Detection device and method for underground semi-filled or non-filled karst cave

Abstract

The invention relates to the technical field of geological detection equipment, and discloses a detection device and method for an underground semi-filled or non-filled karst cave in order to solve the technical problems that a traditional karst cave detection method is low in detection precision, prone to interference, heavy and expensive in equipment and poor in visibility in a turbid water body. Comprising a steel casing, a rotary camera, a cable signal wire, a water hose, a transparent balloon and a portable terminal, wherein the steel casing penetrates through the earth surface and is communicated with a karst cave; the cable signal wire is connected with the rotary camera; the cable signal wire is sleeved with the water hose; through a multi-sensor fusion technology and in combination with scanning, pressure and temperature data cross validation, the actual form and the water flow direction recognition degree of semi-filled and non-filled karst caves are improved; the puncture-resistant transparent balloon is adapted to a narrow or inclined detection environment; the intelligent data analysis module based on a machine learning algorithm can automatically distinguish the karst cave and the crack, and image the true form of the karst cave and the water flow direction of the karst cave.

Description

Technical Field

[0001] The present invention relates to the technical field of geological exploration equipment, and in particular to a detection device and method for underground semi-filled or unfilled karst caves, which are applicable to fields such as karst areas, tunnel engineering, and mine exploration. Background Art

[0002] Traditional karst cave detection methods, such as ground penetrating radar, seismic wave method, and borehole sampling, have many limitations in practical applications. When detecting semi-filled karst caves with ground penetrating radar, due to the small dielectric constant difference between the filling material and the surrounding rock, the signal reflection is weak, which is prone to missed judgment. The seismic wave method is easily interfered by environmental vibrations, resulting in complex data analysis and reduced accuracy. In addition, existing detection equipment is usually bulky and costly, and it is difficult to be flexibly deployed under complex terrain conditions. Even when borehole sampling is used and a professional industrial camera is used for observation, it is difficult to overcome the turbid environment caused by the water flow inside the karst cave after the borehole is formed, resulting in extremely low visibility and inability to effectively obtain the situation inside the hole.

[0003] Therefore, there is an urgent need for a karst cave detection device with a simple structure, low cost, high sensitivity, and good visualization degree. Summary of the Invention

[0004] The purpose of the present invention is to provide a detection device and method for underground semi-filled or unfilled karst caves, so as to solve the technical problems of low detection accuracy, susceptibility to interference, bulky and expensive equipment, and poor visibility in turbid water bodies existing in traditional karst cave detection methods.

[0005] To achieve the above purpose, as one aspect of the present invention, there is provided a detection device for underground semi-filled or unfilled karst caves, including a steel casing provided with a connection through the ground surface and the karst cave, a rotating camera for underwater use, a cable signal line for connecting the rotating camera, a flexible water pipe sleeved outside the cable signal line, a transparent balloon wrapping the rotating camera, and a portable terminal connected to the cable signal line; the transparent balloon is communicated with the flexible water pipe and is used to expand according to the water injection volume; the portable terminal is used to receive data and generate a three-dimensional karst cave model in real time, and display the water flow direction of the karst cave; the rotating camera and the transparent balloon are located inside the karst cave during use.

[0006] The underground semi-filled or unfilled karst cave detection device provided by the present invention connects the ground surface and the karst cave through a steel casing, and uses an inflatable transparent balloon to wrap the rotating camera, enabling it to move and detect flexibly in narrow and inclined karst cave spaces, and overcoming the problem of poor visibility in turbid water bodies in traditional methods. At the same time, the portable terminal generates a three-dimensional karst cave model in real time and displays the water flow direction, providing an intuitive and efficient means for karst cave exploration and rescue, and effectively solving the technical problems of low accuracy, susceptibility to interference, bulky and expensive equipment, etc. in traditional karst cave detection methods.

[0007] Furthermore, the cable signal line is provided with meter scales that facilitate obtaining the depth of the downhole exploration.

[0008] During the downhole exploration process, the operator can directly and accurately understand the depth position of the rotating camera by observing the meter scales on the cable signal line, without the need for additional depth measurement equipment, which simplifies the operation process and improves work efficiency.

[0009] Furthermore, the rotating camera includes a high-precision scanning imaging sensor with 3D scanning function, a pressure sensor for analyzing the water flow direction in the karst cave, and an infrared thermal imager for identifying the temperature difference between the filling material and the karst cave.

[0010] Integrating 3D scanning, pressure sensing, and infrared thermal imaging functions, it can comprehensively and multi-dimensionally obtain the internal information of the karst cave. 3D scanning constructs an accurate karst cave structure model; the pressure sensor analyzes the water flow direction to judge potential dangerous areas; the infrared thermal imager identifies the temperature difference between the filling material and the karst cave to assist in judging the stability of the karst cave and potential cavities, thus significantly improving the accuracy and reliability of the detection.

[0011] Furthermore, the rotating camera includes a signal processing unit, and the signal processing unit includes an A / D converter and a filter circuit module.

[0012] Integrating the signal processing unit inside the rotating camera can perform digital conversion and noise filtering on the analog signals collected by the sensors, effectively improving the signal quality and signal-to-noise ratio, thereby enhancing the accuracy and reliability of data processing.

[0013] Furthermore, the material of the transparent balloon is a rubber material with puncture resistance and high strength.

[0014] Using a rubber material with puncture resistance and high strength as the material of the transparent balloon can effectively prevent the balloon from being scratched or punctured by sharp stones or other obstacles inside the karst cave, ensuring the continuity and stability of the detection process. The high-strength material can also withstand the water pressure and external pressure inside the karst cave, avoiding deformation or rupture of the balloon, thus guaranteeing the safe operation of the rotating camera and related sensors, extending the service life of the detection device, and reducing the risk of detection interruption caused by equipment damage.

[0015] As another aspect of the present invention, a detection method for semi-filled or unfilled underground karst caves is provided, including the following steps:

[0016] Step S100, synchronously collect data through multiple sensors, and the data includes:

[0017] Internal images of the karst cave and 3D scanning point cloud data captured by the rotating camera;

[0018] The water flow direction pressure value recorded by the pressure sensor;

[0019] The temperature difference distribution data obtained by the infrared thermal imager;

[0020] Step S200: Perform spatial alignment and denoising processing on the data to construct a unified three-dimensional coordinate system;

[0021] Step S300: Extract multi-modal features and perform weighted fusion to generate a comprehensive feature vector;

[0022] Step S400: Input the feature vector into the machine learning model to output the karst cave boundary probability map and the water flow direction vector field;

[0023] Step S500: Based on the probability map and the point cloud data, generate a three-dimensional karst cave model and perform real-time visualization display.

[0024] The underground semi-filled or unfilled karst cave detection method provided by the present invention synchronously collects images, 3D point clouds, pressure, and temperature difference data through multiple sensors, and performs spatial alignment and denoising processing, effectively improving the data quality and reliability. Through multi-modal feature fusion and machine learning model analysis, it can accurately identify the karst cave boundary and the water flow direction, overcoming the problems of dependence on single-sensor data and insufficient recognition accuracy in traditional methods. Finally, the three-dimensional karst cave model generated based on the probability map and the point cloud data realizes the real-time visualization display of the internal structure of the karst cave.

[0025] Furthermore, the feature extraction in step S300 includes:

[0026] Visual feature extraction: Extract dissolution fissure edges, rock textures, etc. from the images;

[0027] Pressure feature extraction: Calculate the water flow direction vector field and mark the high pressure gradient regions;

[0028] Thermal feature extraction: Identify the regions with significant temperature differences as potential karst cave boundaries;

[0029] The feature fusion in step S300 includes:

[0030] The weight of the visual feature is 0.6, the weight of the pressure feature is 0.3, and the weight of the thermal feature is 0.1;

[0031] The visual features include dissolution fissure edges and rock textures;

[0032] The pressure features include the water flow direction vector field;

[0033] The thermal features include the region marked with a temperature difference ≥ 2°C.

[0034] Furthermore, the machine learning model includes a convolutional neural network for classifying the karst cave boundary and a random forest regression model for predicting the water flow direction. Its input features include pressure values and the direction of temperature difference gradient.

[0035] Furthermore, the spatial alignment in step S200 includes:

[0036] Determining the camera depth through the meter scale of the cable signal line;

[0037] Using a gyroscope to correct the camera attitude and construct a three-dimensional coordinate system;

[0038] Adopting robust principal component analysis to remove image noise and eliminate thermal imaging artifacts.

[0039] Furthermore, the 3D modeling in step S500 includes:

[0040] Converting the point cloud data into a mesh model using a fast polygon generation algorithm;

[0041] Labeling the karst cave boundary and fracture area according to the probability map output in step S400.

[0042] A detection device and method for underground semi-filled or unfilled karst caves provided by the present invention have the following advantages:

[0043] By encapsulating an underwater high-definition rotating camera in a puncture-resistant transparent balloon and using a water pipe for control and data transmission, the detection of karst caves in narrow or inclined environments is realized. Through multi-sensor fusion technology, scanning, pressure, and temperature data are integrated, and the recognition accuracy of the actual shape and water flow direction of semi-filled / unfilled karst caves is significantly improved through cross-validation. At the same time, the system is equipped with an intelligent data analysis module based on machine learning algorithms, which can automatically distinguish karst caves from fractures and construct a true 3D morphology map of the karst caves based on this, accurately presenting the internal structure of the karst caves and the water flow direction, providing reliable data support for underground engineering construction. Description of the Drawings

[0044] Figure 1 It is a schematic layout diagram of the detection device provided by the present invention.

[0045] In the figure: 10, portable terminal; 20, steel casing; 30, flexible water pipe; 40, cable signal line; 50, rotating camera; 60, karst cave; 70, transparent balloon. Detailed Embodiments

[0046] In order to make the purpose, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0047] Refer to Figure 1 , the present invention provides a detection device for underground semi-filled or unfilled karst caves, including a steel casing 20 provided with a through-hole communicating the ground surface with the karst cave 60, a rotating camera 50 for underwater use, a cable signal line 40 for connecting the rotating camera 50, a flexible water pipe 30 sleeved outside the cable signal line 40, a transparent balloon 70 wrapping the rotating camera 50, and a portable terminal 10 connected to the cable signal line 40;

[0048] The steel casing 20 is used for supporting the hole wall to prevent cave-ins, and the steel casing 20 needs to penetrate through the rock layer and extend into the top of the karst cave 60; the flexible water pipe 30 extends from the steel casing 20 into the interior of the karst cave 60 and can inject and pump water; the transparent balloon 70 is communicated with the flexible water pipe 30 and is used for expanding according to the water injection volume; the portable terminal 10 is used for receiving data and generating a three-dimensional model of the karst cave 60 in real time to display the water flow direction of the karst cave 60; the rotating camera 50 and the transparent balloon 70 are located in the karst cave 60 during use.

[0049] Further, the cable signal line 40 is provided with a meter scale for facilitating the acquisition of the depth of descent.

[0050] Further, the rotating camera 50 includes a high-precision scanning imaging sensor with 3D scanning function, a pressure sensor for analyzing the water flow direction in the karst cave 60, and an infrared thermal imager for identifying the temperature difference between the filling material and the karst cave 60. It has high-definition picture quality, is equipped with a high-resolution lens, supports 4K / 8K shooting, captures underwater details (such as dissolution fissures, rock textures, etc.), has a 3D scanning function, can adjust the detection angle, and can take pictures in multiple directions;

[0051] Further, the rotating camera 50 includes a signal processing unit, and the signal processing unit includes an A / D converter and a filter circuit module.

[0052] Further, the material of the transparent balloon 70 is a rubber material with puncture resistance and high strength, having higher tensile strength and toughness, enabling it to withstand greater external forces without breaking, and being environmentally friendly and safe.

[0053] The present invention also provides a detection method for underground semi-filled or unfilled karst caves, including the following steps:

[0054] Step S100, synchronously collect data through multiple sensors, and the data includes:

[0055] Internal images of the karst cave 60 and 3D scanning point cloud data taken by the rotating camera 50;

[0056] Water flow direction pressure values recorded by the pressure sensor;

[0057] Temperature difference distribution data obtained by the infrared thermal imager;

[0058] Step S200: Align the data spatially and denoise it to construct a unified three-dimensional coordinate system;

[0059] Step S300: Extract multi-modal features and fuse them with weights to generate a comprehensive feature vector;

[0060] Step S400: Input the feature vector into a machine learning model to output the boundary probability map of the karst cave 60 and the water flow direction vector field;

[0061] Step S500: Generate a three-dimensional karst cave 60 model based on the probability map and point cloud data and display it in real-time visually.

[0062] Furthermore, the feature extraction in step S300 includes:

[0063] Visual feature extraction: Extract the edges of dissolution fissures, rock textures, etc. from the images;

[0064] Pressure feature extraction: Calculate the water flow direction vector field and mark the high pressure gradient areas;

[0065] Thermal feature extraction: Identify the areas with significant temperature differences as potential boundaries of the karst cave 60;

[0066] The feature fusion in step S300 includes:

[0067] The weight of the visual feature is 0.6, the weight of the pressure feature is 0.3, and the weight of the thermal feature is 0.1;

[0068] The visual features include the edges of dissolution fissures and rock textures;

[0069] The pressure features include the water flow direction vector field;

[0070] The thermal features include the area markings with a temperature difference ≥ 2°C.

[0071] Furthermore, the machine learning model includes a convolutional neural network for classifying the boundaries of the karst cave 60 and a random forest regression model for predicting the water flow direction. Its input features include the pressure value and the temperature difference gradient direction, and the refresh frequency of the water flow direction prediction is once every 5 seconds.

[0072] Furthermore, the spatial alignment in step S200 includes:

[0073] Determine the camera depth through the meter scale of the cable signal line 40;

[0074] Use the gyroscope to correct the camera attitude and construct a three-dimensional coordinate system;

[0075] Adopt robust principal component analysis to remove image noise and eliminate thermal imaging artifacts.

[0076] Further, the 3D modeling in step S500 includes:

[0077] Converting the point cloud data into a mesh model using a fast polygon generation algorithm;

[0078] Marking the boundary of the karst cave 60 and the fracture area according to the probability map output in step S400.

[0079] Working principle

[0080] System composition: A puncture-resistant transparent balloon 70 encapsulates an underwater high-definition rotating camera 50 inside. The balloon is connected to a portable terminal 10 on the ground through a flexible water pipe 30.

[0081] Detection process:

[0082] Slowly insert the balloon and the camera into the entrance of the target karst cave 60.

[0083] Inject clear water into the balloon through the water pipe, causing the balloon to gradually expand to a diameter of about 2 meters. During the expansion process, the balloon can effectively displace the turbid water body and closely adhere to the rock contour of the karst cave 60 wall.

[0084] The expanded balloon covers the karst cave 60 space within a range of about 2 meters, and the high-definition camera captures the detailed information inside the karst cave 60 in all directions, including dissolution fissures, rock textures, karst cave 60 structures, etc.

[0085] At the same time, the pressure sensor carried by the system monitors the different pressures in the water flow direction to infer the general direction of the water flow in the karst cave 60.

[0086] Combined with the pre-set 2x2 meter drilling spacing data, the system can construct the detailed contour and boundary information of the karst cave 60.

[0087] Data transmission and recovery:

[0088] After the detection is completed, drain the clear water in the balloon through the flexible water pipe 30 to make the balloon shrink.

[0089] The camera is retracted to the ground together with the balloon.

[0090] The data captured by the camera is processed and then transmitted to the portable terminal 10 through the cable signal line for display and storage for subsequent analysis and research.

[0091] It can be seen that this system utilizes the characteristics of the expandable balloon to achieve non-destructive detection of karst caves in the underwater environment, and combined with a high-definition camera and multiple sensors, it can effectively obtain the morphology, structure, and hydrological information of the karst cave, providing an important basis for underground engineering construction.

[0092] The detection device and method for underground semi-filled or unfilled karst caves provided by the present invention have the following advantages:

[0093] Through multi-sensor fusion technology, combined with cross-verification of scanning, pressure, and temperature data, the recognition of the actual shape and water flow direction of semi-filled and unfilled karst caves is improved; through puncture-resistant transparent balloons, it adapts to narrow or inclined detection environments; an intelligent data analysis module based on machine learning algorithms can automatically distinguish karst caves from fractures and image the true shape of karst caves and the water flow direction of karst caves.

[0094] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the present invention shall be included in the protection scope of the present invention.

Claims

1. A detection device for underground semi-filled or unfilled karst caves, characterized in that, It includes a steel casing (20) which is arranged to penetrate the ground surface and communicate with a karst cave (60) and is used to support the side hole walls, a rotating camera (50) for underwater use, a cable signal line (40) for connecting the rotating camera (50), a flexible water pipe (30) sleeved outside the cable signal line (40), a transparent balloon (70) wrapping the rotating camera (50), and a portable terminal (10) connected to the cable signal line (40); The transparent balloon (70) is communicated with the flexible water pipe (30) and is used to expand according to the water injection volume; The portable terminal (10) is used to receive data and generate a three-dimensional model of the karst cave (60) in real time and display the water flow direction of the karst cave (60); The rotating camera (50) and the transparent balloon (70) are located in the karst cave (60) during use.

2. The detection device for underground semi-filled or unfilled karst caves according to claim 1, wherein, The cable signal line (40) is provided with a meter scale for facilitating the acquisition of the depth of the downward exploration.

3. The detection device for underground semi-filled or unfilled karst caves according to claim 2, wherein, The rotating camera (50) includes a high-precision scanning imaging sensor with 3D scanning function, a pressure sensor for analyzing the water flow direction in the karst cave (60), and an infrared thermal imager for identifying the temperature difference between the filling material and the karst cave (60).

4. The detection device for underground semi-filled or unfilled karst caves according to claim 3, characterized in that, The rotating camera (50) includes a signal processing unit, and the signal processing unit includes an A / D converter and a filter circuit module.

5. The detection device for underground semi-filled or unfilled karst caves according to claim 1, characterized in that The material of the transparent balloon (70) is a rubber material with puncture resistance and high strength.

6. A detection method for underground semi-filled or unfilled karst caves, characterized in that, It includes the following steps: Step S100, synchronously collect data through multiple sensors, and the data includes: Internal images of the karst cave (60) and 3D scanning point cloud data taken by the rotating camera (50); Water flow direction pressure values recorded by the pressure sensor; Temperature difference distribution data obtained by the infrared thermal imager; Step S200, perform spatial alignment and denoising processing on the data to construct a unified three-dimensional coordinate system; Step S300, extract multi-modal features and perform weighted fusion to generate a comprehensive feature vector; Step S400, input the feature vector into a machine learning model, and output a karst cave (60) boundary probability map and a water flow direction vector field; Step S500, generate a three-dimensional model of the karst cave (60) based on the probability map and the point cloud data and perform real-time visualization display.

7. The detection method for underground semi-filled or unfilled karst caves according to claim 1, characterized in that The feature extraction in step S300 includes: Visual feature extraction, extracting dissolution fissure edges and rock textures from the image; Pressure feature extraction, calculating the water flow direction vector field and marking high pressure gradient regions; Thermal feature extraction, identifying regions with significant temperature differences as potential karst cave (60) boundaries; The feature fusion in step S300 includes: The weight of the visual feature is 0.6, the weight of the pressure feature is 0.3, and the weight of the thermal feature is 0.1; The visual features include dissolution fissure edges and rock textures; The pressure features include the water flow direction vector field; The thermal features include area markings with a temperature difference ≥ 2°C.

8. The detection method for underground semi-filled or unfilled karst caves according to claim 6, characterized in that, The machine learning model includes a convolutional neural network for karst cave (60) boundary classification and a random forest regression model for water flow direction prediction, and its input features include pressure values and temperature difference gradient directions.

9. The detection method for underground semi-filled or unfilled karst caves according to claim 6, characterized in that, The spatial alignment in step S200 includes: Determining the depth of the rotating camera (50) through the meter scale of the cable signal line (40); Use a gyroscope to correct the camera attitude and construct a three-dimensional coordinate system; Adopt robust principal component analysis to remove image noise and eliminate thermal imaging artifacts.

10. The detection method for underground semi-filled or unfilled karst caves according to claim 6, characterized in that, The three-dimensional modeling in step S500 includes: Use a fast polygon generation algorithm to convert point cloud data into a mesh model; According to the probability map output in step S400, label the boundaries of the karst cave (60) and the fracture area.

Images