River danger hidden trouble intelligent identification method based on multi-source data fusion
By using a multi-source data fusion-based intelligent identification method for river hazards, and combining water level, seepage pressure, displacement, and image data with spatial grids and judgment indices, the method achieves automatic identification and graded early warning of river embankment hazards. This solves the problems of inaccurate monitoring and lack of automation in existing technologies, and improves the safety and management efficiency of river embankments.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-24
- Publication Date
- 2026-07-14
AI Technical Summary
Existing river embankment monitoring technologies are insufficient for all-weather real-time monitoring, lack accuracy in identifying hazards, and lack multi-source data fusion and quantitative evaluation, resulting in low accuracy in locating hidden dangers and failing to meet the real-time and intelligent requirements for the safe operation of river embankments.
By employing a multi-source data fusion method, a joint judgment model of water level, seepage pressure, displacement, and image is established. Combined with spatial grid division and weighting coefficients, a leakage and bank collapse judgment index is constructed. Convolutional neural networks are used to identify image anomalies, thereby achieving risk assessment and graded early warning.
It has improved the accuracy and reliability of identifying potential hazards in river embankments, enabling automatic identification and precise location of hazards, reducing the workload of manual patrols, and enhancing the automation level of river embankment safety monitoring.
Smart Images
Figure CN122390652A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of river engineering safety monitoring technology, specifically to an intelligent identification method for river hazards and risks based on multi-source data fusion, applicable to safety monitoring and risk warning of river embankments, revetment projects, and flood channels. Background Technology
[0002] River embankments are susceptible to various hazards during operation, including flood erosion, long-term seepage, rainfall erosion, and foundation deformation. These hazards can lead to seepage, piping, bank collapse, and landslides. If these hazards are not detected and addressed in a timely manner, they may cause embankment instability or even dam failure, resulting in serious safety accidents.
[0003] Current methods for monitoring river hazards include manual patrols, single-sensor monitoring, video surveillance, and remote sensing. Manual patrols rely heavily on on-site observation by patrol personnel, and their effectiveness is limited by experience and patrol frequency, making 24 / 7 real-time monitoring difficult. Single-sensor monitoring typically focuses on individual indicators such as water level, seepage pressure, or displacement, failing to comprehensively reflect the overall operational status of the dike. While video surveillance can continuously observe the surface of the dike, it struggles to effectively identify hidden hazards such as internal seepage and piping. Remote sensing can acquire large-scale river imagery, but it is usually used independently, lacking integration with ground monitoring data, leading to insufficient accuracy in hazard assessment.
[0004] Furthermore, existing technologies process monitoring data from different sources separately, lacking a unified data fusion mechanism. This makes it difficult to comprehensively assess the operational status of river embankments, resulting in the inability to detect potential hazards such as seepage and bank collapses in a timely manner. Simultaneously, existing methods lack quantitative evaluation models for the degree of hazard risk, hindering automatic identification and tiered early warning. The accuracy of hazard location is also low, failing to meet the requirements for real-time, accurate, and intelligent monitoring for the safe operation of river embankments.
[0005] Therefore, it is necessary to propose a method for automatically identifying potential hazards in river embankments based on multi-source data fusion analysis, so as to improve the accuracy and reliability of river embankment hazard identification.
[0006] The information disclosed in this background section is only intended to enhance the understanding of the background technology of this application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention
[0007] The purpose of this invention is to provide an intelligent identification method for river embankment seepage and bank collapse risks based on multi-source data fusion. By establishing a joint judgment model of water level-seepage pressure-displacement-image, the method can realize automatic identification, risk assessment and early warning of embankment risks.
[0008] To achieve the above objectives, the present invention adopts the following technical solution: A method for intelligent identification of river hazards and potential risks based on multi-source data fusion includes the following steps: S1. Monitoring equipment deployment and coordinate establishment: Water level sensors, piezometers, displacement monitors and video or remote sensing monitoring equipment are deployed along the river embankment, and spatial coordinate information corresponding to each monitoring device is established. S2. Multi-source monitoring data acquisition and time synchronization: Acquire monitoring data, including water level data. H (t), Isostatic data P (t), levee displacement data D (t) and image data I (t), and perform time synchronization processing on various types of data according to preset time intervals; S3. Spatial grid division and data mapping: The monitoring area is divided into spatial grids according to the location of the dike chainage, and each monitoring data is mapped to the corresponding spatial grid unit; S4. Leakage Judgment Index Calculation: Based on the water level change rate Δ H and the rate of change of osmotic pressure Δ P Constructing a leakage judgment index L , L = a Δ H + b Δ P； in, Δ H = H (t) - H (t-1); Δ P = P (t) - P (t-1); a , b These are the weighting coefficients; when L Greater than the preset threshold L At time 0, the spatial grid is determined to have a risk of leakage; S5. Calculation of Bank Collapse Judgment Index: Based on the displacement change rate Δ D and shoreline change Δ S Constructing a bank collapse judgment index C , C = α Δ D + β Δ S； in, ΔD = D (t)- D (t-1); Δ S = S (t)- S (t-1); α , β These are the weighting coefficients; when C When the value exceeds the preset threshold C0, it is determined that there is a risk of bank collapse; S6. Image Anomaly Recognition Probability Calculation: Target recognition is performed on the image data to obtain the recognition probability of cracks, collapses, or turbid water bodies. Pr , Pr = f ( I (t)); in, I (t) represents the image data at the current time. f (·) represents the convolutional neural network recognition model; S7. Multi-source data fusion risk calculation: Using spatial grid cells as the basic analysis object, the risk calculation of various monitoring data is performed within the same grid, and the result is calculated based on the leakage assessment index. L A comprehensive risk value is constructed using the collapse judgment index C and the identification probability Pr. R , R = w 1 L + w 2 C + w 3 Pr； in w 1. w 2. w 3 represents the weighting coefficient; S8. Risk Classification Judgment and Early Warning Output: Based on the comprehensive risk value R and the preset risk threshold, the risk classification judgment is performed, and the risk type, spatial location and risk level are output to realize intelligent identification of potential risks to river embankments.
[0009] Preferably, the time synchronization processing in step S2 includes unifying all monitoring data to a fixed time step Δt, and using linear interpolation to compensate for asynchronously acquired data.
[0010] Preferably, in step S3, the spatial grid is divided into units every 5 to 20 m along the direction of the dike.
[0011] Preferably, in step S6, the image recognition uses a convolutional neural network model to classify and identify the images to obtain the probability of cracks, collapses, slope slippage, or abnormal turbidity. Pr .
[0012] Preferably, the comprehensive risk value in step S8 R Classification is based on the following rules: when R A level I warning is issued when R1 is ≥ 1. when R An alarm of level II is triggered when R2 is ≥ 2. when R A value ≥ R3 indicates a Level III emergency. Where R3>R2>R1.
[0013] Preferably, the hazard output in step S8 includes: hazard type, occurrence time, levee location, latitude and longitude coordinates, and risk level, and is displayed on the geographic information system interface.
[0014] Based on the same inventive concept, this invention also provides an intelligent identification system for river channel hazards and risks based on multi-source data fusion, comprising: The data acquisition module is used to collect water level data, seepage pressure data, displacement data, and image data; The time synchronization module is used to align the time of data from multiple sources. The spatial matching module is used to perform spatial grid mapping according to the dike station number; The leakage assessment module is used to calculate the leakage assessment index based on the water level change rate and the seepage pressure change rate. The bank collapse determination module is used to calculate the bank collapse determination index based on the displacement change rate and the amount of shoreline change. Image recognition module, used for target recognition in images; The fusion judgment module is used to calculate the comprehensive risk value; The early warning output module is used to output information on the location and risk level of the hazard.
[0015] The above fusion determination module is based on the formula R = w 1 L + w 2 C + w 3 Pr； Calculate the overall risk value and output the risk level based on the preset threshold.
[0016] Based on the same inventive concept, the present invention also provides an electronic device, comprising: The system includes a processor, a memory, and a communication interface; the memory stores a computer program; when the computer program is executed by the processor, it implements the above-mentioned intelligent identification method for river hazards and risks based on multi-source data fusion.
[0017] Based on the same inventive concept, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned intelligent identification method for river channel hazards based on multi-source data fusion.
[0018] Compared with the prior art, the present invention has the following beneficial effects: (1) By synchronizing and matching the time and space of multi-source monitoring data such as water level, seepage pressure, displacement and images, and performing fusion analysis, compared with single data monitoring methods, it can more comprehensively reflect the operation status of river embankments and improve the accuracy of identifying dangers such as seepage and bank collapse. (2) By constructing a leakage judgment index, a bank collapse judgment index and a comprehensive risk value model, various monitoring data are weighted and calculated to realize automatic identification of danger and risk assessment, which is conducive to timely detection of abnormalities in the early stage of danger and improves the reliability of dike safety monitoring. (3) Establish a spatial positioning grid based on the dike station number, and output the location and risk level of the danger by combining the multi-source fusion results, so as to realize the accurate positioning and graded early warning of the danger, improve the automation level of river dike safety management, and reduce the workload of manual inspection. Attached Figure Description
[0019] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description of the specific embodiments will be briefly introduced below. Obviously, the following description is only a part of the embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1 This is a schematic diagram of the method flow of the present invention.
[0021] Figure 2 This is a schematic diagram of the multi-source data fusion judgment model of the present invention. Detailed Implementation
[0022] This invention proposes an intelligent identification method for river channel hazards based on multi-source data fusion. To facilitate understanding of this invention by those skilled in the art, the specific implementation of this invention is described below with reference to the accompanying drawings.
[0023] In this invention, unless otherwise specified, all equipment used is commercially available or commonly used in the art. The methods in the following embodiments, unless otherwise specified, are conventional methods in the art. Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.
[0024] I. Monitoring Equipment Deployment and Coordinate Establishment First, monitoring equipment is deployed along the river embankment. In this embodiment, potential hazards are identified along a 5 km long embankment. A monitoring section is set up every 50 m along the embankment. Each monitoring section is equipped with one water level sensor, two piezometers, and one displacement monitor. Video monitoring equipment is also set up on the outside of the embankment.
[0025] A stationing and coding system is established along the main line of the dike, with the starting point of the dike as the reference point, and the starting point is set as... K 0, the endpoint is K 5000 indicates that the total length of the dike is 5000 m. A monitoring section is set up every 50 m along the dike, and numbered sequentially as follows: K 0、 K 50. K 100 K 150. K 200…… K 5000, where the station number represents the distance from the starting point in meters. The station number and latitude / longitude coordinates of each monitoring device are also recorded for subsequent spatial positioning.
[0026] II. Multi-source monitoring data acquisition and time synchronization The monitoring system collects water level data at fixed time intervals. H (t), Isostatic data P (t), displacement data D (t) and image data I (t), where the sampling period for water level and seepage pressure is 1 min, and the sampling period for displacement is 5 min. The image data is collected in real time by video equipment and once a day by drone. All data are unified to a time step of Δt=1 min, and asynchronously collected data are supplemented by linear interpolation to ensure that all types of data are calculated on the same time scale.
[0027] III. Spatial Grid Generation and Data Mapping In this embodiment, the monitoring equipment sets up a monitoring section every 50 m along the embankment. However, the spatial grid division accuracy of this part is higher than the spacing between the monitoring sections. The embankment is subdivided into multiple grid units with a grid length of 10 m. Gi. For grid cells without monitoring equipment, the data is obtained from data from adjacent monitoring sections through nearest neighbor matching or linear interpolation, ensuring that each grid cell has corresponding water level data. H ( K i,t), osmotic pressure data P ( K i,t) displacement data D ( K i,t) and image data I ( K i,t), thereby achieving a unified spatial mapping of multi-source data and forming a unified data set. G i(t).
[0028] IV. Calculation of Leakage Judgment Index After completing time synchronization and spatial matching, a leakage judgment index is constructed for each grid cell. L .
[0029] First, calculate the rate of change of water level Δ at adjacent time points. H and the rate of change of osmotic pressure Δ P ,in, Δ H = H (t)- H (t-1)(1) Δ P = P (t)- P (t-1)(2) Set weight coefficients a and b In this embodiment a =0.6, b =0.4, calculate the leakage judgment index according to the following formula. L = a Δ H + b Δ P (3) A leakage detection threshold L0 = 0.15 is set; when L > L0, the grid cell is considered to have a leakage risk. For example, when Δ H =0.20 m, Δ P When the pressure is 0.10 MPa, L = 0.16, which is greater than the threshold of 0.15, and is therefore judged as an abnormal leakage.
[0030] V. Calculation of Bank Collapse Judgment Index First, displacement data is collected using a displacement monitoring instrument. D (t), calculate the displacement change Δ D , Δ D = D(t)- D (t-1)(4) Simultaneously, the shoreline location was extracted based on drone imagery. S (t), and calculate the change in shoreline Δ S , Δ S = S (t)- S (t-1)(5) Set weight coefficients α and β In this embodiment α =0.7, β =0.3, calculate the bank collapse judgment index according to the following formula, C = α Δ D + β Δ S (6) Set threshold C 0 = 0.10, when C > C At time 0, a risk of bank collapse was determined. For example, Δ D =0.12 m, Δ S When =0.05 m, C =0.099, close to the threshold, indicating a bank collapse warning state.
[0031] VI. Calculation of Anomaly Probability in Image Recognition After obtaining the leakage index L and bank collapse index C Subsequently, auxiliary identification is performed on the image data. Convolutional neural networks are used to identify targets in video surveillance images or drone footage, including cracks, collapses, landslides, and abnormally muddy water areas, and the probability of anomalies is output. Pr , Pr = f (I(t))(7) In the formula, I (t) represents the image data at the current time. f (·) represents the convolutional neural network recognition model. Pr This represents the probability value of the presence of cracks, collapses, or abnormal water bodies in the image.
[0032] when Pr Greater than the preset threshold P At time 0, the image is considered to have abnormal features. In this embodiment, P 0 is taken as 0.6, for example Pr When the value is 0.72, the grid is determined to have image anomalies.
[0033] VII. Risk Calculation of Multi-Source Data Fusion Based on the leakage assessment index L Collapse Judgment Index C and image recognition probability Pr Constructing a comprehensive risk value R The calculation formula is as follows: R = w 1 L + w 2 C + w 3 Pr (8) in w 1. w 2. w 3 represents the weighting coefficient, which is set in this embodiment. w 1 = 0.4 w 2 = 0.4, w 3 = 0.2.
[0034] Based on the risk value R, a risk level threshold is set. R ≥ 0.20 indicates a Level I warning. An R value ≥ 0.30 indicates a Level II alarm. A risk level of R ≥ 0.40 indicates a Level III emergency.
[0035] VIII. Risk Classification Assessment and Early Warning Output The system outputs the levee station location, latitude and longitude coordinates, hazard type, and risk level for the corresponding grid unit, and displays the results on the geographic information system interface. Simultaneously, it sends early warning information to the monitoring platform. An alarm is sent when the risk level reaches Level II or above, and an emergency alert is activated when it reaches Level III, thus achieving automatic identification and early warning of river levee seepage and bank collapse hazards.
[0036] The embodiments of the present invention described above do not constitute a limitation on the scope of protection of the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the claims of the present invention.
Claims
1. A method for intelligent identification of river channel hazards and potential risks based on multi-source data fusion, characterized in that, Includes the following steps: S1. Monitoring equipment deployment and coordinate establishment: Water level sensors, piezometers, displacement monitors and video or remote sensing monitoring equipment are deployed along the river embankment, and spatial coordinate information corresponding to each monitoring device is established. S2. Multi-source monitoring data acquisition and time synchronization: Acquire monitoring data, including water level data. H (t), Isostatic data P (t), levee displacement data D (t) and image data I (t), and perform time synchronization processing on various types of data according to preset time intervals; S3. Spatial grid division and data mapping: The monitoring area is divided into spatial grids according to the location of the dike chainage, and each monitoring data is mapped to the corresponding spatial grid unit; S4. Leakage Judgment Index Calculation: Based on the water level change rate Δ H and the rate of change of osmotic pressure Δ P Constructing a leakage judgment index L , L = a D H + b D P ; in, Δ H = H (t) - H (t-1); Δ P = P (t) - P (t-1); a , b These are the weighting coefficients; when L Greater than the preset threshold L At time 0, the spatial grid is determined to have a risk of leakage; S5. Calculation of Bank Collapse Judgment Index: Based on the displacement change rate Δ D and shoreline change Δ S Constructing a bank collapse judgment index C , C = α D D + β D S； in, Δ D = D (t)- D (t-1); Δ S = S (t)- S (t-1); α , β These are the weighting coefficients; when C When the value exceeds the preset threshold C0, it is determined that there is a risk of bank collapse; S6. Image Recognition Anomaly Probability Calculation: Target recognition is performed on image data to obtain the recognition probability of cracks, collapses, or turbid water bodies. Pr , Pr = f （ I (t)）； in, I (t) represents the image data at the current time. f (·) represents the convolutional neural network recognition model; S7. Multi-source data fusion risk calculation: Using spatial grid cells as the basic analysis object, the risk calculation of various monitoring data is performed within the same grid, and the result is calculated based on the leakage assessment index. L A comprehensive risk value is constructed using the collapse judgment index C and the identification probability Pr. R , R = w 1 L + w 2 C + w 3 Pr； in w 1. w 2. w 3 represents the weighting coefficient; S8. Risk Classification Judgment and Early Warning Output: Based on the comprehensive risk value R and the preset risk threshold, the risk classification judgment is performed, and the risk type, spatial location and risk level are output to realize intelligent identification of potential risks to river embankments.
2. The intelligent identification method for river channel hazards based on multi-source data fusion according to claim 1, characterized in that: The time synchronization process in step S2 includes unifying all monitoring data to a fixed time step Δt, and using linear interpolation to compensate for asynchronously acquired data.
3. The intelligent identification method for river channel hazards based on multi-source data fusion according to claim 1, characterized in that: In step S3, the spatial grid is divided into units every 5 to 20 m along the direction of the dike.
4. The intelligent identification method for river channel hazards based on multi-source data fusion according to claim 1, characterized in that: In step S6, the image recognition uses a convolutional neural network model to classify and identify the images to obtain the probability of cracks, collapses, slope slippage, or abnormal turbidity. Pr .
5. The intelligent identification method for river channel hazards based on multi-source data fusion according to claim 1, characterized in that: Comprehensive risk value in step S8 R Classification is based on the following rules: when R A level I warning is issued when R1 is ≥ 1. when R An alarm of level II is triggered when R2 is ≥ 2. when R A value ≥ R3 indicates a Level III emergency. Where R3>R2>R1.
6. The intelligent identification method for river channel hazards based on multi-source data fusion according to claim 1, characterized in that: The hazard output in step S8 includes: hazard type, occurrence time, levee location, latitude and longitude coordinates, and risk level, and is displayed on the geographic information system interface.
7. A smart identification system for river hazards and potential risks based on multi-source data fusion, used to execute the smart identification method for river hazards and potential risks based on multi-source data fusion as described in any one of claims 1 to 6, characterized in that, include: The data acquisition module is used to collect water level data, seepage pressure data, displacement data, and image data; The time synchronization module is used to align the time of data from multiple sources. The spatial matching module is used to perform spatial grid mapping according to the dike station number; The leakage assessment module is used to calculate the leakage assessment index based on the water level change rate and the seepage pressure change rate. The bank collapse determination module is used to calculate the bank collapse determination index based on the displacement change rate and the amount of shoreline change. Image recognition module, used for target recognition in images; The fusion judgment module is used to calculate the comprehensive risk value; The early warning output module is used to output information on the location and risk level of the hazard.
8. The intelligent identification system for river hazards and potential risks based on multi-source data fusion according to claim 7, characterized in that: The fusion determination module follows the formula R = w 1 L + w 2 C + w 3 Pr； Calculate the overall risk value and output the risk level based on the preset threshold.
9. An electronic device, characterized in that, include: The system includes a processor, a memory, and a communication interface; the memory stores a computer program; when the computer program is executed by the processor, it implements the intelligent identification method for river hazards and hidden dangers based on multi-source data fusion as described in any one of claims 1 to 6.
10. A computer-readable storage medium, characterized in that: The medium stores a computer program, which, when executed by a processor, implements the intelligent identification method for river hazards and risks based on multi-source data fusion as described in any one of claims 1 to 6.