Refined methods and devices for identifying and inspecting the degree of seepage in earth-rock dams

Abstract

This invention relates to a refined method and apparatus for identifying and inspecting the degree of seepage in earth-rock embankments. The method involves using a drone equipped with a hyperspectral imager to photograph the embankment surface, extracting seepage areas using two threshold settings, and defining the diffusion and water content changes of the seepage area based on variations in the area and spectral reflectance statistical parameters. The degree of seepage is then assessed using the numerical values ​​of water content change and diffusion. This invention improves the accuracy of identifying abnormal seepage risks in embankments and enhances the automation and intelligence of disaster prevention.

Description

Technical Field

[0001] This invention belongs to the field of seepage detection technology for earth and rock dams, specifically relating to a refined identification and inspection method and device for the degree of seepage in earth and rock dams. Background Technology

[0002] Earth-rock dams are a crucial component of water conservancy projects, safeguarding people's lives and property. Seepage, a major hidden danger in earth-rock dams, possesses a degree of concealment and uncertainty, and in severe cases, can trigger dam failure. Monitoring and diagnosing seepage hazards in earth-rock dams has become a critical issue urgently needing resolution in interpreting their health status and predicting their service life. Current dam inspection and leak detection techniques have several shortcomings: ① Insufficient grassroots emergency response experience and a lack of technical personnel make it extremely difficult to accurately identify highly concealed abnormal seepage hazards; ② Geophysical detection technologies based on electrical, electromagnetic, and seismic methods are developing in diverse ways, with each technology having multiple solutions and limitations; ③ Airborne infrared thermal imaging technology is used for dam seepage detection, using low-temperature areas as the basis for identifying seepage points. However, the dam surface is generally characterized by pits, water accumulation, and vegetation cover, all of which are at low temperatures, causing "seepage-like" interference and resulting in a very high misjudgment rate; ④ Isotope tracer methods require seepage sources, artificial tracer deployment is difficult and the observation period is long, and radioactive tracers have adverse environmental impacts.

[0003] Hyperspectral imaging technology is often used in precision agriculture, environmental monitoring, food safety and oil and gas exploration. However, due to the complexity of the dam surface environment and the special nature of the working environment, there is a lack of refined evaluation indicators to distinguish between "leakage-like" situations and to establish leakage degree levels. Therefore, the feasibility of this technology in seepage detection of earth and rock dams still needs to be verified by adapting it to inspection methods. Summary of the Invention

[0004] The purpose of this invention is to overcome the defects of the prior art and provide a refined method and device for identifying and inspecting the degree of seepage in earth-rock dams.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A refined method for identifying and inspecting the degree of seepage in earth-rock dams includes:

[0007] Using drones equipped with hyperspectral imagers to inspect the downstream face of earth and rock dams, spectral images of the downstream face of the dams are captured, and suspected water bodies are extracted by using preset characteristic band spectral reflectance thresholds.

[0008] The drone resumed flight to inspect suspected water bodies and take hyperspectral images. Areas with pH ≥ threshold were extracted from the hyperspectral images as leakage areas. Statistical parameters of the spectral reflectance of the leakage areas were obtained, and the area of ​​the leakage areas was calculated.

[0009] The drone will fly again after a preset time interval T to inspect the leakage area and take hyperspectral images, and recalculate the statistical parameters of the spectral reflectance of the leakage area and the area of ​​the leakage area.

[0010] The ratio of statistical parameters of spectral reflectance from two re-flights and the ratio of the leakage area are used to characterize the change in water content and diffusivity of the leakage area. The degree of leakage is assessed using numerical values ​​of water content change and diffusivity.

[0011] As a preferred embodiment, the pH threshold is determined as follows:

[0012] Using a hyperspectral imager, identified rainwater accumulation areas and seepage areas within the earth-rock embankment were photographed. The spectral reflectance values ​​of these areas in the characteristic wavebands of the water body were analyzed. Based on these values, a threshold was set to distinguish between rainwater accumulation and seepage. Preferably, the pH threshold was set to 7.

[0013] As a preferred implementation, after identifying suspected water bodies or leakage areas, the drone's location and flight parameters are recorded to control the drone's re-flight position.

[0014] As a preferred implementation method, during the first flight, a drone inspection route is set. If the dam height H < 15m, one aerial photography route is set, and the drone's altitude is adjusted so that the hyperspectral imager's field of view can cover the entire downstream surface of the dam. Then, an aerial photograph is taken along the dam axis to complete the inspection of the entire downstream surface of the dam. If the dam height 15m ≤ H < 30m, two parallel aerial photography routes are set, and the drone's altitude is adjusted so that the field of view of the two aerial photography routes can cover the entire downstream surface of the dam. Then, an aerial photograph is taken along the dam axis to complete the inspection of the entire downstream surface of the dam.

[0015] As a preferred implementation, an alarm is triggered after a suspected body of water is identified, and the drone's location and flight parameters are recorded simultaneously.

[0016] As a preferred implementation, the spectral reflectance of the hyperspectral image is processed by first-order differentiation to extract areas suspected of containing water and areas of leakage. First-order differentiation enhances the spectral differences in water content within the leakage area, facilitating the analysis of reflectance changes in characteristic bands during re-flight and subsequent re-flight attempts.

[0017] In a preferred embodiment, the statistical parameter of the spectral reflectance is the minimum value of the spectral reflectance of the characteristic band at the leakage point.

[0018] As a preferred embodiment, the method of assessing the degree of leakage using changes in water content and diffusivity is as follows:

[0019] When 1 < r < 1.2 and 1 < k < 1.3, it is determined that this is the stage of slight seepage;

[0020] When 1.2 ≤ r and 1.3 ≤ k < 1.5, it is determined as the stage of rapid leakage;

[0021] When r = 1 and 1.5 < k, it is determined as the stage of seepage failure.

[0022] As a preferred implementation manner, the method further includes, after determining the leakage area in the first go-around, using PVC pipes to enclose a regular quadrilateral that includes the leakage area, leaving a leakage space between the quadrilateral and the leakage area; using the area of the quadrilateral as a calculation reference to calculate the actual leakage area during the go-around and the second go-around.

[0023] Another object of the present invention is to provide a device for using the above method, including a drone, a hyperspectral imager carried on the drone, a ground remote control station, PVC pipes, and a gray cloth;

[0024] The PVC pipes enclose a regular quadrilateral that includes the leakage area;

[0025] The gray cloth is placed in the shooting area;

[0026] The ground remote control station receives the hyperspectral images taken by the hyperspectral imager and performs data processing, including:

[0027] A classification and identification and alarm module, a regional area estimation module, a spectral feature extraction module, and a leakage degree evaluation module; the classification and identification and alarm module extracts suspected water body parts and leakage areas respectively based on the spectral reflectance threshold and pH threshold of the preset characteristic bands, and gives an alarm when a suspected water body part is found; the regional area estimation module estimates the area of the extracted leakage area; the spectral feature extraction module obtains the statistical parameters of the spectral reflectance of the leakage area; the leakage degree evaluation module characterizes the change in water content and the diffusion degree of the leakage area based on the ratio of the statistical parameters of the spectral reflectance of the two go-arounds and the ratio of the leakage area, and evaluates the leakage degree using the values of the change in water content and the diffusion degree.

[0028] The recognition principle of the present invention is as follows: During the first flight, due to the different spectral information of different objects, different species classification and recognition are carried out on the hyperspectral image data. By setting the threshold R0 of the spectral reflectance in the characteristic band, irrelevant information interference such as vegetation and potholes is excluded, and an alarm is given and recorded for the suspected water areas. During the re-flight, a new threshold is set to distinguish water bodies with different pH values. Spectral information analysis is carried out on the suspected water alarm areas to screen whether it is rainwater accumulation or reservoir leakage water, and the spectral information of the leakage water is collected to draw the reflectance R1 curve, observe the peak and trough of the band, obtain the statistical parameter R1' of the spectral data reflectance, and calculate the leakage area S1. Since rainwater accumulation dissolves carbon dioxide in the air to form substances such as carbonic acid, rainwater accumulation is weakly acidic, while dam leakage water is usually alkaline, and a pH threshold can be set for distinction. During the second re-flight, the reflectance R2 curve is drawn, the statistical parameter R2' of the spectral data reflectance and the leakage area S2 are obtained, and the reflectance change r = R1' / R2' (indicating the change in soil water content, that is, the water content) and the area change rate k = S2 / S1 (indicating the change in the degree of leakage area diffusion, that is, the diffusion degree) are calculated, and the leakage situation is comprehensively evaluated using the water content and the diffusion degree.

[0029] Since the continuous increase in the water volume in the leakage area will cause the leakage area to increase, and the increase in water content will cause the spectral reflectance to decrease. When 1 < r < 1.2 and 1 < k < 1.3, the water at the seepage point wets the surrounding soil, the water content of the soil in the leakage area increases, and the leakage area expands, but since no visible water flow is formed, the values of R1' and R2' are large, and it is determined that this is the stage of slight seepage; when 1.2 ≤ r and 1.3 ≤ k < 1.5, visible water continuously overflows, and both the leakage area and the water content of the leakage area continuously increase, and it is determined to be the stage of rapid leakage; when r = 1 and 1.5 < k, the soil in the permeation area is approximately saturated, the water content does not change significantly, and the values of R1' and R2' are small, while the area of the dam surface wetting the soil will rapidly expand, and it is determined to be the stage of permeation failure.

[0030] The method and device of the present invention have the following beneficial effects:

[0031] (1) High inspection efficiency: An unmanned aerial vehicle is used to carry a hyperspectral imager, and an automatic alarm module is set to achieve all-day, non-contact, large-area, fast, efficient, flexible and mobile inspection of earth-rock dams. Compared with manual walking inspection, the efficiency can be increased by more than 3 times, and it is especially suitable for the inspection of earth-rock dam body leakage with a large quantity and wide area;

[0032] (2) High recognition accuracy: The collected spectral images are rich in information. Due to the differences in material composition and structure, the material selectively absorbs and emits photons of different wavelengths. The differences between different ground objects can be well reflected in the images. At the same time, the reflection spectrum of the object has a "fingerprint" effect, which distinguishes different material information by different spectra of different objects.

[0033] (3) Quantitative Leakage Degree: Based on the images and spectral information obtained by the hyperspectral imager, the water content and area of ​​the leakage area are quantitatively evaluated. The leakage level is reasonably classified by comprehensively considering the various state parameters, which will provide a strong theoretical basis and basis for the leakage treatment plan. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the structure of the seepage device of an earth-rock dam, which is inspected using a drone equipped with a hyperspectral imager platform.

[0035] Figure 2 These are typical visible light and hyperspectral images of dam surface seepage, where a is the visible light image, b is the hyperspectral image, and c is the reflectance curve of different ground features on the dam surface.

[0036] In the image: 1. Drone; 2. Hyperspectral imager; 3. Downstream face of the dam; 4. Suspected seepage point of the dam; 5. Ground remote control station; 6. Classification and alarm module; 7. Area estimation module; 8. Spectral feature extraction module; 9. Leakage degree assessment module; 10. PVC pipe; 11. Gray cloth. Implementation

[0037] The technical solution of the present invention will be further described below through specific embodiments.

[0038] like Figure 1 and Figure 2 As shown, a refined identification method for detecting the degree of seepage in earth-rock dams based on a UAV platform equipped with a hyperspectral imager includes the following steps:

[0039] (1) Instrument inspection

[0040] Ensure the drone frame is intact, the propellers and arms are fully extended, and the hyperspectral imager is securely installed; check the battery levels of the drone and remote controller to ensure they are sufficient for operation; check the signal connections of the flight control system and the onboard control system to ensure the hyperspectral imager can sample normally and the ground remote controller can receive spectral information smoothly; turn on the GPS signal to ensure the flight position information is dynamically updated.

[0041] (2) Inspection preparation

[0042] The ground conditions of the dam were observed to estimate the field of view of the drone, and the drone's flight altitude and inspection flight path were set. The dam is 6m high, and the drone's flight altitude was set to 13m, ensuring the field of view covers the downstream dam surface. The inspection proceeded from the left shoulder to the right shoulder, completing the entire dam inspection in a single line. Considering that the hyperspectral image data acquired by the hyperspectral imager might be affected by atmospheric and water vapor factors after the drone reaches a certain altitude, a 2m*2m gray cloth calibrated by the National Institute of Metrology was placed in the shooting area before drone takeoff. During hyperspectral image acquisition, only one hyperspectral scene needed to be covered by the gray cloth.

[0043] (3) Inspection flight

[0044] The drone 1 is controlled to take off from the left shoulder of the dam and ascend to an inspection height of 13m, allowing the hyperspectral imager 2 to cover the downstream dam surface 3 within its field of view. Aerial photography of the downstream dam surface 3 is then conducted from left to right. During the inspection, the image data is preprocessed, including radiometric calibration, atmospheric correction, and orthorectification. Based on this, and considering the actual conditions of the study area, a classification identification and alarm module 6 is determined for the study area, integrating various classification principles. The classification identification and alarm module 6 reads the reflectance of each frame of the spectral image in real time and calculates the alarm conditions. If an alarm is detected at a certain location, that location is identified as a suspected seepage point 4 on the dam, and the flight parameters of the drone 1 are automatically recorded. The inspection then continues, completing the inspection of the entire downstream dam surface 3, until no further alarms are detected.

[0045] (4) Inspection and re-flight

[0046] Based on the recorded flight parameters of UAV 1 at suspected seepage point 4 of the dam, the drone re-flyed to suspected seepage point 4. Using the pH threshold set in the seepage area extraction module of the classification, identification, and alarm module 6, the drone extracted the feature bands of the hyperspectral image of suspected seepage point 4 and found the actual seepage. The drone 1 was controlled to hover and a quadrilateral composed of PVC pipes 9 was placed to surround the seepage area, and the area S of the quadrilateral composed of PVC pipes was measured and calculated. The relative area of ​​the seepage area and the quadrilateral composed of PVC pipes obtained by the area area estimation module 7 was converted with the quadrilateral area S to obtain the actual seepage area S1. At the same time, the spectral feature extraction module 8 can obtain the minimum value R1' of the reflectance of the feature band at the seepage point.

[0047] (5) Resumption of flights

[0048] After a 15-minute interval, the drone 1 flies back to the leak location based on the recorded flight parameters. The actual leak area S2 is obtained by converting the image of the leak area and the quadrilateral formed by the PVC pipe obtained by the area estimation module 7 with the quadrilateral area S. At the same time, the spectral feature extraction module 8 can obtain the minimum value R2' of the reflectance of the characteristic band at the leak location.

[0049] The leakage assessment module 9 diagnoses the leakage situation at a leakage point by comparing the values ​​of R1' and R2' and combining the ranges of r and k.

[0050] (6) Unmanned aerial vehicle (UAV) recovery

[0051] After the inspection and detection operation is completed, the operator lands the drone in a safe area; backs up the collected visible light and hyperspectral information, fills in the flight record; turns off the power of the drone, hyperspectral imager, remote control and other equipment in sequence, and checks the drone for component damage.

[0052] In this embodiment, the drone 1 is a DJI M300 RTK drone, and the hyperspectral imager 2 is a GaiaSky-mini3-VN airborne hyperspectral imager.

[0053] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made in accordance with the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for fine identification and patrol of the degree of leakage of earth and rockfill dams, characterized in that, The method comprises the following steps: The downstream surface of the earth-rock dam is patrolled by a UAV carrying a hyperspectral imager, and a spectral image of the downstream surface of the dam is taken. By using a preset characteristic band spectral reflectance threshold R0, irrelevant information interference such as vegetation and pits is excluded, and a suspected water body part is extracted; The UAV flies again to patrol the suspected water body part and take a hyperspectral image, and the threshold is reset to distinguish water bodies of different pH values, so as to screen rainfall accumulated water or reservoir seepage water. The area with pH≥threshold value is extracted from the hyperspectral image as a seepage area, a reflectance R1 curve is drawn, the wave band peak and trough are observed, the statistical parameters R1' of the spectral reflectance of the seepage area are obtained, and the area S1 of the seepage area is calculated; The UAV flies again after a preset time T, patrols the seepage area and takes a hyperspectral image, and the statistical parameters of the spectral reflectance of the seepage area and the area of the seepage area are recalculated; The ratio of the statistical parameters of the spectral reflectance and the ratio of the areas of the seepage areas in the two flights represent the change in water content and the diffusion degree of the seepage area, and the values of the water content change and the diffusion degree are used to evaluate the seepage degree; The pH threshold is determined by taking a hyperspectral image of the rainfall accumulated water area and the seepage water area in the earth-rock dam, and analyzing the spectral reflectance values of the rainfall accumulated water area and the seepage water area in the water body characteristic band. Based on the values, a threshold value is set to distinguish rainfall accumulated water and seepage water; The spectral reflectance of the hyperspectral image is subjected to first-order differential processing, and the suspected water body part and the seepage area are extracted; The statistical parameters of the spectral reflectance are the minimum values of the spectral reflectance of the seepage characteristic band; When flying again, a reflectance R2 curve is drawn, the statistical parameters R2' of the spectral data reflectance are obtained, the area S2 of the water seepage area is calculated, the reflectance change r=R1' / R2' is calculated, that is, the water content; The area change rate k=S2 / S1 is calculated, that is, the diffusion degree; The method for evaluating the seepage degree by using the values of the water content change and the diffusion degree is as follows: When 1 When 1.2≤r and 1.3≤k<1.5, it is determined to be a rapid seepage stage; When r=1 and 1.5 2. The method of claim 1, wherein, The pH threshold is set to 7.

3. The method of claim 1, wherein, After identifying the suspected water body part or the seepage area, the positioning and flight parameters of the UAV are recorded to control the flight position of the UAV.

4. The method of claim 1, wherein, An alarm is triggered after identifying the suspected water body part, and the positioning and flight parameters of the UAV are recorded.

5. The method of claim 1, wherein, Further, after determining the seepage area in the first flight, a PVC pipe is used to form a regular quadrilateral containing the seepage area, and a seepage space is reserved between the quadrilateral and the seepage area. The area of the quadrilateral is used as a reference to calculate the actual seepage area in the second flight and the third flight.

6. Apparatus for use in the method of any one of claims 1 to 5 characterised in that, The method comprises a UAV (1), a hyperspectral imager (2) carried on the UAV (1), a ground remote control console (5), a PVC pipe (10) and a gray cloth (11); The PVC pipe (10) forms a regular quadrilateral containing the seepage area; The gray cloth (11) is placed in the shooting area. The ground remote control station (5) receives hyperspectral images taken by the hyperspectral imager (2) and performs data processing, including: A classification identification and alarm module (6), a regional area estimation module (7), a spectral feature extraction module (8), and a leakage degree evaluation module (9); the classification identification and alarm module (6) extracts suspected water body parts and leakage areas based on preset characteristic band spectral reflectance thresholds and pH thresholds, and alarms when a suspected water body part is found; the regional area estimation module (7) estimates the area of the extracted leakage area; the spectral feature extraction module (8) obtains statistical parameters of the spectral reflectance of the leakage area; and the leakage degree evaluation module (9) represents the water content change and the diffusion degree of the leakage area based on the ratio of the statistical parameters of the spectral reflectance of two times of reflight and the area ratio of the leakage area, and evaluates the leakage degree by using the numerical values of the water content change and the diffusion degree.

Images