing flood control wall. It is used to calculate the degree of insufficient dike top elevation and to determine the risk of overflow. Key auxiliary data include engineering quality data, hydrodynamic condition data, riverbed condition data, and flood control channel connectivity data, among which: Engineering quality data includes wall tilting, subsidence, and wall aging, which are used to assess the long-term operational health of the dike; Hydrodynamic data include river flow velocity, corresponding to the scouring force and impact of floods on dikes; Riverbed data includes river cross-sectional morphology, river scouring and deposition, distribution of river channel grooves, and slope coefficient; Data on the connectivity of flood control channels includes the detour length of flood control channels, which corresponds to traffic accessibility and rescue efficiency during emergency rescue operations. Real-time hydrological data includes riverbed erosion and deposition rate data and ship wave scouring intensity data, among which: Riverbed scour and deposition rate data include the change in the thickness of riverbed scour or deposition per unit time, corresponding to the impact of the dynamic evolution of riverbed morphology on the foundation of dikes. Ship wave scouring intensity data includes the scouring force of waves generated by ship navigation on the toe of the dike, corresponding to dynamic risk factors in plain river dikes.

[0008] In the preferred embodiment, step S1 further includes the following steps: S11. Collect essential core data; The slope anti-sliding stability was monitored by a total station for displacement monitoring, the seepage safety performance was tested by a permeameter in the field, and the structural stability was tested by an ultrasonic detector for internal defects. Slope anti-sliding stability is achieved by setting up multiple monitoring sections on the embankment slope, with multiple monitoring points on each section, and using a total station to collect three-dimensional coordinate data to calculate the displacement change between adjacent periods. Structural stability is assessed by using an ultrasonic testing instrument to set up testing points at intervals along the length of the embankment, scanning and testing the integrity of the foundation and pile foundation, and recording the location and extent of defects; The seepage safety performance was assessed by arranging multiple seepage test holes in the embankment, conducting constant head seepage tests using a permeameter, continuously collecting seepage flow rate and head difference data, and calculating the permeability coefficient. Simultaneously, an outlier handling mechanism is introduced using the 3σ criterion, and the mean is calculated: (1) in For displacement monitoring data, This represents the total number of displacement monitoring data. Sum of standard deviation: (2) Eliminate those that satisfy the following conditions: (3) Abnormal values ​​in displacement monitoring data; The elevation data of the embankment crest was measured using a combination of a GPS positioning instrument and a leveling instrument. Multiple monitoring points are set up along the top of the embankment at preset intervals. The plane coordinates of the monitoring points are determined by GPS positioning instrument. The elevation is measured by leveling instrument using nearby known elevations as references, and the current embankment top elevation is recorded. The design top elevation is obtained by consulting the embankment design drawings and the elevation difference is calculated.

[0009] S12. Collect important auxiliary data; Among them, the engineering quality data was collected by laser rangefinders to measure the tilt of the wall; the aging of the wall was visually inspected by drones equipped with high-definition cameras. The tilt angle of a wall is determined by setting up distance measuring points at the same height on both sides of the wall, measuring the horizontal distance, and then calculating the tilt angle. Wall aging is assessed by using drones to capture images of the wall surface and then using the YOLOv8 algorithm to automatically identify aging phenomena such as cracks and peeling, and quantify the project quality level according to the proportion of the affected area. The hydrodynamic data were obtained by measuring flow velocity using a Doppler current meter. By setting up a flow meter at the center of the river cross-section, measuring the flow velocity at intervals and layers, taking the average value of multiple measurements for each layer, and calculating the average flow velocity of the cross-section. The riverbed data was obtained by using a multibeam echo sounder to measure the riverbed topography, and the slope coefficient was measured in the field using a total station. By setting up multiple survey lines along the river, survey boats navigate the river to collect riverbed elevation data, generate a three-dimensional topographic model of the riverbed, and calculate the thickness of scour and sedimentation. Positive values ​​indicate sedimentation, while negative values ​​indicate scour. At the same time, measurement points are set up on the upstream and downstream slopes of the embankment to measure the horizontal distance and vertical height and calculate the slope coefficient. Data on the connectivity of flood control channels was collected using a trajectory recorder combined with map software. By planning the shortest route for flood control channels, measuring the ideal length, recording the actual route length on-site, and calculating the detour length.

[0010] S13. Collect real-time hydrological data; The riverbed scouring and deposition rate data were collected using a multibeam echo sounder system shared with the riverbed condition data acquisition system, combined with a water level gauge. The riverbed topography was measured using a multibeam echo sounder system, and the scouring and sedimentation thickness was calculated by comparing it with the previous measurement results. The scouring and sedimentation thickness was corrected by combining the average water level recorded by the water level gauge. The scouring and sedimentation rate was calculated by comparing it with the previous measurement time, with sedimentation being positive and scouring being negative.

[0011] The wave scouring intensity data of the ship was measured by combining wave height data with current velocity data using a wave height meter. Wave height meters were deployed near the toe of the breakwater to record the wave height and period of the waves as the ship's waves passed, and the wavelength was calculated. (4) in It is the acceleration due to gravity. For periodicity; Calculate wave speed using formula : (5) in Wavelength; scouring intensity Calculated using the formula: (6) in For wave height.

[0012] In the preferred embodiment, in step S2: The dual-module weighting system includes a basic weight matrix construction module and a dynamic correction module, wherein: The basic weight matrix is ​​constructed using the Analytic Hierarchy Process (AHP), consisting of a target layer (Z), an indicator layer (A), and a second-level indicator layer (B), where: The target layer (Z) is for calculating the flood control risk of the dikes; The indicator layer (A) includes four primary indicators: risk of dike safety and stability factors (A1), risk of overflow factors (A2), risk of river morphology factors (A3), and risk of emergency rescue and disaster relief factors (A4). The secondary indicator layer (B) includes 7 secondary indicators, of which A1 corresponds to engineering quality (B1) and structural safety (B2); A2 corresponds to the degree of insufficient dike crest elevation (B3) and surrounding land use (B4); A3 corresponds to hydrodynamic conditions (B5) and riverbed conditions (B6); and A4 corresponds to the flood control channel connectivity (B7). After the hierarchy is established, an expert scoring process is carried out. Experts are invited to use the 1-9 scale to compare the importance of each level of indicators. Here, 1 means that the two indicators are equally important, 3 means that the former is significantly more important than the latter, 5 means that the former is strongly more important than the latter, 7 means that the former is extremely more important than the latter, 9 means that the former is absolutely more important than the latter, and the remaining 2, 4, 6, and 8 are intermediate transition scales.

[0013] In the preferred embodiment, step S2 further includes the following steps: S21. Construct judgment matrices for each layer; The indicator layer (A) constructs a 4×4 judgment matrix. This represents the importance scale value of the i-th primary indicator relative to the j-th primary indicator, satisfying... That is, reciprocity and That is, reflexivity; Then the weights are calculated, starting with the product of the elements in each row of the judgment matrix: (7) Recalculate The nth root: (8) Where n is the order of the judgment matrix; Next to After normalization, the basic weights are obtained: (9) Calculate the product of the judgment matrix and the weight vector: (10) Calculate the largest eigenvalue again (11) Consistency Indicators: (12) Next, find the average random consistency index RI; Finally, calculate the consistency ratio: (13) when If the value is less than the preset value, the matrix is ​​judged to be consistent and the weights are valid, thus passing the consistency test. The same operation applies to other layers.

[0014] S22. Introduce a dynamic correction module; Constructing hydrological time series characteristic coefficients Based on dynamic risk factors such as riverbed scouring and deposition rate and ship wave scouring intensity, and combined with a quantitative analysis of the impact of historical hazard data on dike risks, a weighted ratio of strong riverbed scouring and deposition and weak ship wave scouring is used for calculation. The formula is as follows: (14) in The riverbed erosion and deposition rate; This represents the average daily intensity of wave scouring. This represents the riverbed scouring and deposition weighting coefficient. The weighting coefficient for the ship's wave scavenging; Then on Normalization is performed: (15) in The largest in the region's history value; Avoid over-adjusting weights; Furthermore, dynamic weight calculation is introduced, namely: (16) in The original base weights; This is a correction factor; These are the normalized hydrological time-series characteristic coefficients; when When it increases, Adjust accordingly upwards; conversely, adjust downwards. Finally, a safety index correction term is introduced to avoid excessive dynamic weight correction that could lead to deviation from actual risks. The safety index is used to verify and adjust the dynamic weights. The formula for calculating the safety index is: (17) in Let i be the risk-causing characteristic value of the i-th secondary indicator. Its dynamic weight; Through safety index Dynamic adjustment of correction coefficient Reduce the correction factor when overcorrection occurs. Increase the correction factor when the correction is insufficient. The dynamic weights are recalculated after adjustment. (18) in This is the adjusted correction factor; The final dynamic weight matrix is ​​output, which solves the problem of incompatibility between traditional AHP static weights and dynamic environments, and provides support for the quantitative accuracy of risk assessment.

[0015] In the preferred embodiment, in step S3: Scene adaptation factor calibration through weight correction magnitude Triggered, where the weight adjustment magnitude is: (19) when When the value exceeds the preset value, threshold calibration for this indicator is triggered; Then, a linear adjustment model was used to calibrate the mathematical model, as shown in the formula: (20) in This is the calibrated threshold for the k-th risk level; This is the baseline threshold for the k-th risk level; This is the threshold adjustment coefficient; The weight adjustment range for the i-th indicator; Special operating condition factor calibration uses the threshold offset calibration method, and the formula is: ;(twenty one) in This represents the threshold for the k-th risk level under special operating conditions. This is the threshold offset coefficient; This refers to the strength coefficient for the corresponding working condition; Finally, verify whether the threshold calibration meets the standard. First, collect multiple sets of historical incident data, and define the threshold accuracy rate as: ;(twenty two) The threshold used to determine the risk level is consistent with the actual situation. The representation is inconsistent; like If the value is greater than the preset value, the threshold calibration is effective; if If it is less than the preset value, adjust k or Recalibrate until Meets the standards; This addresses the contradiction arising from the mismatch between dynamic weights and static thresholds, improves the accuracy of risk classification, and adapts to scenarios where grassroots data is scarce, thus ensuring the consistency and accuracy of risk assessment.

[0016] In the preferred embodiment, in step S4: Indicator grouping is based on the degree of influence of indicators on risk assessment and the difficulty of data collection. The K-means clustering algorithm is used to group indicators into core essential indicators, important auxiliary indicators, and optional supplementary indicators. First, select the risk impact weight. ;(twenty three) and the difficulty level of data collection ;(twenty four) in For data collection completeness rate; As clustering features, the data is then standardized using a formula: (25) (26) Next, initialize 3 cluster centers. Calculate the Euclidean distance from each sample to the cluster center: (27) Iteratively update the cluster centers until a convergence threshold is reached; finally, calculate the silhouette coefficients. (28) in The average distance between a sample and other samples in the same cluster. The average distance between a sample and its nearest heterogeneous sample is given by the average silhouette coefficient. If the value is greater than the preset value, output the grouping results; The parallel computing pipeline architecture breaks down the risk assessment process into three sub-tasks: threshold after data matching and calibration, score calculation, and weighting. It builds an independent computing pipeline for each set of indicators and shortens the overall computing time through parallel computing of multi-core processors. First, the preprocessed index data for each group is distributed to the corresponding computation pipeline; Then, the grouped index scores are calculated in parallel using the formula: (29) in The preset core score; The weights are dynamic; n represents the corresponding pipeline. Furthermore, by directly assigning scores to grouped indicators based on threshold intervals, the output rate is accelerated; Finally, the comprehensive numerical value is calculated using the comprehensive numerical formula: (30) in The corresponding dynamic weights for the primary indicators; The corresponding dynamic weights for the secondary indicators; Based on comprehensive values Determine the risk level based on the corresponding interval; By grouping metrics to reduce unnecessary calculations, parallel pipeline architecture improves computing speed, reduces resource consumption, and shortens the evaluation cycle.

[0017] In the preferred embodiment, in step S5: First, the contribution of all secondary indicators is calculated using the risk contribution formula: (31) in The risk contribution of the j-th indicator, Score it; Then Sort the data in descending order and select only the top five indicators. Among these five indicators, those with scores below the medium-risk threshold are identified as weak points and recommended for improvement. Further develop a standardized handling template library, which includes templates for different risk levels, such as material quotas per unit length, dike length, risk level, number of construction personnel, and completion time. This will allow non-professionals to operate directly according to the templates, thus solving the problem of the disconnect between assessment and actual operation.

[0018] This invention provides a method for the dynamic weighting and threshold-based collaborative assessment and handling of flood control risks in dikes. It utilizes a dynamic weighting dual-module computation mechanism combined with a threshold self-calibration system to adapt to dynamic hydrological environments. The AHP (Advanced Hierarchical Method) base weights, combined with dynamic correction, retain the advantages of engineering experience while adapting to environmental changes in real time. Collaborative calibration of weights and thresholds avoids secondary misjudgments, improving the alignment of assessment results with actual engineering scenarios. A lightweight parallel computing architecture enhances computational efficiency, and a core indicator priority computation mechanism can quickly output preliminary risk levels, meeting the real-time needs of emergency decision-making during flood season and adapting to portable computing devices at the grassroots level. A data collection system that requires core data and allows optional auxiliary data adapts to scenarios with scarce grassroots data. A three-dimensional closed-loop output system transforms complex quantitative results into standardized templates, facilitating direct operation by non-professional personnel at the grassroots level. It can dynamically respond to various complex working conditions, ensuring accurate assessment under extreme environments through special working condition factor calibration and dynamic weight adjustment. It supports the assessment needs of dikes of different types, scales, and operating years, making it widely applicable. It does not employ a mixed subjective and objective weighting approach, does not introduce non-core indicators outside the dike itself, and has no complex function calculation steps, making it highly operable and easy to promote in engineering applications. Attached Figure Description

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of a method for collaborative assessment and handling of flood control risks of dikes using dynamic weights and thresholds, as per the present invention. Detailed Implementation Example 1 like Figure 1 As shown, a method for collaborative assessment and handling of flood control risks of dikes using dynamic weights and thresholds, with a dike section of the Yangtze River in a plain city as the application scenario, specifically includes the following steps: In the preferred scheme, S1 involves constructing a three-layer data system and completing data collection; The three-tiered data system includes core essential data, important auxiliary data, and real-time hydrological data, among which: The core essential data include the anti-sliding stability of the flood control wall slope, the structural stability of the flood control wall, the seepage safety performance, and the top elevation of the dike; Key auxiliary data include wall tilt, subsidence, wall aging, hydrodynamic conditions, riverbed conditions, and flood control channel connectivity. Real-time hydrological data includes riverbed erosion and deposition rate data, and ship wave scouring intensity data; The slope anti-sliding stability was assessed using a Leica TS60 total station. One monitoring section was set up every 500m along the length of the embankment. Each section had one monitoring point at the top, middle, and toe of the slope, for a total of 20 sections and 60 monitoring points. Three-dimensional coordinate data were collected every 7 days to calculate the displacement change between adjacent periods. The seepage safety performance was tested using a YS-2 constant head permeameter. One permeation test hole was set up every 800m on the water-facing side of the dike, for a total of 13 holes. The hole depth had to penetrate the seepage barrier layer of the dike. The seepage flow rate and head difference were continuously collected, and the permeability coefficient was calculated according to Darcy's law. Structural stability was assessed using a ZBL-U520 ultrasonic testing instrument, with one testing point every 300m along the length of the embankment, for a total of 34 points. The instrument scanned the integrity of the foundation and pile foundation and recorded the location of defects. Data acquisition of levee crest elevation: Using a HiTarget V98 GNSS receiver in conjunction with a DSZ2 level, with the nearby national second-order leveling point as the benchmark, one elevation monitoring point was set up every 200m, for a total of 50 points, to measure the current crest elevation and calculate the underestimation value compared with the design elevation; The wall inclination was assessed using a Bosch GLM 500 laser rangefinder, with measuring points set up at the same height on both sides of the wall to measure the difference in horizontal distance. The wall aging was assessed by taking images of the wall surface with a DJI Mavic 3 drone, and the YOLOv8 algorithm was used to automatically identify cracks and peeling, and the engineering quality level was quantified according to the proportion of the affected area. Hydrodynamic data were obtained using an NDJ-8S Doppler current meter. The flow velocity was measured in five layers at 0.5m intervals at the center of the river cross-section. Each layer was measured three times and the average value was taken to calculate the average flow velocity of the cross-section. Riverbed data were collected using the Kongsberg EM 2040 multibeam echo sounder system. Twenty survey lines were laid out along the river channel. The survey vessel collected riverbed elevation data at a speed of 5 km / h, generated a three-dimensional terrain model, and calculated the thickness of erosion and deposition. The slope coefficient was calculated by measuring the horizontal distance and vertical height of the upstream and downstream slopes of the dike using a Leica TS60 total station. Data on the connectivity of flood control channels was obtained using a Garmin eTrex track recorder combined with Gaode Maps. The shortest ideal length of the flood control channel was planned, the actual route length was recorded during on-site driving, and the detour length was calculated. The riverbed scouring and deposition rate data are reused using a multibeam echo sounder system. The riverbed topography is measured once every 15 days, and the scouring and deposition thickness is calculated by comparing it with the previous data. The scouring and deposition rate is then divided by the time. The wave scouring intensity data was obtained by using an SBE 26plus wave height meter deployed near the toe of the dike to record the wave height H and period T as the wave passed. The wavelength was then calculated using formula (4). Calculate the wave speed according to formula (5) Then calculate the scouring intensity according to formula (6). ; Subsequently, the displacement monitoring data was processed using the 3σ criterion. The mean of the displacement data of the monitoring points was calculated according to formula (1); the standard deviation σ was calculated according to formula (2); and outliers were judged according to formula (3). If an outlier was found, the data was removed.

[0020] In the preferred scheme, S2 involves constructing a dual-module weighting system and calculating dynamic weights; First, a basic weight matrix is ​​constructed using the AHP method. The target layer Z corresponds to the calculation of flood control risk of the dike. In the index layer A, A1 corresponds to the risk of dike safety and stability factors, A2 corresponds to the risk of overflow factors, A3 corresponds to the risk of river morphology factors, and A4 corresponds to the risk of emergency rescue and disaster relief factors. In the secondary index layer B, B1 corresponds to engineering quality, B2 corresponds to structural safety, B3 corresponds to the degree of insufficient dike crest elevation, B4 corresponds to the surrounding land use situation, B5 corresponds to hydrodynamic conditions, B6 corresponds to riverbed conditions, and B7 corresponds to the flood control channel connectivity. Subsequently, five experts were invited to score the data using a 1-9 scale to construct the judgment matrix for indicator layer A:

[0021] Calculate the product of each row according to formula (7). Calculate the nth root according to formula (8) Then, according to formula (9) Normalization is performed to obtain the basic weights. ; After calculating A×W according to formula (10), calculate the maximum eigenvalue according to formula (11). Then bring in Calculate the consistency index CI according to formula (12); Finally, the average random consistency index RI is searched, and the consistency ratio CR is calculated according to formula (13). The consistency of the matrix is ​​judged. If it is reasonable, the basic weights are effective. Next, a dynamic correction module is introduced, and the calculation method is as follows: Take the riverbed scouring and deposition weight coefficient Ship wave wash weight coefficient The regional historical hydrological time series characteristic coefficient is the largest. value Calculate according to formula (14) Normalized hydrological time series characteristic coefficients are obtained by normalization according to formula (15). ; Take the initial correction coefficient λ, and calculate the dynamic weight according to formula (16). ; Based on the risk-causing characteristic values ​​of secondary indicators The safety index is calculated according to formula (17). Through safety index Dynamic adjustment of correction coefficient Reduce the correction factor when overcorrection occurs. Increase the correction factor when the correction is insufficient. After adjustment, recalculate the dynamic weights and recalculate according to formula (18). The final output is a dynamic weight matrix.

[0022] In the preferred scheme, S3 involves constructing a two-factor threshold calibration system and completing threshold calibration; First, the scene adaptation factor is calibrated: Taking the secondary indicator B2 as an example, the calculation of the weight adjustment magnitude is as follows. , Calculate the weight correction magnitude according to formula (19) , The preset trigger threshold is set to 15%, so the trigger threshold calibration and the weight correction of other indicators are calculated in the same way. Taking the secondary indicator B2 as an example, the linear adjustment calibration uses the high-risk benchmark threshold of indicator B2. A score below 80 indicates high risk, and the threshold adjustment coefficient is... Calculate according to formula (20) If the corrected score is below 91, it is considered high risk; the same applies to the linear adjustment calibration of other indicators. Next, the special operating condition factor is calibrated: The special working condition is set for this section of the dike to encounter torrential rains during the flood season. First, the working condition intensity coefficient is set. Threshold offset coefficient Calculate according to formula (21) ; The validity of the threshold was verified by collecting 100 sets of historical hazard data for the levee section over the past 10 years and calculating the threshold accuracy using formula (22). The preset threshold is 90%. If the threshold is met, the threshold calibration is effective.

[0023] In the preferred solution, S4 involves constructing a lightweight parallel computing architecture and shortening the evaluation cycle. First, the indicators are grouped using K-means clustering: Calculate the risk impact right according to formula (23) ; Calculate the data collection difficulty coefficient according to formula (24) If the completeness rate of B2 collection is 95%, then ; according to Get as well as Calculate according to formula (25) ;according to Get

[0024] as well as Calculate according to formula (26) ; Initialize 3 cluster centers Calculate the Euclidean distance from the sample to the center using formula (27), iteratively update the cluster center until it converges to 0.001, and calculate the average silhouette coefficient using formula (28). The output grouping results are as follows: core essential indicators (B2, B3, B5), important auxiliary indicators (B1, B6, B7), and optional supplementary indicators (B4). The architecture was built using an Intel Core i7-12700H processor, with three independent computing pipelines corresponding to three types of metrics. Calculate the core indicator score according to formula (29), such as , ,but ; The comprehensive risk level is determined by calculating the comprehensive value using formula (30). Combined with the calibrated threshold set: Low risk: 90≤R≤100; Low to medium risk: 80 ≤ R < 90; Medium risk: 60 ≤ R < 80; High risk: 0 ≤ R < 60; Determine the current risk level of this section of the dike.

[0025] In the preferred solution, S5 involves constructing a three-dimensional closed-loop output system and generating practical instructions; In the process of identifying the weakest link indicators: The risk contribution rate is calculated using formula (31) to determine the contribution rate of the secondary indicators. ; Then The scores were sorted in descending order as follows: C2=3.59, C5=2.87, C1=2.15, C6=1.39, C3=1.12. The top three indicators were selected, and the corresponding secondary indicators with scores below the medium-risk threshold of 80 were identified as weak indicators, indicating that structural safety monitoring, hydrodynamic control, and riverbed dredging need to be strengthened. Based on the level, the corresponding template in the disposal template library is invoked: For example, medium risk corresponds to: Material quota per unit length: 12 sandbags / meter, 5m² / meter geotextile, 2 shovels / 100 meters; Length of dike treatment: 3km of dike section involved in the short-board indicator; Number of construction workers: 30 people per kilometer, totaling 90 people; Completion time limit: 12 hours; Key operational guidelines: For section X of the short plank, a temporary seepage barrier layer is constructed using sandbags, with a stacking height of 0.5m; for section Y of the short plank, flow velocity meters are deployed for real-time monitoring to limit the speed of passing vessels; for section Z of the short plank, dredging is carried out using a dredger to a depth of 0.3m.

[0026] Through the above steps, the flood control risks of this section of the Yangtze River embankment can be accurately assessed and effectively addressed, taking into account the accuracy of the assessment, the real-time nature of the emergency response, and the practicality at the grassroots level. The above embodiments are merely preferred technical solutions of the present invention and should not be considered as limitations on the present invention. The scope of protection of the present invention should be limited to the technical solutions described in the claims, including equivalent substitutions of the technical features described in the claims. That is, equivalent substitutions and improvements within this scope are also within the scope of protection of the present invention.

Claims

1. A method for dynamically weighting and thresholding co-assessment of flood risk of embankments, characterized in that, Comprise the following steps: S1, construct a three-layer data system covering the risk assessment of embankment, which contains core essential data, important auxiliary data and real-time hydrological data, to provide real-time data support for subsequent dynamic weight calculation and threshold calibration; S2, construct a dual-module weighting system of basic weight and dynamic correction, construct a basic weight matrix through AHP, calculate dynamic correction coefficients combined with real-time hydrological data, and dynamically adjust the index weight; S3, construct a dual-factor threshold calibration system of scene adaptation factor and special working condition factor, trigger scene adaptation threshold calibration according to weight correction amplitude, and trigger special working condition threshold calibration combined with extreme working condition intensity coefficient; S4, construct an architecture of index grouping, pipeline parallel and lightweight calculation, adopt clustering algorithm for index grouping, and build multiple parallel computing pipelines; S5, construct a three-dimensional closed-loop output system of risk level, short-board index and disposal template, calculate the risk contribution of index to identify short-board index, call preset standardized disposal template library, and generate directly operable operation guide.

2. The method of claim 1, wherein, In step S1, The core essential data includes structural safety data and embankment crest elevation data; The structural safety data includes anti-flood wall slope anti-sliding stability, anti-flood wall structure stability, and seepage safety performance; The embankment crest elevation data includes embankment design crest elevation and present anti-flood wall crest elevation; The important auxiliary data includes engineering quality data, hydrodynamic condition data, riverbed condition data, and anti-flood passage penetration condition data; The real-time hydrological data includes riverbed erosion and deposition rate data and ship wave scouring intensity data.

3. The method of claim 2, wherein, In step S1, The structural safety data is collected by using total station to monitor slope anti-sliding stability; seepage test is carried out by using permeameter, and structure stability is detected by using ultrasonic detector; After collecting slope anti-sliding stability data, 3σ rule is used to process abnormal values, and mean value ； wherein is displacement monitoring data, is the total number of displacement monitoring data; and standard deviation are calculated: ； Abnormal values of displacement monitoring data satisfying ； are removed.

4. The method of claim 1, wherein, In step S2, The basic weight matrix is constructed by AHP, which has target layer, index layer and secondary index layer, the target layer is embankment flood control risk calculation, the index layer includes embankment safety and stability factor risk, overflow factor risk, river regime factor risk, and rescue factor risk, and the secondary index layer includes engineering quality, structure safety, embankment crest elevation deficiency degree, surrounding land use condition, hydrodynamic condition, riverbed condition, and anti-flood passage penetration condition; after construction, consistency test is used to determine whether the basic weight is effective.

5. The method of claim 4, wherein, In step S2, The dynamic correction module is constructed by building hydrological time series characteristic coefficient: ； wherein is the riverbed erosion and deposition rate; is the daily average ship wave scouring intensity; is the riverbed erosion and deposition weight coefficient; is the ship wave scouring weight coefficient; After normalization processing: ； wherein is the area history maximum value; Dynamic weight is calculated according to formula: ； wherein is the original basis weight; is the correction coefficient; is the normalized hydrological time series characteristic coefficient; When increases, correspondingly up-regulated; otherwise down-regulated. And safety index is introduced: ； wherein is the hazard characteristic value of the i-th secondary indicator, is its dynamic weight; According to Recalculate the dynamic weights after adjusting the correction factor λ: ； wherein is the adjusted correction factor.

6. The method of claim 1, wherein, In step S3, Scene adaptation factor calibration is carried out by weight correction amplitude: ； wherein is a dynamic weight; is a base weight; Triggered when Greater than a preset value, linearly adjusted by the formula: ； wherein is a calibrated threshold value for the kth risk level; is a reference threshold value for the kth risk level; is a threshold adjustment coefficient; is a weight correction amplitude for the ith indicator; Special working condition factor calibration is carried out according to formula: ； wherein is a threshold value for the kth risk level under the special working condition; is a threshold offset coefficient; is a corresponding working condition intensity coefficient; Threshold offset is carried out, and after calibration, threshold accuracy is verified: ; the representative threshold value decision risk level is consistent with the actual, the representative is inconsistent; When Acc is greater than or equal to preset value, calibration is effective.

7. The method of claim 1, wherein, In step S4, Risk influence weight ； wherein is the corresponding first level indicator weight; and data acquisition difficulty coefficient ； wherein is the data collection completeness rate; are selected as clustering characteristics, after standardization, K-means clustering algorithm is used to divide them into core essential index, important auxiliary index and optional supplementary index, and after clustering, profile coefficient is used to verify grouping rationality: ； wherein is the average Euclidean distance of the sample to other samples of the same cluster, is the average Euclidean distance of the sample to the nearest sample of a different cluster. When the average profile coefficient If the value is greater than the preset value, output the grouping result.

8. The method of claim 7, wherein, In step S4, The parallel computing pipeline splits the risk assessment into three sub-tasks of data matching calibration threshold, score calculation, and weight weighting, and calculates the group score according to the formula: ； wherein is a preset core score; is a dynamic weight; n is a corresponding pipeline; The comprehensive numerical value is calculated by the comprehensive numerical formula: ； wherein is the dynamic weight corresponding to the first-level index; is the dynamic weight corresponding to the second-level index; According to the comprehensive numerical value The risk level is determined according to the corresponding interval.

9. The method of claim 1, wherein, In step S5, The short board index is calculated by the risk contribution formula: ； wherein is the risk contribution of the jth indicator, is its score; according to Sort in descending order and filter the indicators whose scores are below a specified threshold from the top few indicators, and use them as the weakest indicators; if the filter results are empty, retain the indicators with the lowest scores.

10. The method of claim 9, wherein, In step S5, The standardized treatment template library presets the unit length material quota, the number of construction personnel, and the completion time limit according to the risk level, contains operation steps, material list, and responsibility division, supports self-defined adjustment according to the area type, and generates practical operation guide.

Images