A method and device for detecting soil particle start-up loss during dam leakage process

Abstract

The present invention discloses a method and device for detecting the start-up loss of soil particles during the dam leakage process. The method first collects temperature data, visible light images and thermal infrared images of the downstream dam surface of the dam, and calibrates and fuses the visible light image and the thermal infrared image to obtain a dual-light fusion image; then the dam leakage area is determined according to the dual-light fusion image and temperature data; finally, the isothermal circle is extracted from the dual-light fusion image, and a temperature gradient detection model is established according to the threshold value and horizontal semi-axis length of the isothermal circle and the temperature data, and the temperature gradient detection model is used to determine whether the start-up loss of soil particles occurs in the dam leakage area. The present invention can efficiently sense the critical start-up of soil particles during the dam leakage process and realize real-time monitoring of the evolution of the dam leakage lesion.

Description

Technical Field

[0001] The invention relates to the technical field of dam leakage detection, in particular to a method and a device for detecting the start-up loss of soil particles during a dam leakage process. Background Art

[0002] The leakage of earth-rock dams is characterized by concealment, temporal and spatial randomness, and a small initial magnitude. The time from the occurrence of leakage to the serious damage of the dam is very short. When seepage occurs, the water drag force on the cohesive soil particles destroys the original static equilibrium conditions of the particles. The fine particles of the soil and even the finer particles in the skeleton will flow out with the seepage. The pores between the coarser particles are not filled with skeleton particles with smaller particle sizes, forming a seepage channel inside the soil. That is, under the action of seepage, the finer particles continue to migrate and lose through the matrix gaps formed by the coarser particles, causing the internal structure of the soil to change, resulting in the instability of the entire soil and collapse. According to statistics, about 50% of the dam failures in the world are caused by the loss of soil particles caused by seepage. The continuous loss of soil particles causes the diseased parts to be gradually eroded and hollowed out, which will have serious adverse effects on projects such as gravel foundations and embankment structures. Timely discovery and accurate positioning of the start-up loss of cohesive soil particles is the key to ensuring the safety of earth-rock dams.

[0003] The initial loss of cohesionless soil particles is an intuitive manifestation of seepage deformation, which is affected by both the geometric conditions of the soil itself and the external hydraulic conditions. Geometric conditions include particle grading, density, porosity, etc., which determine the stability of the internal structure of the soil, and the stability of the internal structure is determined by the particle grading and relative density of the soil; hydraulic conditions include the size and direction of the hydraulic gradient, which is the external condition that drives the initial loss of soil particles and causes seepage deformation, and is generally described by the critical hydraulic gradient. Although a lot of research has been done on particle initiation, most of them are carried out by establishing scaled models in the laboratory, ignoring the complexity of actual leakage in dam projects on site, and unable to truly restore the leakage form of the dam, making it difficult to efficiently and accurately determine the initial loss of dam leakage particles. Summary of the invention

[0004] Purpose of the invention: The purpose of the present invention is to provide a method and device for detecting the initial loss of soil particles in the dam leakage process, which is simple to calculate and is not affected by factors such as the dam particle grading, density and porosity.

[0005] Technical solution: The method for detecting the start-up loss of soil particles during the dam leakage process of the present invention comprises the following steps:

[0006] Collect temperature data, visible light images and thermal infrared images of the downstream dam surface;

[0007] Calibrate and fuse the visible light image and the thermal infrared image to obtain a dual-light fusion image;

[0008] Determine the leakage area of ​​the dam according to the dual-light fusion image and the temperature data;

[0009] Extracting isotherms from the dual-light fusion image, and establishing a temperature gradient detection model according to the threshold and horizontal semi-axis length of the isotherms and the temperature data;

[0010] It is determined whether soil particles start to lose in the leakage area of ​​the dam according to the temperature gradient detection model.

[0011] Furthermore, establishing a temperature gradient detection model according to the threshold value and horizontal semi-axis length of the isothermal circle and the temperature data includes:

[0012] The temperature data includes the reservoir water temperature T 水 and the temperature T 气 ;

[0013] When T 气 >T 水 When the temperature gradient is determined by the model k = G / g, where g = (T1-T 水 ) / x1, G=(T2-T 水 ) / x2, T1 and T2 are respectively the first isothermal threshold and the second isothermal threshold in the dual-light fusion image, and x1 and x2 are respectively the horizontal semi-axis lengths of T1 and T2;

[0014] When T 气 <T 水 When the temperature gradient is determined by the model k′=G′ / g′, where g′=|(T1′-T 水 ) / x1′|,G′=|(T2′-T 水 ) / x2′|, T1′ and T2′ are respectively the third isothermal threshold and the fourth isothermal threshold in the dual-light fusion image, and x1′ and x2′ are respectively the horizontal semi-axis lengths of T1′ and T2′;

[0015] When k or k′ exceeds the gradient threshold, it is determined that soil particles have started to lose in the leakage area of ​​the dam.

[0016] Furthermore, the threshold increment ΔT in the dual-light fusion image is 气 -T 水 ) / 2|;

[0017] When T 气 >T 水 When T1=T 水 +ΔT, T2 = T 水 +2ΔT;

[0018] When T 气 <T 水 When T1′=T水 -ΔT, T2′=T 水 -2ΔT.

[0019] Furthermore, the visible light image and the thermal infrared image are calibrated and fused by a method based on non-subsampled Contourlet transform to obtain a dual-light fusion image.

[0020] Furthermore, isothermal circles are extracted from the dual-light fusion image by using an adaptive threshold binarization algorithm based on the maximum inter-class variance method.

[0021] Furthermore, visible light images and thermal infrared images of the downstream dam surface are collected by using a drone equipped with a visible light camera and an infrared thermal imager.

[0022] The device for detecting soil particle start-up loss during dam leakage of the present invention comprises:

[0023] A collection unit, used to collect temperature data, visible light images and thermal infrared images of the downstream dam surface;

[0024] A dam leakage area determination unit is used to calibrate and fuse the visible light image and the thermal infrared image to obtain a dual-light fusion image; and determine the dam leakage area according to the dual-light fusion image and the temperature data;

[0025] The soil particle start-up loss detection unit is used to extract the isotherm circle from the dual-light fusion image, establish a temperature gradient detection model according to the threshold value and horizontal semi-axis length of the isotherm circle and the temperature data; and judge whether the soil particle start-up loss occurs in the dam leakage area according to the temperature gradient detection model.

[0026] Furthermore, in the acquisition unit, the temperature data includes the reservoir water temperature T 水 and the temperature T 气 ;

[0027] In the soil particle start-up loss detection unit, when T 气 >T 水 When the temperature gradient is determined by the model k = G / g, where g = (T1-T 水 ) / x1, G=(T2-T 水 ) / x2, T1 and T2 are respectively the first isothermal threshold and the second isothermal threshold in the dual-light fusion image, T1=T 水 +ΔT, T2 = T 水 +2ΔT,ΔT=|(T 气 -T 水 ) / 2|; x1 and x2 are the horizontal semi-axis lengths of T1 and T2 respectively;

[0028] When T 气 <T水 When the temperature gradient is determined by the model k′=G′ / g′, where g′=|(T1′-T 水 ) / x1′|,G′=|(T2′-T 水 ) / x2′|, T1′ and T2′ are respectively the third isothermal threshold and the fourth isothermal threshold in the dual-light fusion image, ΔT=|(T 气 -T 水 ) / 2|, T1′=T 水 -ΔT, T2′=T 水 -2ΔT; x1′ and x2′ are the horizontal semi-axis lengths of T1′ and T2′ respectively;

[0029] When k or k′ exceeds the gradient threshold, it is determined that soil particles have started to lose in the leakage area of ​​the dam.

[0030] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method for detecting the start-up loss of soil particles during the dam leakage process is implemented.

[0031] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the method for detecting the start-up loss of soil particles during the dam leakage process is implemented.

[0032] Beneficial effect: Compared with the prior art, the advantages of the present invention are: the present invention can not only effectively avoid the calculation of complex hydraulic models to obtain the law of soil particle migration, but also can efficiently sense the critical start of soil particles during dam leakage without considering the influence of factors such as dam particle grading, density and porosity, and realize real-time monitoring and early warning of the evolution of dam leakage lesions. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A diagram of a detection framework for soil particle startup loss according to an embodiment of the present invention;

[0034] Figure 2 It is a schematic diagram of thermal infrared image characteristics of the downstream dam slope leakage area when the soil particles do not start to lose in the embodiment of the present invention;

[0035] Figure 3 Schematic diagram of thermal infrared image characteristics of the downstream dam slope leakage area when soil particles start to lose in an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0037] The detection of soil particle start-up loss during dam leakage of the present invention comprises the following steps:

[0038] Step 1: Use a drone equipped with an infrared thermal imager to inspect the entire downstream dam surface and calculate the water temperature T 水 and the temperature T 气 Set the temperature control range [k min ,k max ]; When the infrared thermal image of the downstream dam surface shows abnormal temperature within the temperature control range, it is a suspected dam leakage area.

[0039] Temperature control range [k min ,k max ] can help locate the target object quickly and accurately, k min Reservoir water temperature T 水 and the temperature T 气 The minimum value of k max Reservoir water temperature T 水 and the temperature T 气 The maximum value of .

[0040] Step 2: Since the visible light and thermal infrared dual-light lenses work together, they complement each other and use the heterogeneous image fusion criteria to register the visible light and thermal infrared dual-light heterogeneous images to obtain a dual-light fusion image, so as to accurately determine the leakage area of ​​the dam.

[0041] The heterogeneous image fusion criterion is based on the non-subsampled Contourlet transform (NSCT), which makes the temperature, contour, texture and other detailed information of the target area clearly discernible. With the help of visible light images, it is possible to identify whether there is leakage water flow in the temperature anomaly area in the thermal infrared image and eliminate the interference of "quasi-leakage" temperature anomaly points.

[0042] Step 3: The drone has an AI reset module. Through the precise reset function enabled by AI, it automatically reaches the target leakage area and uses the segmentation algorithm to extract the isotherms of the leakage area.

[0043] The AI ​​algorithm will autonomously compare the target area with the current real-time image, and correct the camera's shooting angle accordingly to ensure that the same target area can be photographed each time.

[0044] The segmentation algorithm uses an adaptive threshold binarization algorithm based on the maximum between-class variance method (OTSU), and the threshold increment ΔT = |(T 气 -T 水 ) / 2|, the dual-light fusion image is segmented. When T 气 >T 水 When the isotherm threshold is extracted, T1 = T 水 +ΔT, isothermal threshold T2 = T 水 +2ΔT; when T 气 <T 水 When the isotherm threshold is extracted, T1′=T 水 -ΔT, isothermal threshold T2′=T水 -2ΔT.

[0045] Step 4: Establish a training temperature gradient judgment model to monitor and identify in real time whether the seepage soil particles in the dam are moving.

[0046] When T 气 >T 水 When the temperature gradient is determined, the model uses k = G / g as the criterion. Where g = (T1-T 水 ) / x1, G=(T2-T 水 ) / x2, x1, x2 are the horizontal semi-axis lengths of the isothermal circles T1, T2 respectively; when T 气 <T 水 When the temperature gradient determination model is based on k'=G' / g' as the determination criterion. Where g'=|(T1'-T 水 ) / x1′|,G′=|(T2′-T 水 ) / x2′|, x1′ and x2′ are the horizontal semi-axis lengths of the isothermal circles T1′ and T2′ respectively.

[0047] like Figure 2 and Figure 3 As shown in the figure, in the early stage of seepage, the overflow position of the downstream dam surface has stable thermal diffusion, that is, x2 = 2x1, x2′ = 2x1′. Once the soil particles start to lose, an obvious leakage channel is formed, and the overflow position of the downstream dam surface has unstable thermal diffusion. At this time, x2 < 2x1, x2′ < 2x1′. Therefore, the threshold of the temperature gradient identification model is set to 1. When k> 1 or k′> 1, it means that the soil particles in the leakage channel of the dam have started to lose.

[0048] like Figure 1 The diagram is a diagram of the detection architecture of soil particle start-up loss in this embodiment, in which 1, leakage channel; 2, earth-rock dam; 3, leakage outlet; 4, downstream dam surface; 5, drone; 6, infrared thermal imager; 7, control console; 8, temperature control interval module; 9, image fusion module; 10, A1 reset module; 11, image segmentation module; 12, temperature gradient recognition model; 13, remote control system; 14, cabin; 15, thermometer. Figure 1 The method of the present invention comprises the following steps:

[0049] (1) Planning inspection system

[0050] The drone (5) uses the DJI Matrice 350RTK, which has IP55 protection, 55 minutes of ultra-long flight time (maximum flight time when unloaded), and can withstand winds of up to level 7, making it suitable for operations in complex environments. In addition, the mounted infrared thermal imager (6) uses the Zenmuse H20T, which intuitively presents visible light and thermal imaging images. The drone smart cabin 14 uses the Zhongke Yuntu Zhizhen G900, which has 4 sets of drone batteries built in. The robot arm can realize automatic battery replacement of the drone every 120 seconds, allowing the drone to operate uninterruptedly. It is also equipped with industrial-grade three-dimensional monitoring, constant temperature and humidity, and meteorological systems to fully guarantee the safe operation of the drone and the smart cabin.

[0051] (2) Flight mission preparation

[0052] The height of a certain earth-rock dam (2) is 5m. According to the inspection task requirements, the remote control system (13) in the control console (7) is used to plan the take-off position to be the left shoulder of the dam body, the inspection route is from the left shoulder of the dam body to the right shoulder, and the landing position is the right shoulder of the dam body, completing the inspection of the entire dam body at one time; the flight altitude of the drone during the inspection is 20m, the thermal imager field of view needs to cover the entire width of the downstream dam surface, and the flight process is not affected by obstacles. The reservoir water temperature is measured by the thermometer (15) to be 15℃ and the atmospheric temperature is 27℃.

[0053] (3) Deployment of flight missions

[0054] After the drone route is planned online through the control console (7), a task instruction is sent to the cabin (14) through the Internet or the local area network. After receiving the instruction, the cabin (14) automatically starts the drone (5) and loads the battery. After passing the safety self-check, it takes off and cruises by relying on the navigation and positioning system. When the drone (5) completes the inspection task, it automatically flies back to the cabin (14), lands in the cabin (14) by relying on visual positioning, and automatically takes out the drone battery to charge it, completing the entire automated inspection process.

[0055] (4) Real-time online monitoring

[0056] The entire downstream dam surface (4) is inspected by using an infrared thermal imager mounted on a drone. The temperature range module (8) in the control console (7) sets the temperature control range based on the reservoir water temperature of 15°C and the air temperature of 27°C [15,27], thereby narrowing the target temperature range and determining the suspected leakage area. Furthermore, based on the non-subsampled Contourlet transform (NSCT), the image fusion module (9) in the control console (7) uses the heterogeneous image fusion criterion to register the visible light and thermal infrared images. It is found that the dam surface leakage outlet (3) has abnormal temperature and water flow, which can accurately determine the leakage area of ​​the dam.

[0057] Next, the control console (7) uses the AI ​​reset module (10) to control the drone (5) to accurately monitor the evolution of the leakage channel (1) in real time. The image segmentation module (11) in the control console (7) uses the maximum between-class variance method (OTSU) segmentation algorithm and an adaptive threshold binarization algorithm with a threshold increment ΔT = |(27-15) / 2| = 6°C to segment the dual-light fusion image of the leakage outlet (3) on the dam surface and extract the isothermal zone threshold T1 = T 水 +ΔT=15+6=21℃, isothermal threshold T2=T 水 +2ΔT=15+12=27℃.

[0058] (5) Temperature gradient identification model

[0059] The control panel (7) uses k=G / g in the temperature gradient identification model (12) as the judgment criterion. In the control panel (7), the horizontal semi-axis lengths x1 and x2 of the isothermal circle of the leakage outlet (3) are observed to be 8 mm and 10 mm respectively, then Temperature gradient determination model When the temperature recognition model threshold 1 is reached, the image fusion module (9) shows that water flows at the leakage outlet (3) and the water quality is relatively turbid, indicating that soil particles in the dam leakage channel have started to flow away.

[0060] The device for detecting soil particle start-up loss during dam leakage of the present invention comprises:

[0061] A collection unit, used to collect temperature data, visible light images and thermal infrared images of the downstream dam surface;

[0062] A dam leakage area determination unit is used to calibrate and fuse the visible light image and the thermal infrared image to obtain a dual-light fusion image; and determine the dam leakage area according to the dual-light fusion image and the temperature data;

[0063] The soil particle start-up loss detection unit is used to extract the isotherm circle from the dual-light fusion image, establish a temperature gradient detection model according to the threshold value and horizontal semi-axis length of the isotherm circle and the temperature data; and judge whether the soil particle start-up loss occurs in the dam leakage area according to the temperature gradient detection model.

[0064] Furthermore, in the acquisition unit, the temperature data includes the reservoir water temperature T 水 and the temperature T 气 ;

[0065] In the soil particle start-up loss detection unit, when T 气 >T 水 When the temperature gradient is determined by the model k = G / g, where g = (T1-T 水 ) / x1, G=(T2-T水 ) / x2, T1 and T2 are respectively the first isothermal threshold and the second isothermal threshold in the dual-light fusion image, T1=T 水 +ΔT, T2 = T 水 +2ΔT,ΔT=|(T 气 -T 水 ) / 2|; x1 and x2 are the horizontal semi-axis lengths of T1 and T2 respectively;

[0066] When T 气 <T 水 When the temperature gradient is determined by the model k′=G′ / g′, where g′=|(T1′-T 水 ) / x1′|,G′=|(T2′-T 水 ) / x2′|, T1′ and T2′ are respectively the third isothermal threshold and the fourth isothermal threshold in the dual-light fusion image, ΔT=|(T 气 -T 水 ) / 2|, T1′=T 水 -ΔT, T2′=T 水 -2ΔT; x1′ and x2′ are the horizontal semi-axis lengths of T1′ and T2′ respectively;

[0067] When k or k′ exceeds the gradient threshold, it is determined that soil particles have started to lose in the leakage area of ​​the dam.

[0068] The electronic device described in the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method for detecting the start-up loss of soil particles during the dam leakage process is implemented.

[0069] The computer-readable storage medium of the present invention stores a computer program, and when the computer program is executed by a processor, the method for detecting the start-up loss of soil particles during the dam leakage process is implemented.

[0070] The computer-readable storage media may include RAM, ROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, flash memory, or any other medium that can be used to store the desired program code in the form of instructions or data structures and that can be accessed by a computer.

[0071] The processor is used to execute the computer program stored in the memory to implement each step of the method involved in the above embodiment.

Claims

1. A method for detecting soil particle start-up loss during dam leakage, characterized in that: The steps include: Collect temperature data, visible light images and thermal infrared images of the downstream dam surface; The visible light image and the thermal infrared image are calibrated and fused by a method based on non-subsampled Contourlet transform to obtain a dual-light fusion image; Determine the leakage area of ​​the dam according to the dual-light fusion image and the temperature data; Extracting isotherms from the dual-light fusion image, and establishing a temperature gradient detection model according to the threshold and horizontal semi-axis length of the isotherms and the temperature data; The temperature data includes the reservoir water temperature T 水 and the temperature T 气 ; The threshold increment ΔT in the dual-light fusion image is |(Tair-Twater) / 2|; When T 气 >T 水 When T1=T 水 +ΔT, T2 = T 水 +2ΔT; When T 气 <T 水 When T1′=T 水 -ΔT, T2′=T 水 -2ΔT; When Tgas>Twater, the temperature gradient determination model is k=G / g, where g=(T1-Twater) / x1, G=(T2-T 水 ) / x2, T1 and T2 are respectively the first isothermal threshold and the second isothermal threshold in the dual-light fusion image, and x1 and x2 are respectively the horizontal semi-axis lengths of T1 and T2; When T_gas < T_water, the temperature gradient determination model is k' = G' / g', where g' = |(T1' - T_water) / x1'|, G' = |(T2' - T 水 ) / x2'|, T1' and T2' are respectively the third isothermal circle threshold and the fourth isothermal circle threshold in the dual-light fusion image, and x1' and x2' are respectively the horizontal semi-axis lengths of T1' and T2'; When k or k′ exceeds the gradient threshold, that is, when k>1 or k′>1, it is judged that soil particles start to lose in the leakage area of ​​the dam; It is determined whether soil particles start to lose in the leakage area of ​​the dam according to the temperature gradient detection model.

2. The method for detecting soil particle start-up loss during dam leakage according to claim 1, characterized in that: The isotherm is extracted from the dual-light fusion image by using an adaptive threshold binarization algorithm based on the maximum inter-class variance method.

3. The method for detecting soil particle start-up loss during dam leakage according to claim 1, characterized in that: Visible light images and infrared light images of the downstream dam surface are collected by using a drone equipped with a visible light camera and an infrared thermal imager.

4. A device for detecting soil particle loss during dam leakage, characterized in that: include: A collection unit, used to collect temperature data, visible light images and thermal infrared images of the downstream dam surface; A dam leakage area determination unit is used to calibrate and fuse the visible light image and the thermal infrared image by a method based on non-subsampled Contourlet transform to obtain a dual-light fusion image; and determine the dam leakage area according to the dual-light fusion image and the temperature data; A soil particle start-up loss detection unit is used to extract the isotherm in the dual-light fusion image, establish a temperature gradient detection model according to the threshold and horizontal semi-axis length of the isotherm, and the temperature data; and determine whether soil particle start-up loss occurs in the dam leakage area according to the temperature gradient detection model; In the acquisition unit, the temperature data includes the reservoir water temperature T 水 and the temperature T 气 ; In the soil particle start-up loss detection unit, when Tgas>Twater, the temperature gradient determination model is k=G / g, where g=(T1-T 水 ) / x1, G=(T2-T 水 ) / x2, T1 and T2 are respectively the first isotherm threshold and the second isotherm threshold in the dual-light fusion image, T1=Twater+ΔT, T2=Twater+2ΔT, ΔT=|(Tgas-Twater) / 2|; x1 and x2 are respectively the horizontal semi-axis lengths of T1 and T2; When T_gas < T_water, the temperature gradient determination model is k′ = G′ / g′, where g′ = |(T1′ - T_water) / x1′|, G′ = |(T2′ - T 水 ) / x2′|, T1′ and T2′ are the third isothermal circle threshold and the fourth isothermal circle threshold in the dual - light fusion image respectively, ΔT = |(T 气 - T 水 ) / 2|, T1′ = T 水 - ΔT, T2′ = T 水 - 2ΔT; x1′ and x2′ are the horizontal semi - axis lengths of T1′ and T2′ respectively; When k or k′ exceeds the gradient threshold, that is, when k>1 or k′>1, it is determined that soil particles have started to lose in the leakage area of ​​the dam.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is loaded into the processor, the method for detecting the start-up loss of soil particles in the dam leakage process according to any one of claims 1 to 3 is implemented.

6. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for detecting the start-up loss of soil particles in a dam leakage process according to any one of claims 1 to 3 is implemented.

Images