Testing device and method for measuring leakage process of tunnel face based on machine vision

Abstract

The invention relates to a test device and method for measuring the leakage process of a tunnel face based on machine vision, and the device comprises a water tank, and is characterized in that a concrete outer box which encloses the water tank is arranged outside the water tank, one side wall of the concrete outer box is set as a simulated tunnel face, and the other side wall of the concrete outer box is set as a water tank. A plurality of crack-shaped holes and point-shaped holes are formed in the simulated tunnel face in a chiseling mode, an annular gap layer is formed between the concrete outer box and the water tank, the gap layer is filled with soil and gravel, and a plurality of water outlets communicated with the gap layer are formed in one side of the water tank. According to the test device for measuring the tunnel face leakage process based on machine vision, the mountain tunnel face water leakage process is accurately restored by simulating the real surrounding rock environment, multi-dimensional data acquisition and controllable test working conditions are achieved, and a high-quality data set is generated; and a solid data support is provided for accurate prediction of tunnel leakage water and training of an intelligent monitoring model.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering experimental technology, specifically to an experimental device and method for measuring the leakage process at the tunnel face based on machine vision. Background Technology

[0002] Water leakage is a common problem during the construction and operation of mountain tunnels, seriously threatening the structural safety and operational stability of the tunnels. It not only directly threatens construction safety but also easily leads to long-term deterioration of the tunnel structure, severely affecting the durability and operational stability of the project. In order to achieve accurate prediction and intelligent monitoring of tunnel water leakage, the training of deep learning models urgently requires a large amount of real, multi-scenario water leakage process data as support.

[0003] However, on-site data acquisition suffers from significant drawbacks, including high costs, lengthy cycles, uncontrollable operating conditions, difficulty in data collection, and insufficient data volume to meet model training requirements. Furthermore, the actual leakage volume in real-world engineering projects is difficult to measure. Existing indoor simulation devices are mostly limited to simulating single leakage scenarios and have limited data acquisition dimensions, failing to provide rich and realistic multi-source datasets for deep learning model training. Therefore, there is an urgent need for an indoor experimental device capable of accurately reproducing the leakage process at the working face of mountain tunnels and collecting data from multiple dimensions to generate high-quality model training datasets, which has become a pressing technical problem to be solved in this field. Summary of the Invention

[0004] In view of the shortcomings of the prior art, the technical problem to be solved by the present invention is to provide an experimental device and method for measuring the seepage process of tunnel face based on machine vision. It is not only reasonably structured, but also realizes the real simulation of the seepage process by restoring the interaction between the surrounding rock (soil, gravel layer) and the seepage water channel (point-like, crack-like pores). Furthermore, it uses a standing 3D laser scanner, camera and other equipment to collect multi-dimensional data at the same time, providing a rich and accurate dataset for subsequent deep learning model training.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is: a test device for measuring the leakage process of tunnel face based on machine vision, including a water tank, and a concrete outer box surrounding the water tank. One side wall of the concrete outer box is set as a simulated tunnel face. The simulated tunnel face is engraved with a number of crack-like holes and dot-like holes. An annular void layer is formed between the concrete outer box and the water tank. The void layer is filled with soil and gravel. A number of water outlets communicating with the void layer are opened on one side of the water tank.

[0006] Furthermore, a vertically extending scale is embedded in the inner wall of the water tank, and the scale has graduation lines along the vertical extension direction.

[0007] Furthermore, the water tank is enclosed by acrylic glass panels, with an open top.

[0008] Furthermore, the water outlets are arranged in an array on the side of the water tank near the simulated tunnel face.

[0009] Furthermore, a camera for observation is installed on the outer side of the simulated tunnel face.

[0010] Furthermore, a three-dimensional laser scanner for scanning the seepage area is installed on the outer side of the simulated tunnel face.

[0011] Furthermore, both the crack-like holes and the dot-like holes are sealed with silicone plugs.

[0012] Furthermore, an L-shaped steel support is fixed to the bottom of the outer side of the simulated tunnel face for reinforcement, and the horizontal section of the steel support is fixed to the ground.

[0013] The working method of the test device for measuring the leakage process of tunnel face based on machine vision is as follows: First, before the test: the void layer is not filled with soil and gravel, and the sealing performance of the water tank and the void layer is checked to complete the sealing performance verification; During the test: the void layer is filled with soil and gravel, and all crack-like holes and point-like holes on the simulated tunnel face are sealed with silicone plugs. Test water is added to the water tank, and the water level in the cavity is monitored and recorded in real time using a ruler; ensure that the soil and gravel materials in the void layer are fully saturated with water, and then selectively remove the silicone plugs of the corresponding point-like holes or crack-like holes on the simulated tunnel face; at this time, the test water seeps into the void layer filled with soil and gravel, and under the action of gravity, it seeps along the material pores to the simulated tunnel face, and finally flows out through the opened holes, realizing the accurate simulation of the target seepage phenomenon, and is recorded by a camera and a 3D laser scanner.

[0014] Furthermore, the three-dimensional laser scanner continuously performs three-dimensional scanning of the seepage area of ​​the simulated tunnel face at a frequency of 1 million points per second, collecting point cloud data of the seepage point location, crack seepage range, and water flow coverage area. The camera continuously captures dynamic images of the seepage area and records changes in water flow morphology. The water level data on the scale is read once every 15 minutes to record the rate of water level drop in the tank.

[0015] Compared with existing technologies, the present invention has the following advantages: The present invention uses a pre-reserved void layer between the water tank composed of a concrete outer box and an acrylic glass plate, filled with soil and gravel, and combined with the water outlets on both sides to regulate the water flow infiltration path, which can realistically reproduce the surrounding rock environment of the tunnel; the simulated tunnel face is made of cement with pre-reserved point-like and crack-like holes, which can simulate various scenarios such as point-like and crack-like seepage / rushing water in mountain tunnels, solving the problem of the single scenario of the test device; by using a 3D laser scanner to collect 3D point cloud of the seepage area and a camera to collect dynamic video, multi-dimensional information of spatial morphology and dynamic process is obtained simultaneously, and subsequent registration processing forms a continuous time series point cloud dataset, which can effectively improve the training effect of deep learning models.

[0016] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. Attached Figure Description

[0017] Figure 1 This is a schematic diagram illustrating the structure of an embodiment of the present invention; Figure 2 This is a top view of an embodiment of the present invention; Figure 3 This is a front view of an embodiment of the present invention.

[0018] In the diagram: 1-Water tank, 2-Concrete outer casing, 3-Tunnel face, 4-Crack-like pores, 5-Point-like pores, 6-Void layer, 7-Soil and gravel, 8-Water outlet, 9-Acrylic glass plate, 10-Scale ruler, 11-Camera, 12-3D laser scanner, 13-Steel support. Detailed Implementation

[0019] To make the above features and advantages of the present invention more apparent and understandable, specific embodiments are described below in conjunction with the accompanying drawings for detailed explanation.

[0020] like Figures 1-3 As shown, the experimental device for measuring the leakage process of a tunnel face based on machine vision includes a water tank 1. A concrete outer box 2 is set outside the water tank to enclose the water tank. One side wall of the concrete outer box is set as a simulated tunnel face 3. Several crack-like holes 4 and dot-like holes 5 are carved on the simulated tunnel face. An annular void layer 6 is formed between the concrete outer box and the water tank. The void layer is filled with soil and gravel 7. Several water outlets 8 communicating with the void layer are opened on one side of the water tank.

[0021] In this embodiment of the invention, the water tank is composed of acrylic glass panels to simulate a surrounding rock environment. The water tank is square with an open top. Two acrylic glass panels, each 37cm x 35cm, correspond to the length of the simulated tunnel seepage system; the other two panels, each 17cm x 35cm, correspond to the width. Adjacent acrylic glass panels are sealed with adhesive to prevent leakage from the sidewalls of the tank.

[0022] In this embodiment of the invention, the water tank serves as a water source storage device for the entire experiment, providing test water for the entire experimental device. The water tank can hold up to 32.34L, and the test water can be directly transported to the surrounding rock environment system.

[0023] In this embodiment of the invention, a 5cm space is reserved between the water tank and the concrete outer box to form a void layer; the void layer is filled with soil and gravel according to the particle size distribution ratio of the actual surrounding rock of the mountain tunnel, so as to restore the real surrounding rock environment.

[0024] In this embodiment of the invention, three water outlets with a diameter of 20mm are opened along the length direction on an acrylic glass plate near the simulated tunnel face. These three water outlets form a group, and multiple groups are set at 10cm intervals along the height direction. At the start of the test, water is injected into the water tank so that the water can flow out from the reserved water outlets.

[0025] In this embodiment of the invention, a vertically extending scale 10 is embedded in the inner wall of the water tank. The scale has scale lines along the vertical extension direction for real-time monitoring of water level changes in the surrounding rock environment cavity.

[0026] In this embodiment of the invention, a cuboid with dimensions of 50cm×30cm×35cm is formed by cement casting, with an inner cavity of 42cm×22cm×35cm and an opening at the top (without a closed top surface). One of the 50cm×35cm surfaces is selected as a simulated tunnel face, and the surface is manually chiseled to form an uneven cross-section to reproduce the rough shape of a real tunnel face.

[0027] Different sizes of point-like holes and crack-like holes were pre-set at the simulated tunnel face: six point-like holes with a diameter of 5-8 mm were randomly distributed; six crack-like holes with a length of 5-15 cm and a width of 1-3 mm were also randomly distributed. All pre-set holes were initially sealed with removable silicone plugs to ensure the cavity was airtight before the test. During the test, the silicone plugs of the corresponding holes could be selectively removed according to the working conditions, allowing the point-like holes and crack-like holes to be opened individually or in combination. The point-like holes were used to simulate localized seepage points at the tunnel face, while the crack-like holes were used to simulate seepage from rock fissures at the tunnel face, achieving accurate simulation of different seepage types.

[0028] In this embodiment of the invention, the water outlets are arranged in an array on the side of the water tank near the simulated tunnel face.

[0029] In this embodiment of the invention, a camera 11 for observation is provided on the outer side of the simulated tunnel face.

[0030] In this embodiment of the invention, a three-dimensional laser scanner 12 for scanning the seepage area is provided on the outer side of the simulated tunnel face.

[0031] In this embodiment of the invention, both the crack-like holes and the dot-like holes are sealed with silicone plugs.

[0032] In this embodiment of the invention, an L-shaped steel bracket 13 is fixedly connected to the bottom of the outer side of the simulated tunnel face for reinforcement. The horizontal section of the steel bracket is fixed to the ground. The simulated tunnel water leakage system is fixed as a whole by the steel bracket. The height of the steel bracket is adjusted to make the device horizontal, so as to avoid water flow deviation due to tilting during the test.

[0033] In this embodiment of the invention, a three-dimensional laser scanner is placed 1.5m in front of the simulated tunnel face. By adjusting the scanning angle, the scanning range completely covers the entire water leakage area of ​​the simulated tunnel face—including point-like holes and crack-like holes. Then, the scanner is started and its scanning parameters are confirmed to meet the preset requirements: scanning speed of 1 million points per second, point accuracy of 1.9mm, and distance measurement accuracy of 1mm+10ppm.

[0034] In this embodiment of the invention, a high-definition video camera is mounted 10cm to the right of the 3D laser scanner. The height of the instrument is adjusted so that the center of the lens is aligned with the center of the simulated tunnel face. The lens focal length is adjusted to clearly capture the water seepage area. The core parameters are confirmed as follows: resolution 1920×1080, frame rate 25fps, panoramic horizontal field of view ≥160°, and vertical field of view ≥80°.

[0035] The working method of the test device for measuring the leakage process at the tunnel face based on machine vision is as follows: Before the test: Do not fill the void layer with soil and gravel, and completely seal all the pre-set holes on the simulated tunnel face with silicone plugs; slowly inject test water into the water tank, controlling the water depth to about 5cm, and then let it stand for 30 minutes; during this period, pay close attention to whether there is any leakage on the side wall of the cavity and the water outlet, and continuously check whether the scale reading is stable; after confirming that there is no leakage in the device, completely drain the water in the cavity to complete the sealing test; The 3D laser scanner was activated to pre-scan the simulated tunnel face to obtain initial 3D point cloud data. The integrity of the point cloud data and the accuracy of the points were checked to ensure they met the test requirements. A 10-minute test video was recorded using a high-definition video camera. The video was then played back to confirm that the water flow observation area was clearly captured without any stuttering or blurring, ensuring that the data acquisition equipment was operating normally.

[0036] During the experiment: the void layer was filled with soil and gravel, and all the pre-set holes on the simulated tunnel face were sealed with silicone plugs. 30L of test water was added to the water tank, and the water level in the cavity was monitored and recorded in real time using a ruler. When the water level rose to 20cm, it was left to stand for 10-15 minutes to ensure that the soil and gravel materials in the void layer were fully saturated with water, laying the foundation for simulating the real seepage environment. According to the pre-set seepage type in the experiment, the silicone plugs of corresponding point-like holes or crack-like holes on the simulated tunnel face are selectively removed; at this time, the test water in the surrounding rock environment cavity seeps into the void layer filled with soil and gravel through the outlet of the acrylic glass inner box, and under the action of gravity, it seeps along the material pores to the simulated tunnel face, and finally flows out through the opened holes, so as to achieve accurate simulation of the target seepage phenomenon. Simultaneously, a 3D laser scanner is activated to continuously scan the seepage area of ​​the simulated tunnel face at a frequency of 1 million points per second, collecting point cloud data on the location of seepage points, the seepage range of fissures, and the area covered by water flow, ensuring data integrity and storing the data in .las format; a high-definition video camera continuously captures dynamic images of the seepage area throughout the process, recording changes in water flow patterns, and the image data is automatically segmented and stored every 30 minutes for subsequent matching and analysis with the 3D point cloud data; The testers read the water level data on the scale every 15 minutes, recorded the rate of water level drop in the surrounding rock environment cavity, and observed whether there was any abnormal leakage at the simulated tunnel face. If a blockage was found, the test was immediately suspended, and the blockage was cleared with a 0.5mm diameter metal wire before the test continued. When the water level in the simulated surrounding rock environment chamber drops to 5cm, first turn off the 3D laser scanner, then turn off the high-definition video camera, wait for the test water in the water tank to be drained, clean the soil and gravel material in the void layer, and properly store it for reuse in the next test. The collected 3D point cloud data, dynamic image data, and water level monitoring data are systematically organized and accurately labeled, classified and archived according to the preset seepage types, and a standardized dataset is generated. This dataset can be directly used for the training and optimization of deep learning models, providing data support for the subsequent research and development of intelligent identification technology for seepage at the working face of mountain tunnels.

[0037] The device of this invention can accurately simulate the dynamic process of water seepage at the working face of mountain tunnels. It synchronously collects water seepage test data using a 3D laser scanner and a camera, and after data preprocessing and registration, forms a continuous time-series point cloud dataset. This dataset is rich, realistic, and reproducible, providing high-quality samples for training deep learning models. It has significant engineering value and broad application prospects for improving the level of intelligent monitoring and prediction technology for tunnel water seepage.

[0038] This invention is not limited to the preferred embodiment described above. Anyone inspired by this invention can derive various other forms of experimental apparatus and methods for measuring tunnel face leakage based on machine vision. All equivalent variations and modifications made within the scope of the claims of this invention should be considered within the scope of this invention.

Claims

1. An experimental device for measuring the leakage process at the tunnel face based on machine vision, comprising a water tank, characterized in that: The water tank is surrounded by a concrete outer box. One side wall of the concrete outer box is designed to simulate a tunnel face. Several crack-like holes and dot-like holes are carved on the simulated tunnel face. An annular void layer is formed between the concrete outer box and the water tank. The void layer is filled with soil and gravel. Several water outlets communicating with the void layer are opened on one side of the water tank.

2. The experimental device for measuring tunnel face leakage process based on machine vision according to claim 1, characterized in that: The inner wall of the water tank is fitted with a vertically extending scale.

3. The experimental device for measuring tunnel face leakage process based on machine vision according to claim 2, characterized in that: The water tank is enclosed by acrylic glass panels, with an open top.

4. The experimental device for measuring tunnel face leakage process based on machine vision according to claim 1, characterized in that: The water outlets are arranged in an array on the side of the water tank near the simulated tunnel face.

5. The experimental device for measuring tunnel face leakage process based on machine vision according to claim 1, characterized in that: A camera for observation is installed on the outside of the simulated tunnel face.

6. The experimental device for measuring the seepage process at the tunnel face based on machine vision according to claim 1, characterized in that: A three-dimensional laser scanner is installed on the outer side of the simulated tunnel face to scan the seepage area.

7. The experimental device for measuring tunnel face leakage process based on machine vision according to claim 1, characterized in that: Both the crack-like and dot-like holes were sealed with silicone plugs.

8. The experimental device for measuring the seepage process at the tunnel face based on machine vision according to claim 1, characterized in that: The simulated tunnel face has an L-shaped steel support fixed to the bottom of the outer side for reinforcement.

9. The working method of the experimental device for measuring the leakage process at the tunnel face based on machine vision, characterized in that, The test apparatus for measuring the leakage process of tunnel face based on machine vision, as described in any one of claims 1-8, is used, and the following steps are performed: First, before the test: the void layer is not filled with soil and gravel, and the sealing performance of the water tank and the void layer is checked to complete the sealing performance verification; During the test: the void layer is filled with soil and gravel, and all crack-like holes and point-like holes on the simulated tunnel face are sealed with silicone plugs. Test water is added to the water tank, and the water level in the cavity is monitored and recorded in real time using a ruler; ensure that the soil and gravel materials in the void layer are fully saturated with water, and then selectively remove the silicone plugs corresponding to point-like holes or crack-like holes on the simulated tunnel face; at this time, the test water seeps into the void layer filled with soil and gravel, and under the action of gravity, it permeates along the material pores to the simulated tunnel face, and finally flows out through the opened holes, realizing the accurate simulation of the target seepage phenomenon, and is recorded by a camera and a 3D laser scanner.

10. The working method of the experimental device for measuring the seepage process at the tunnel face based on machine vision according to claim 9, characterized in that: The three-dimensional laser scanner continuously performs three-dimensional scanning of the seepage area of ​​the simulated tunnel face at a frequency of 1 million points per second, collecting point cloud data of the seepage point location, the seepage range of the cracks, and the area covered by the water flow. The camera continuously captures dynamic images of the seepage area and records the changes in the water flow pattern. The water level data of the scale is read once every 15 minutes to record the rate of decline of the water level in the tank.

Images