Test system and test method for subsurface erosion void disease of slope drainage structure and soil contact interface under rainstorm

Abstract

The invention discloses a test system and a test method for subsurface erosion void damage of a slope drainage structure and soil contact interface under rainstorm. Comprising a transparent model box, a rainfall device, a slope runoff device and a data acquisition unit, a rectangular opening is cut in a top plate of the model box, an overflow plate is lapped, the overflow plate is lapped on a slope confluence water storage tank, a hole is formed in the bottom of the water storage tank, a flow meter and a faucet are sequentially connected through a hose, and the slope runoff device is constructed. Two hydraulic jacks are arranged at the bottom of the model box, and camera positions are arranged on the sides, the top and the bottom of the model box. The invention provides two testing methods, a machine learning method is used for carrying out correlation analysis on measured hydrodynamic parameters and sensor data, a deep learning method is used for constructing a data set through camera photos, and a related deep learning model is trained so as to realize classification of different development stages of subsurface erosion and evaluation of damage degrees. The research on the subsurface erosion void disease mechanism of the slope drainage structure and the soil contact interface under the rainstorm is realized.

Description

Technical Field

[0001] This invention belongs to the field of geological disaster testing technology, and specifically relates to a testing system and method for testing the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain. Background Technology

[0002] Brick-concrete drainage structures are a common water control measure for mountain slopes. They intercept runoff from the upper slope via intercepting channels and then discharge it to drainage ditches at the bottom of the slope, thus playing a role in drainage and energy dissipation. According to field investigations, the interface between the drainage structure and the soil in mountainous areas suffers from severe erosion and seepage damage. Small splash craters, small gullies, and shear slippage of the surrounding soil are observed at the interface on various slope sections. In some sections, the bottom of the intercepting channels has become hollowed out, and the drainage channels are either suspended or broken. This erosion and hollowing out at the drainage structure-soil interface damages the slope drainage system and directly affects slope stability.

[0003] Slope erosion and voiding at the soil-soil interface of brick-concrete drainage structures and urban underground structures can be collectively classified as engineering geological interface erosion and voiding diseases. Currently, research on engineering geological interface erosion is limited. Existing research on soil-structure interfaces mainly focuses on the contact characteristics between soil and structures, and seepage at the soil-structure interface under hydraulic or shear forces. Current soil erosion testing equipment primarily concentrates on erosion visualization and seepage erosion testing within slopes, lacking relevant testing equipment and methods for slope drainage structure-soil interface erosion and voiding diseases. This significantly limits research on the mechanisms and prevention of engineering geological interface erosion diseases. Summary of the Invention

[0004] In order to overcome the shortcomings of the prior art, the present invention aims to provide a test system and test method for the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain. The test system uses a rainfall device to simulate extreme rainfall and a slope confluence device to simulate the flow from the upper slope. The test system studies the mechanism of erosion and voiding disease at the interface between the slope drainage structure and the soil under the coupled effect of extreme rainfall and slope confluence.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A testing system for erosion and voiding at the interface between slope drainage structure and soil under heavy rain includes a multi-slope adjustable model box, a rainfall device, a slope runoff device, and a data acquisition unit. The multi-slope adjustable model box includes a transparent model box with a built-in slope drainage structure model. The transparent model box is an inclined box with an open top surface. The upper side plate in the inclined direction has a water inlet, and the lower side plate has a water outlet connected to a water collection tank. The rainfall device includes several rainfall nozzles located above the top surface of the transparent model box, which apply simulated rainfall to the slope drainage structure model; The slope runoff device includes a slope runoff collection and storage tank and an inclined slope runoff overflow plate. The slope runoff collection and storage tank supplies water to the inlet of the transparent model box through the slope runoff overflow plate, thereby applying slope runoff to the slope drainage structure model. The data acquisition unit includes a flow meter and a camera. The flow meter acquires the flow data of the slope runoff device, and the camera acquires image data of the slope drainage structure model during the testing process.

[0006] In one embodiment, the slope drainage structure model includes a brick drainage ditch and loess. Along the runoff direction, the bottom and both sides of the brick drainage ditch are loess, and a pore pressure sensor, a moisture content sensor, and a soil pressure sensor are embedded in the loess.

[0007] In one embodiment, a rectangular opening is cut into the upper side panel of the transparent model box as the water inlet, and the lower edge of the slope overflow plate rests against the water inlet; the lower side panel of the transparent model box has densely packed drainage holes as the water inlet, and the top surface of the water collection tank is open, and the top surface is lower than the height of the lowest drainage hole.

[0008] In one embodiment, the inner wall of the transparent model box is provided with permeable geotextile, and one side of its bottom surface is mounted on a hydraulic jack. The hydraulic jack is used to tilt the transparent model box and adjust its tilt angle.

[0009] In one embodiment, the inlet end of the slope confluence water storage tank is connected to a faucet and the flow meter is installed on the connecting pipe. The slope confluence overflow plate is an inclined plate structure, with its upper edge resting on or connected to the edge of the slope confluence water storage tank and its lower edge connecting to the inlet of the transparent model box.

[0010] In one embodiment, the camera is positioned above, below, and to the side of the transparent model box.

[0011] This invention also provides a testing method for the erosion and voiding disease at the interface between the drainage structure and the soil on a slope under heavy rain. The method is based on the testing system for the erosion and voiding disease at the interface between the drainage structure and the soil on a slope under heavy rain, and the steps are as follows: Step 1: Activate the rainfall device to pre-rain the water. After the water flows in streams on the slope surface of the slope drainage structure model, stop the rainfall and let it stand for 20-30 hours. Step 2: Delineate runoff plots. Each runoff plot has a complete path from top to bottom of the slope drainage structure model and includes the drainage structure-soil contact interface in the width direction. Measure the runoff flow rate, runoff velocity, and runoff width of the runoff plots and calculate the hydrodynamic parameters. Step 3: Use the rainmaking device to generate rain again until a stream of water is formed. Then, activate the slope runoff device to apply stable slope runoff from top to bottom to the slope drainage structure model through the slope confluence overflow plate. Step 4: Use a camera to record the slope erosion process and the changes in the wetting peak within the slope in real time. Step 5: Based on the runoff plots defined in Step 2, measure and calculate the hydrodynamic parameters again; Step 6: Continue rainfall until the contact interface of the slope drainage structure model shows signs of erosion and voiding. Step 7: Change the rainfall intensity, runoff, and slope angle, and repeat the experiment to study the relationship between loess-structure erosion failure and corresponding hydrodynamic parameters and rainfall intensity, runoff, and slope angle.

[0012] In one embodiment, in step 6, after the occurrence of contact interface erosion and voiding, an endoscope and a supplementary light plate are used to record the slope morphology of the erosion pits, and a three-dimensional laser scanner is used to collect the changes in the amount of erosion of the erosion pits before and after rainfall in real time; and the collected data are subjected to correlation analysis of influencing factors to identify the key factors affecting the contact interface erosion and voiding.

[0013] In one embodiment, step 6 may also involve using a camera to take photos of the entire slope erosion process, combining these photos with the erosion failure mechanism of the drainage structure-soil contact interface, manually annotating them, and introducing real erosion failure photos as test data to train and verify the prediction model based on deep learning technology.

[0014] In one embodiment, a CNN-LSTM-MLP integrated framework can also be constructed. CNN extracts spatial features of the contact interface detachment image, and LSTM captures the temporal dynamics of erosion evolution. Simultaneously, MLP is combined to process normalized numerical factors such as slope and flow velocity to capture the correlation of the physical environment. The spatiotemporal image features and numerical features are vectorized and concatenated through a fusion layer to achieve classification and prediction of the erosion stage.

[0015] Compared with existing technologies, this invention, based on the real-world situation of burial erosion and voiding at the interface between drainage structures and soil on mountain slopes, proposes a testing system and method for burial erosion and voiding at this interface under heavy rainfall. This system can simulate the development stages of burial erosion and voiding at the interface between drainage structures and soil under the coupled effects of extreme rainfall and slope runoff. It acquires hydrodynamic parameters, sensor data, and image data during the burial development process, visualizes the burial erosion from different camera positions, and uses machine learning methods to perform correlation analysis on the measured data. Through deep learning, using the captured photographs as a dataset, a relevant deep learning model is trained to classify different development stages of burial erosion and evaluate the degree of damage. This testing system is simple to operate, economical, and convenient. Combined with machine learning and deep learning testing methods, it can comprehensively study the mechanism of burial erosion and voiding at the interface between drainage structures and soil under heavy rainfall. Attached Figure Description

[0016] Figure 1 Southeast isometric projection of a test device for testing the interface between a slope drainage structure and soil under heavy rain and the soil, showing the erosion and voiding disease.

[0017] Figure 2 This is a rear view of a test device for testing the erosion and voiding of the interface between a slope drainage structure and soil under heavy rain.

[0018] Figure 3 This is a schematic diagram of a transparent model box for testing a test device for the erosion and voiding disease at the interface between a slope drainage structure and the soil under heavy rain.

[0019] Figure 4 This is a schematic diagram of a rainfall device used to test the erosion and voiding at the interface between a slope drainage structure and the soil under heavy rain.

[0020] Figure 5 This is a schematic diagram of a slope confluence device, which is used to test the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain.

[0021] Figure 6 This is a schematic diagram of the lifting device for a test apparatus for the erosion and voiding disease at the interface between a slope drainage structure and soil under heavy rain.

[0022] Among them: 1-Water collection tank; 2-Transparent model box; 3-Faucet; 4-Flow meter; 5-Drain hole; 6-Brick structure drainage ditch; 7-Loess; 8-Rain spray head; 9-Metal frame base; 10-Slope overflow plate; 11-Slope water storage tank; 12-Water tank; 13-T-pipe; 14-Water pipe; 15-Camera; 16-Hydraulic jack; 17-Self-priming diaphragm pump. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] The present invention will now be described in further detail with reference to the accompanying drawings: like Figure 1 and Figure 2 As shown, the present invention provides a testing system for the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain. It mainly includes four functional parts: a multi-slope adjustable model box, a rainfall device, a slope runoff device, and a data acquisition unit.

[0026] Among them, reference Figure 3 The multi-slope adjustable model box uses a transparent model box 2, whose main body is a cuboid structure. It is tilted to simulate a slope, with an open top surface and an internal slope drainage structure model. The upper part of the tilt direction is defined as the upper side plate, the lower part as the lower side plate, and the part opposite the open top surface as the bottom plate. The other two are the left and right side plates, which are obviously parallel to the main runoff direction. The upper side plate of the transparent model box 2 has an inlet for the slope runoff device to apply slope runoff to the slope runoff device, while the lower side plate has an outlet for simulated rainfall and / or slope runoff to flow out of the transparent model box 2. It is connected to a water collection tank 1 to collect water for volume measurement.

[0027] The slope drainage structure model of this invention is used to simulate the actual state of the interface between the slope drainage structure and the soil. Figure 3In the structure shown, the slope drainage structure model includes a brick drainage channel 6 and loess 7. Along the runoff direction, the bottom and both sides of the brick drainage channel 6 are loess 7, with the sides forming the interface between the slope drainage structure and the soil. To test relevant data, this invention embeds pore pressure sensors, moisture content sensors, and soil pressure sensors at different locations and depths within the loess 7.

[0028] The main structure of the rain-making device can be referenced. Figure 4 As shown, it mainly includes several rain nozzles 8, which are located above the top surface of the transparent model box 2. Their function is to apply simulated rainfall to the slope drainage structure model.

[0029] The main structure of the slope runoff device can be referenced. Figure 5 As shown, the system includes a slope runoff collection and storage tank 11 and an inclined slope runoff overflow plate 10. The slope runoff collection and storage tank 11 should have a flow statistics function, and its function is to store water and supply water to the slope runoff overflow plate 10. The slope runoff overflow plate 10 is an inclined plate used to simulate slope runoff. The slope runoff collection and storage tank 11 supplies water to the inlet of the transparent model box 2 through the slope runoff overflow plate 10, thereby applying slope runoff to the slope drainage structure model.

[0030] The data acquisition unit of the present invention mainly includes a flow meter 4 and a camera 15, and further includes various sensors buried in the soil as described above. The flow meter 4 collects runoff data provided by the slope runoff device, and the camera 15 collects image data of the slope drainage structure model during the test process.

[0031] Based on the above structure, the present invention simultaneously introduces a rainfall device and a slope runoff device, which can simulate the confluence phenomenon under extreme rainfall. With the addition of a data acquisition unit, it can test the latent erosion and voiding disease at the interface between the slope drainage structure and the soil, and further explore its formation mechanism and main contributing factors.

[0032] In a further embodiment of the invention, a transparent model box 2 is placed on a metal base, the metal base is connected to a vertical rod, and the vertical rod is then connected to a top metal frame 9. The metal frame 9 is horizontally positioned, and pipes are laid on it. Rain nozzles 8 are connected to these pipes, pointing downwards. Figure 1 As shown. A water distribution pipe 3 is installed on the metal frame 9. The inlet end of the water distribution pipe 3 is connected to the water tank 12 through the water pipe 14, and the outlet end is connected to each pipeline. Figure 1 In this embodiment, the metal frame 9 has two parallel pipes, with the water distribution pipe 3 being a tee pipe. In this embodiment, the multi-slope adjustable model box and the rainfall device are integrated, and a self-priming diaphragm pump 17 can be installed on the water pipe 14.

[0033] In a further embodiment of the present invention, the transparent model box 2 can be made of acrylic, satisfying both transparency and load-bearing capacity requirements. A rectangular opening is cut into its upper side plate as the water inlet of the transparent model box 2, facilitating the lower edge of the slope overflow plate 10 to be placed against it, allowing slope runoff to be applied through the water inlet; dense drainage holes 5 are opened on the lower side plate of the transparent model box 2 as its water inlet, and the water collection tank 1 is placed at the drainage holes 5 to receive the drainage. For simplicity, the top surface of the water collection tank 1 is open, and the top surface is lower than the height of the lowest drainage hole 5. To prevent soil loss, the inner wall of the transparent model box 2 of the present invention is provided with permeable geotextile, allowing only water to pass through and not soil to pass through, restricting soil particles and draining water to the water collection trough 1 below.

[0034] In a further embodiment of the present invention, to achieve multi-slope adjustable transparent model box 2, the present invention provides a lifting and lowering device on its bottom surface, such as... Figure 6 As shown, the lifting device specifically selects hydraulic jack 16. The bottom side of the transparent model box 2 is placed on the hydraulic jack 16. The transparent model box 2 is tilted by the hydraulic jack 16 and its tilt angle is adjusted. For balance, two hydraulic jacks 16 are arranged at the bottom of the box.

[0035] In a further embodiment of the invention, the inlet end of the slope confluence water storage tank 11 is connected to a faucet 3, and a flow meter 4 is designed to monitor the flow rate on the connecting pipeline. The slope confluence overflow plate 10 is an inclined plate, with its upper edge resting against or connected to the edge of the slope confluence water storage tank 11, and its lower edge connecting to the inlet of the transparent model box 2. Alternatively, an elongated hole can be opened in the lower part of the water storage tank, and the upper edge of the slope confluence overflow plate 10 can rest against this elongated hole.

[0036] In a further embodiment of the invention, a suggested placement of the camera 15 is provided, which should be positioned above, below, and to the side of the transparent model box 2. The top camera position captures the top image of the slope drainage structure model, the side camera position captures images of the slope and contact interface, and the bottom camera position captures the bottom image of the slope drainage structure model.

[0037] Based on the above system, the testing method of the present invention includes the following main steps: Step 1: Adjust the slope angle of the built-in slope drainage structure model by adjusting the tilt angle of the transparent model box 2.

[0038] For example, the lifting height of one side of the transparent model box 2 can be adjusted using the hydraulic jack 16, thereby adjusting its tilt angle.

[0039] Step two: Install permeable geotextile on the inner wall of the lower side panel of the transparent model box 2, and embed pore pressure sensors and moisture content sensors in layers within the filled soil. Embed earth pressure sensors in the loess soil 7 below the brick structure drainage ditch 6 and on the side of the contact interface to collect moisture content, pore pressure, and earth pressure data in real time.

[0040] Step 3: Fill the loess layer by layer until the soil reaches the specified density, and set up an interception and drainage system or other engineering structures.

[0041] Step 4: Place the water inlet of the transparent model box 2 against the slope confluence overflow plate 10, adjust the angle of the slope confluence overflow plate 10, return the flow meter 4 to zero, and use a hose to connect it to the faucet 3 and the slope confluence water storage tank 11.

[0042] Step 5: Fill water tank 12 with water, turn on self-priming diaphragm pump 17, and use the rainmaking device to pre-rain the water. After the water flows in streams on the slope of the slope drainage structure model, stop the rainmaking and let it stand for 20-30 hours. In this embodiment, 24 hours is selected.

[0043] Step six involves defining runoff plots, measuring runoff flow rate at the slope bottom, determining flow velocity using potassium permanganate solution, and measuring runoff width using a steel ruler. Based on these three fundamental quantities, hydrodynamic parameters such as the Reynolds number (Re), Froude number (Fr), runoff shear force (τ), and runoff power (ω) are calculated. The runoff plots should have a complete top-to-bottom path of the slope drainage structure model and include the drainage structure-soil contact interface in the width direction.

[0044] Step 7: Use the rainmaking device to make rain again. After the rain forms a stream, turn on the tap 3 and apply a stable slope runoff from top to bottom through the slope runoff storage tank 11 and the slope runoff overflow plate 10.

[0045] Step 8: Use camera 15 positioned directly above to record the slope erosion process in real time, and use camera 15 positioned to the side to record the changes in the wetting peak within the slope.

[0046] Step 9: Based on the designated runoff zones, measure the slope runoff velocity again using potassium permanganate chemical reagent, measure the width of the runoff streams using a thin steel ruler, collect the runoff flow rate at the bottom using a graduated cylinder, and calculate the hydrodynamic parameters.

[0047] Step 10: Continue rainfall until the slope drainage structure model shows signs of erosion and voiding at the brick-concrete drainage structure-soil contact interface.

[0048] Step 11: Using an endoscope and a light source, record the slope morphology of the burrowing pits. Use a handheld 3D laser scanner to collect real-time data on the changes in erosion amount of the burrowing pits before and after rainfall.

[0049] Step 12: Change the rainfall intensity, slope runoff, and slope angle, and repeat the above experiment to study the relationship between loess-structure erosion failure and corresponding hydrodynamic parameters and rainfall intensity, runoff, and slope angle.

[0050] Step 13: Using photographs of the entire slope erosion process taken from the overhead camera position, and combining them with the loess-structure contact interface erosion failure mechanism, manual annotations are performed. A CNN convolutional neural network model is then trained, and real loess-structure erosion failure photographs taken during field investigations are used as test data to verify the predictive performance of the trained model. Relevant evaluation metrics are then used to evaluate the trained neural network model.

[0051] Step fourteen involves processing and analyzing the collected time-series and experimental data using impact factors. This process includes visualization analysis, correlation and multicollinearity analysis, relative importance analysis of impact factors, selection of the final impact factors, and construction of a dataset based on these factors. A pipeline is then built based on the constructed dataset, and data optimization is performed according to its numerical and categorical attributes. Numerical attributes include filling missing values, feature transformation, and feature scaling. Categorical attributes include filling missing values ​​and one-hot encoding. Based on this, models are selected and trained, including but not limited to LightGBM, SVM, CNN, and LSTM. Finally, evaluation metrics such as ROC_AUC and PR_AUC are used to evaluate the models.

[0052] Step 15: Construct a CNN-LSTM-MLP integrated framework. CNN extracts spatial features from the contact interface erosion image, while LSTM captures the temporal dynamics of erosion evolution. Simultaneously, MLP is used to process normalized numerical factors such as slope and flow velocity to capture the correlation with the physical environment. A fusion layer concatenates the spatiotemporal image features and numerical features into vectorized forms, achieving classification and prediction of erosion stages.

[0053] In summary, this invention, based on the phenomenon of burial erosion and voiding at the interface between drainage structures and soil on mountain slopes, proposes a testing system and method for this phenomenon under heavy rainfall. This system can simulate the development stages of burial erosion and voiding at the drainage structure-soil interface under extreme rainfall and slope runoff coupling. It acquires hydrodynamic parameters, sensor data, and image data during the burial development process, visualizes the burial erosion from different camera positions, and uses machine learning methods to perform correlation analysis on the measured data. Using deep learning methods, the captured photographs are used as a dataset to train relevant deep learning models to classify different development stages of burial erosion and evaluate the degree of damage. This testing system is simple to operate, economical, and convenient. Combined with machine learning and deep learning testing methods, it can comprehensively study the mechanism of burial erosion and voiding at the drainage structure-soil interface on slopes under heavy rainfall.

[0054] The above content is only for illustrating the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solution based on the technical concept proposed in this invention shall fall within the scope of protection of the claims of this invention.

Claims

1. A testing system for undercutting and voiding defects at the interface between a slope drainage structure and soil under heavy rain, characterized in that, It includes a multi-slope adjustable model box, a rainfall device, a slope runoff device, and a data acquisition unit; The multi-slope adjustable model box includes a transparent model box (2) with a built-in slope drainage structure model. The transparent model box (2) is an inclined box with an open top surface. Its upper side plate in the inclined direction has a water inlet, and its lower side plate has a water outlet connected to the water collection tank (1). The rainfall device includes several rainfall nozzles (8), which are located above the top surface of the transparent model box (2) and apply simulated rainfall to the slope drainage structure model; The slope runoff device includes a slope runoff storage tank (11) and an inclined slope runoff overflow plate (10). The slope runoff storage tank (11) supplies water to the inlet of the transparent model box (2) through the slope runoff overflow plate (10) and applies slope runoff to the slope drainage structure model. The data acquisition unit includes a flow meter (4) and a camera (15). The flow meter (4) acquires the flow data of the slope runoff device, and the camera (15) acquires the image data of the slope drainage structure model during the test process.

2. The testing system for the interface erosion and voiding of slope drainage structure and soil under heavy rain as described in claim 1, characterized in that, The slope drainage structure model includes a brick drainage ditch (6) and loess (7). Along the runoff direction, the bottom and both sides of the brick drainage ditch (6) are loess (7). A pore pressure sensor, a moisture content sensor and a soil pressure sensor are embedded in the loess (7).

3. The testing system for the interface erosion and voiding of slope drainage structure and soil under heavy rain as described in claim 1, characterized in that, The upper side plate of the transparent model box (2) is cut out into a rectangular opening as the water inlet, and the lower edge of the slope overflow plate (10) rests against the water inlet; the lower side plate of the transparent model box (2) has dense drainage holes (5) as the water inlet, and the top surface of the water collection tank (1) is open, and the top surface is lower than the height of the lowest drainage hole (5).

4. The testing system for the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain as described in claim 1, characterized in that, The inner wall of the transparent model box (2) is provided with permeable geotextile, and one side of its bottom surface is placed on a hydraulic jack (16). The transparent model box (2) is tilted and its tilt angle is adjusted by the hydraulic jack (16).

5. The testing system for the interface erosion and voiding of slope drainage structure and soil under heavy rain as described in claim 1, characterized in that, The inlet end of the slope confluence water storage tank (11) is connected to a faucet (3) and the flow meter (4) is installed on the connecting pipe. The slope confluence overflow plate (10) is a plate structure placed at an inclination. Its upper edge rests on or is connected to the edge of the slope confluence water storage tank (11), and its lower edge is connected to the inlet of the transparent model box (2).

6. The testing system for the erosion and voiding disease at the interface between the slope drainage structure and the soil under heavy rain as described in claim 1, characterized in that, The camera (15) is positioned above, below, and to the side of the transparent model box (2).

7. A test method for burial erosion and voiding at the interface between a slope drainage structure and soil under heavy rain, implemented based on the test system for burial erosion and voiding at the interface between a slope drainage structure and soil under heavy rain as described in any one of claims 1 to 6, characterized in that, The steps are as follows: Step 1: Activate the rainfall device to pre-rain the water. After the water flows in streams on the slope surface of the slope drainage structure model, stop the rainfall and let it stand for 20-30 hours. Step 2: Delineate runoff plots. Each runoff plot has a complete path from top to bottom of the slope drainage structure model and includes the drainage structure-soil contact interface in the width direction. Measure the runoff flow rate, runoff velocity, and runoff width of the runoff plots and calculate the hydrodynamic parameters. Step 3: Use the rainmaking device to make rain again until a stream of water is formed. Then, start the slope runoff device and apply stable slope runoff from top to bottom to the slope drainage structure model through the slope confluence overflow plate (10). Step 4: Use a camera (15) to record the slope erosion process and the changes in the wetting peak within the slope in real time; Step 5: Based on the runoff plots defined in Step 2, measure and calculate the hydrodynamic parameters again; Step 6: Continue rainfall until the contact interface of the slope drainage structure model shows signs of erosion and voiding. Step 7: Change the rainfall intensity, runoff, and slope angle, and repeat the experiment to study the relationship between loess-structure erosion failure and corresponding hydrodynamic parameters and rainfall intensity, runoff, and slope angle.

8. The test method for undercurrent erosion and voiding at the interface between the drainage structure and soil on a slope under heavy rain as described in claim 7, is characterized in that... In step 6, after the occurrence of contact interface erosion and voiding, an endoscope and a supplementary light plate are used to record the slope morphology of the erosion pits, and a three-dimensional laser scanner is used to collect the changes in the amount of erosion of the erosion pits before and after rainfall in real time; and the collected data are subjected to correlation analysis of influencing factors to identify the key factors affecting the contact interface erosion and voiding.

9. The test method for undercurrent erosion and voiding at the interface between the drainage structure and soil on a slope under heavy rain as described in claim 7 or 8, characterized in that, In step 6, a camera (15) is used to take photos of the entire process of slope erosion. Combined with the mechanism of latent erosion damage at the drainage structure-soil contact interface, the photos are manually labeled, and real erosion damage photos are introduced as test data to train and verify the prediction model based on deep learning technology.

10. The test method for undercurrent erosion and voiding at the interface between the drainage structure and soil on a slope under heavy rain as described in claim 9, characterized in that, A CNN-LSTM-MLP integrated framework is constructed. CNN extracts spatial features of the contact interface erosion image and LSTM captures the temporal dynamics of erosion evolution. Simultaneously, MLP is combined to process normalized numerical factors and capture the correlation with the physical environment. The spatiotemporal image features and numerical features are vectorized and concatenated through a fusion layer to achieve classification and prediction of the erosion stage.

Images