Tunnel water and mud burst detection method and system and medium

Abstract

The invention provides a tunnel water and mud burst detection method and system and a medium, and relates to the technical field of tunnels and underground engineering, and the method comprises the steps: obtaining a video image corresponding to a target lens; preprocessing the video image to obtain a target video image; extracting a key frame based on the target video image; and according to the key frame and a target detection model, determining whether water burst and mud burst occur in the tunnel. Thus, the water burst and mud burst phenomena in the tunnel are automatically and intelligently recognized by combining deep learning and image processing, the accuracy and effectiveness of water burst and mud burst phenomenon detection are improved, meanwhile, the efficiency and real-time performance of water burst and mud burst phenomenon detection are improved, and the manpower maintenance cost and manpower input are reduced. Meanwhile, the requirement for detecting the complex geological environment is met, and the detection result is not prone to being affected by the surrounding environment.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnels and underground engineering, and in particular to a method, system and medium for detecting water gushing and mud bursting in tunnels. Background Art

[0002] Railway tunnel construction routes must pass through complex geological environments such as high water pressure and high water temperature. Therefore, major disasters such as tunnel water gushing and mud bursts are very likely to occur during tunnel excavation, posing huge risks and challenges to construction safety.

[0003] Currently, there are two methods for detecting water and mud inrush: one based on manual inspection and the other on comprehensive monitoring using multiple sensors deployed at the construction site. The former suffers from poor real-time performance and low efficiency. Due to the complex tunnel construction environment, sensors may be affected by environmental interference, and the deployment of multiple sensors can also affect construction efficiency.

[0004] In response to the problems of the prior art, the present invention provides a method, system and medium for detecting water gushing and mud bursting in a tunnel. Summary of the Invention

[0005] In response to the problems of the current existing technology, the present invention provides a method, system and medium for detecting water and mud inrush in a tunnel, the method comprising:

[0006] Obtain the video image corresponding to the target lens;

[0007] Preprocessing the video image to obtain a target video image;

[0008] Extracting key frames based on the target video image;

[0009] According to the key frames and the target detection model, it is determined whether water gushing or mud bursting occurs in the tunnel.

[0010] According to one embodiment of the present invention, the target video image is obtained by the following steps:

[0011] Performing grayscale transformation on the video image to obtain a grayscale image;

[0012] Performing noise reduction processing on the grayscale image to obtain a noise-reduced image;

[0013] performing contrast enhancement and brightness enhancement on the noise-reduced image to obtain a processed image;

[0014] Perform edge detection and sharpening processing on the processed image to obtain the target video image.

[0015] According to one embodiment of the present invention, the key frame is obtained by the following steps:

[0016] Determining a motion vector of the target video image;

[0017] Determine the area where the motion vector is greater than a target threshold as the video feature image;

[0018] A differential image of the video feature image is determined, and the key frame is extracted according to the differential image.

[0019] According to one embodiment of the present invention, whether water inrush and mud burst occurs in the tunnel is determined by the following steps:

[0020] Determining the location of the water gushing point, the location of the mud, the curvature characteristics of the water gushing column, and the water flow rate level based on the key frame and the target detection model;

[0021] Whether water gushing and mud bursting occurs in the tunnel is determined according to the location of the water gushing point, the location of the mud, the curvature characteristics and the water flow level.

[0022] According to one embodiment of the present invention, the target detection model includes a convolutional neural network;

[0023] The position of the water inflow point and the position of the mud are determined by the following steps:

[0024] Extracting features from the key frames using the convolutional neural network to obtain water gushing features and mud features;

[0025] The water gushing feature and the mud feature are input into the target detection model to generate the position of the water gushing point and the position of the mud.

[0026] According to one embodiment of the present invention, the arc characteristic and the water flow level are determined by the following steps:

[0027] Performing pixel-level semantic segmentation on the key frame using the target detection model to obtain the curvature feature of the water column;

[0028] The key frame is detected by the target detection model to obtain the water flow level of the water column.

[0029] According to one embodiment of the present invention, whether water inrush and mud burst occurs in the tunnel is determined by the following steps:

[0030] Performing multimodal data fusion on the radian feature and the water flow level to obtain the fused feature;

[0031] generating a risk level according to the location of the water inflow point, the location of the mud, and the fusion characteristics;

[0032] If the risk level is a disaster level, it is determined that water gushing and mud bursting occur in the tunnel.

[0033] According to one embodiment of the present invention, the method further includes:

[0034] When water or mud burst occurs in the tunnel, a danger alarm is issued.

[0035] According to another aspect of the present invention, a storage medium is provided, which includes a series of instructions for executing the method steps as described in any one of the above.

[0036] According to another aspect of the present invention, there is also provided a tunnel water and mud burst detection system, which performs the method as described above, and the system comprises: a video acquisition module, a video image processing unit, a key frame extraction module and a target detection module connected in sequence;

[0037] The video acquisition module is used to obtain the video image corresponding to the target lens;

[0038] The video image processing unit is used to pre-process the video image to obtain a target video image;

[0039] The key frame extraction module is used to extract key frames based on the target video image;

[0040] The target detection module is used to determine whether water gushing or mud bursting occurs in the tunnel based on the key frames and the target detection model.

[0041] The present invention provides a method, system, and medium for detecting water and mud inrush in tunnels, which have the following advantages over existing technologies:

[0042] The present invention first acquires video images, then preprocesses them, and then extracts key frames from the target video images. Then, based on the key frames and the target detection model, it determines whether water and mud bursts have occurred in the tunnel. In this way, by combining deep learning and image processing, water and mud bursts in tunnels can be automatically and intelligently identified. This improves the accuracy and effectiveness of water and mud burst detection while also improving the efficiency and real-time performance of water and mud burst detection, reducing labor maintenance costs and input. Furthermore, it meets the needs of detecting complex geological environments, and the detection results are not easily affected by the surrounding environment.

[0043] Other features and advantages of the present invention will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present invention. The purposes and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0045] Figure 1 A flow chart showing a method for detecting water and mud inrush in a tunnel according to an embodiment of the present invention is shown;

[0046] Figure 2 A block diagram of a tunnel water and mud burst detection system according to an embodiment of the present invention is shown;

[0047] Figure 3 A flow chart of a method for detecting water and mud inrush in a tunnel according to yet another embodiment of the present invention is shown.

[0048] In the accompanying drawings, the same reference numerals are used for the same parts. In addition, the accompanying drawings are not drawn according to the actual scale. DETAILED DESCRIPTION

[0049] To make the objectives, technical solutions and advantages of the present invention more clear, embodiments of the present invention are described in further detail below with reference to the accompanying drawings.

[0050] Prior art (CN118379663A) proposes a method and system for predicting tunnel water inrush dynamics based on intelligent visual monitoring. This system includes a visual intelligent monitoring component, a water inrush visual identification component, and a cloud platform. The visual intelligent monitoring component, when installed facing the tunnel construction face, collects video monitoring signals from the tunnel construction face. The video monitoring signals from the visual intelligent monitoring component are uploaded to the cloud platform via a wireless transmission system and processed by a cooperating image processing device to obtain a digitized signal. The water inrush visual identification component further processes the digitized signal to identify the spatial location and water volume of the water inrush.

[0051] This method mainly constructs a video detection method. The water gushing amount detection method adopts the method of substituting the detected water gushing area into the empirical formula for calculation, and only gives a general description of the mentioned algorithm. It is difficult to cope with the complex conditions and diverse water gushing characteristics of complex construction sites, and the intelligence level of the detection process is not high.

[0052] The existing technology (CN116778678A) proposes a method, device and system for unmanned monitoring and early warning of sudden water inrush using machine vision. It includes calculating the optical flow vector of each pixel point in the tunnel face image; determining the position of the sudden water inrush point, the number of sudden water inrush points and the optical flow vector at the sudden water inrush point based on the optical flow vector of each pixel point; judging whether the current sudden water inrush point has disappeared based on the change of the optical flow vector at the sudden water inrush point; judging whether there is a sudden water inrush based on the optical flow vector at the sudden water inrush point, and determining the optical flow vector at the sudden water inrush point; determining the amount of sudden water inrush based on the optical flow vector at the sudden water inrush point, and judging whether to issue an early warning based on the amount of sudden water inrush. By utilizing the motion characteristics of the sudden water inrush flow, the optical flow method is used to monitor the scope of sudden water inrush in the tunnel, realizing real-time monitoring of sudden water inrush in the tunnel.

[0053] This method mainly uses the information represented by the optical flow vector of each pixel point to detect the water gushing point and water volume. It may become inaccurate due to factors such as the weak light in the construction environment, the interference of construction smoke and dust, and the rich diversity of the gushing water column shape.

[0054] In view of the above-mentioned defects of the prior art, the present invention proposes a method, system and medium for detecting water inrush and mud burst in tunnels. Figure 1 A flow chart of a method for detecting water and mud inrush in a tunnel according to an embodiment of the present invention is shown, the method comprising:

[0055] S101, obtaining a video image corresponding to a target lens;

[0056] S102, pre-processing the video image to obtain a target video image;

[0057] S103, extracting key frames based on the target video image;

[0058] S104: Determine whether water or mud inrush occurs in the tunnel based on the key frames and the target detection model.

[0059] The target lens may be a high-definition visible light camera to provide all-weather real-time monitoring of water and mud bursts in the tunnel. The target detection model may be a YOLO model, such as YOLOv10.

[0060] The present invention first acquires video images, then preprocesses them, and then extracts key frames from the target video images. Then, based on the key frames and the target detection model, it determines whether water and mud bursts have occurred in the tunnel. In this way, by combining deep learning and image processing, water and mud bursts in tunnels can be automatically and intelligently identified. This improves the accuracy and effectiveness of water and mud burst detection while also improving the efficiency and real-time performance of water and mud burst detection, reducing labor maintenance costs and input. Furthermore, it meets the needs of detecting complex geological environments, and the detection results are not easily affected by the surrounding environment.

[0061] In a possible embodiment, the target video image is obtained by the following steps:

[0062] Perform grayscale transformation on the video image to obtain a grayscale image;

[0063] Performing noise reduction processing on the grayscale image to obtain a noise-reduced image;

[0064] Perform contrast enhancement and brightness enhancement on the denoised image to obtain a processed image;

[0065] Perform edge detection and sharpening on the processed image to obtain the target video image.

[0066] The video image can be converted into a grayscale image through binarization, reducing the video image data volume while highlighting the characteristics of the water and mud burst. The grayscale image is then subjected to techniques such as median filtering and bilateral filtering to remove noise, ensuring the clarity of the de-noised image. The de-noised image is then enhanced in contrast and brightness through techniques such as histogram equalization and gamma correction, improving the visibility of the water and mud burst in the processed image. Finally, edge detection and sharpening algorithms such as Sobel and Laplacian are used to highlight the edges of the water and mud burst, ensuring that the details of the water and mud burst are clearly displayed.

[0067] In this way, by preprocessing the video image, the target video image is obtained, which reduces the amount of video image data while improving the clarity, contrast and feature visibility of the target video image, providing a basis for the generation of key frames and improving the accuracy and efficiency of water gushing and mud bursting phenomena.

[0068] In a possible embodiment, the key frame is obtained by the following steps:

[0069] determining a motion vector of a target video image;

[0070] The area where the motion vector is greater than the target threshold is determined as the video feature image;

[0071] A differential image of the video feature image is determined, and key frames are extracted based on the differential image.

[0072] For example, the target threshold may be determined according to actual application scenarios, which is not limited in the present invention.

[0073] Among them, the motion vector calculation can be performed on each frame of video image to identify the pixel displacement between two adjacent frames of video image. Then, the motion vector can be compared with the target threshold, and the (pixel) area greater than the target threshold can be screened out. The area is determined as the video feature image, that is, the key frame where water gushing and mud bursting may occur. Afterwards, the differential image technology can be used to calculate the differential image between the video feature images to calculate the pixel difference between the video feature images, and then confirm the time point when the water gushing and mud bursting event occurred. Finally, the image with the largest absolute value of the pixel difference in the differential image is determined as the key frame, that is, the video image corresponding to the moment when the water gushing and mud bursting phenomenon is most significant.

[0074] In this way, motion vectors and differential images are calculated based on the target video image to determine the key frame, which improves the accuracy of the key frame and provides a basis for detecting whether water gushing and mud bursting occur in the tunnel.

[0075] In a possible embodiment, whether a water inrush and mud burst occurs in a tunnel is determined by the following steps:

[0076] Determine the location of the water inflow point, the location of the mud, the curvature characteristics of the water column, and the water flow rate level based on the key frames and target detection model;

[0077] Determine whether water gushing and mud bursting will occur in the tunnel based on the location of the water gushing point, the location of the mud, the curvature characteristics and the water flow level.

[0078] Among them, the key frames can be input into the target detection model to output the location of the water gushing point, the location of the mud, the curvature characteristics of the water gushing column and the water flow level, thereby determining whether water gushing and mud bursting occur in the tunnel.

[0079] Furthermore, the target detection model includes both a model test database and a field acquisition database. The target detection model is trained using a large number of data samples from the model test database. Subsequently, the model is validated against the field acquisition database, resulting in strong generalization and robustness. Furthermore, the model automatically learns the characteristics of target video images, eliminating the need for manual design of complex rules or parameters, significantly reducing the cost and error of manual intervention.

[0080] In one possible embodiment, the target detection model includes a convolutional neural network;

[0081] Determine the location of the water inflow and the location of the mud by the following steps:

[0082] The convolutional neural network is used to extract the features of key frames to obtain the water inrush features and mud features;

[0083] The water gushing characteristics and mud characteristics are input into the target detection model to generate the location of the water gushing point and the location of the mud.

[0084] First, a convolutional neural network (CNN) can be used to extract features from keyframes to obtain water and mud features. Based on these features and a target detection model, water and mud targets in the target video image can be identified, their locations located, and corresponding contour boxes generated. Furthermore, the target detection model can be used to further segment the water and mud points at the pixel level, generating precise contours and region boundaries.

[0085] In this way, based on the key frames and target detection model, the locations of water gushing points and mud can be obtained, providing a basis for determining whether water gushing and mud bursting occur in the tunnel.

[0086] In one possible embodiment, the arc characteristics and water flow level are determined by the following steps:

[0087] The key frames are segmented at the pixel level using the target detection model to obtain the curvature characteristics of the water column.

[0088] The key frames are detected through the target detection model to obtain the water flow level of the water column.

[0089] The object detection model can perform pixel-level semantic segmentation on keyframes, deriving the curvature characteristics of the water column. This accurately depicts the shape and boundaries of the water column, aiding in further analysis of the water flow trajectory. Furthermore, the object detection model can detect the flow rate of the water column and classify it into three levels: low flow, medium flow, and high flow. Furthermore, the object detection model can also detect the flow velocity of the water column.

[0090] In this way, based on the key frame and target detection model, the curvature characteristics and water flow level of the water column can be obtained, which improves the accuracy of the water flow rate and provides a basis for determining whether water flow and mud burst occur in the tunnel.

[0091] In a possible embodiment, whether a water inrush and mud burst occurs in a tunnel is determined by the following steps:

[0092] Perform multimodal data fusion on the arc characteristics and water flow level to obtain fusion features;

[0093] Generate risk levels based on the location of water inflow points, mud location and fusion characteristics;

[0094] If the risk level is disaster level, it is determined that water gushing and mud bursting will occur in the tunnel.

[0095] Multimodal data fusion of arc characteristics and water flow levels can more accurately assess the risk level of water and mud inrush, further improving the accuracy and timeliness of early warnings. This information fusion technology ensures not only the detection of initial water seepage but also more accurate predictions of the severity and evolution of water inrush.

[0096] Afterwards, the fusion features, the location of the water gushing point, and the location of the mud can be input into the corresponding table to generate a risk level. If the risk level is a disaster level, it is determined that water gushing and mud bursting have occurred in the tunnel; if the risk level is a non-disaster level, it is determined that water gushing and mud bursting have not occurred in the tunnel.

[0097] In this way, based on the curvature characteristics, water flow level, location of water gushing point, and location of mud, it is determined whether water gushing and mud bursting occur in the tunnel, thereby improving the accuracy and real-time performance of the detection results.

[0098] In a possible embodiment, the method further includes:

[0099] When water or mud bursts occur in the tunnel, a danger alarm will be issued.

[0100] When water or mud bursts occur in the tunnel, an alarm signal will be automatically issued to warn of danger, reminding on-site construction personnel to respond in time and evacuate personnel. At the same time, when water or mud bursts occur in the tunnel, the power supply can be automatically shut down to improve safety.

[0101] The tunnel water and mud burst detection method provided by the present invention may also be used in conjunction with a computer-readable storage medium having a computer program stored thereon. The computer program is executed to implement the tunnel water and mud burst detection method. The computer program is capable of executing computer instructions, which include computer program code. The computer program code may be in source code form, object code form, an executable file, or some intermediate form.

[0102] Computer-readable storage media may include: any entity or device that can carry computer program code, recording media, USB flash drives, mobile hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0103] It should be noted that the content contained in computer-readable storage media can be appropriately increased or decreased according to the requirements of legislation and patent practices in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practices, computer-readable storage media do not include electrical carrier signals and telecommunications signals.

[0104] According to another aspect of the present invention, a tunnel water inrush and mud burst detection system is provided, which implements a tunnel water inrush and mud burst detection method. Figure 2 A block diagram of a tunnel water and mud burst detection system according to an embodiment of the present invention is shown, wherein the system comprises: a video acquisition module 510, a video image processing unit 520, a key frame extraction module 530, and a target detection module 540 connected in sequence;

[0105] The video acquisition module 510 is used to acquire the video image corresponding to the target lens;

[0106] The video image processing unit 520 is used to pre-process the video image to obtain a target video image;

[0107] A key frame extraction module 530 is used to extract key frames based on the target video image;

[0108] The target detection module 540 is used to determine whether water or mud inrush occurs in the tunnel based on the key frames and the target detection model.

[0109] For example, the video acquisition module 510 , the video image processing unit 520 , the key frame extraction module 530 , and the target detection module 540 may be connected via an internal network to ensure real-time transmission and processing of video images.

[0110] Example 1: Application of the method for detecting water inrush and mud burst in tunnels (eg Figure 3 shown)

[0111] At a tunnel construction site in a mountainous area, a system was installed to monitor tunnel water and mud inrush in real time. The surrounding geological conditions are complex, with numerous cracks and karst formations. Water and mud inrush are frequent, posing a serious threat to construction safety.

[0112] First, conduct on-site monitoring deployment:

[0113] Multiple high-definition visible light cameras are deployed in key areas of the tunnel (such as the tunnel walls, tunnel faces, and crack-prone areas) for 24 / 7 monitoring of these high-risk areas. These cameras transmit the collected visible light video data in real time via wired or wireless networks to a monitoring center located on the ground. The monitoring center is equipped with a video image processing unit to process and analyze the video data.

[0114] Second, the system detects initial water seepage:

[0115] One night, the monitoring system received high-definition camera data near the tunnel wall, indicating slight water seepage in the area. Using the video image processing unit, the system first preprocessed the video data (including grayscale conversion, noise reduction, and image enhancement), clearly displaying the changes in the water seepage area.

[0116] The key frame extraction module successfully detected obvious dynamic changes at the seepage point by calculating the pixel motion vectors of the seepage area in the video. Differential image analysis was then performed to further confirm that significant seepage had occurred in the area. The key frame extraction module then automatically extracted multiple key frames and marked the changes in the seepage area.

[0117] Third, the YOLOv10 model’s water column arc segmentation and water flow classification:

[0118] The system then uses the YOLOv10 model to further analyze the extracted keyframes. First, the YOLOv10 model performs pixel-level arc segmentation on the seepage area, accurately depicting the curvature of the water column. The arc segmentation results show that the curvature of the water column gradually increases, indicating an increase in water flow velocity.

[0119] The YOLOv10 model also performed a grading test on the water flow of the water column, determining the flow rate and velocity at the seepage point. The test results showed that the flow gradually evolved from a low flow rate to a medium flow rate, with the velocity increasing continuously. The system automatically integrated the arc segmentation results with the flow grading test results. Through multimodal data fusion, combined with the location of the water seepage point and the location of the mud, a preliminary assessment of the water inrush risk was made.

[0120] Fourth, real-time risk assessment and alarm triggering:

[0121] Based on the arc segmentation and water flow classification detection results generated by the YOLOv10 model, combined with the location of the water gushing point and the mud location, the system's risk assessment module determines that the current water seepage phenomenon may evolve into a water gushing and mud burst event. The system automatically generates a risk level and evaluates the water flow rate, water velocity, and duration of the water seepage point.

[0122] As the water flow and curvature of the water column gradually increased, the system determined that the seepage had a high probability of escalating into a high-risk water inrush event, raising the risk level to a disaster level. The intelligent early warning system immediately issued a high-risk alert, notifying tunnel construction managers to respond quickly. Simultaneously, the system activated its emergency response mechanism, automatically shutting down electrical equipment within the tunnel to prevent an electrical accident. Evacuation instructions were issued through the public address system, and construction personnel evacuated in an orderly manner according to the plan.

[0123] Fifth, follow-up monitoring and emergency response:

[0124] During the construction workers' evacuation, the monitoring system continued to monitor the tunnel's water inrush in real time. As the water inrush intensified, mud also began to flow into the tunnel, and the water and mud burst incident entered a critical stage. Using the YOLOv10 model, the system continuously monitored the mud burst area, automatically generating new keyframes and marking the expansion of the mud burst area.

[0125] Based on the detailed data provided by the system, including water inflow volume, velocity, arc changes, and mud outflow volume, the construction command center promptly adjusted its emergency response plan and implemented measures such as sealing water inflow openings and reinforcing the tunnel walls. Ultimately, thanks to the system's real-time warnings and data support, the construction team successfully controlled the water and mud outburst disaster and avoided a major safety accident.

[0126] In summary, the present invention provides a method, system, and medium for detecting water and mud inrush in tunnels, which have the following advantages over existing technologies:

[0127] The present invention first acquires video images, then preprocesses them, and then extracts key frames from the target video images. Then, based on the key frames and the target detection model, it determines whether water and mud bursts have occurred in the tunnel. In this way, by combining deep learning and image processing, water and mud bursts in tunnels can be automatically and intelligently identified. This improves the accuracy and effectiveness of water and mud burst detection while also improving the efficiency and real-time performance of water and mud burst detection, reducing labor maintenance costs and input. Furthermore, it meets the needs of detecting complex geological environments, and the detection results are not easily affected by the surrounding environment.

[0128] It should be understood that the embodiments disclosed herein are not limited to the specific structures, processing steps, or materials disclosed herein, but should extend to equivalent substitutions of these features understood by those skilled in the relevant art. It should also be understood that the terminology used herein is for the purpose of describing specific embodiments only and is not intended to be limiting.

[0129] In the description of the present invention, unless otherwise specified, "plurality" means two or more; terms such as "upper," "lower," "left," "right," "inner," "outer," "front end," "rear end," "head," and "tail" indicate positions or relationships based on those shown in the accompanying drawings. These terms are intended solely to facilitate the description of the present invention and simplify the description. They do not indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limiting the present invention. Furthermore, terms such as "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0130] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "connected" and "connection" should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integral connection; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediary. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0131] Certain terms are used throughout this application document to indicate specific system components. As will be appreciated by those skilled in the art, different names may be used to indicate the same component, and thus this application document is not intended to distinguish between components that are only different in name but not in function. In this application document, the terms "comprise," "include," and "have" are used in an open format and should therefore be interpreted as meaning "including, but not limited to...". In addition, the terms "substantially," "substantially," or "approximately" that may be used herein refer to industry-accepted tolerances for the corresponding terms. The term "coupling," as used herein, includes direct coupling and indirect coupling via another component, element, circuit, or module, wherein for indirect coupling, the intervening component, element, circuit, or module does not change the information of the signal but can adjust its current level, voltage level, and / or power level. Inferred coupling (e.g., one element is coupled to another element by inference) includes direct and indirect coupling between two elements in the same manner as "coupling."

[0132] References in this specification to "one embodiment" or "an embodiment" mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Therefore, appearances of the phrases "one embodiment" or "an embodiment" in various places throughout this specification do not necessarily refer to the same embodiment.

[0133] The embodiments of the present invention are presented for purposes of illustration and description and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those skilled in the art. The embodiments are chosen and described in order to better illustrate the principles of the invention and its practical application and to enable those skilled in the art to understand the invention and design various embodiments with various modifications as suited for specific applications.

[0134] Although the embodiments disclosed herein are as described above, the contents described herein are merely embodiments for facilitating understanding of the present invention and are not intended to limit the present invention. Any person skilled in the art of the present invention may make any modifications and changes in the form and details of the implementation without departing from the spirit and scope disclosed herein. However, the scope of patent protection of the present invention shall still be subject to the scope defined by the appended claims.

Claims

1. A method for detecting water inrush and mud burst in a tunnel, characterized in that: The method comprises: Obtain the video image corresponding to the target lens; Preprocessing the video image to obtain a target video image; Extracting key frames based on the target video image; According to the key frames and the target detection model, it is determined whether water gushing or mud bursting occurs in the tunnel.

2. The method according to claim 1, wherein The target video image is obtained by the following steps: Performing grayscale transformation on the video image to obtain a grayscale image; Performing noise reduction processing on the grayscale image to obtain a noise-reduced image; performing contrast enhancement and brightness enhancement on the noise-reduced image to obtain a processed image; Perform edge detection and sharpening processing on the processed image to obtain the target video image.

3. The method according to claim 1 or 2, wherein: The key frame is obtained by the following steps: Determining a motion vector of the target video image; Determine the area where the motion vector is greater than a target threshold as the video feature image; A differential image of the video feature image is determined, and the key frame is extracted according to the differential image.

4. The method according to any one of claims 1 to 3, wherein Determine whether water and mud inrush occurs in the tunnel by the following steps: Determining the location of the water gushing point, the location of the mud, the curvature characteristics of the water gushing column, and the water flow rate level based on the key frame and the target detection model; Whether water gushing and mud bursting occurs in the tunnel is determined according to the location of the water gushing point, the location of the mud, the curvature characteristics and the water flow level.

5. The method according to claim 4, wherein The target detection model includes a convolutional neural network; The position of the water inflow point and the position of the mud are determined by the following steps: Extracting features from the key frames using the convolutional neural network to obtain water gushing features and mud features; The water gushing feature and the mud feature are input into the target detection model to generate the position of the water gushing point and the position of the mud.

6. The method according to claim 4 or 5, characterized in that The arc characteristics and the water flow level are determined by the following steps: Performing pixel-level semantic segmentation on the key frame using the target detection model to obtain the curvature feature of the water column; The key frame is detected by the target detection model to obtain the water flow level of the water column.

7. The method according to any one of claims 4 to 6, wherein: Determine whether water and mud inrush occurs in the tunnel by the following steps: Performing multimodal data fusion on the radian feature and the water flow level to obtain the fused feature; generating a risk level according to the location of the water inflow point, the location of the mud, and the fusion characteristics; If the risk level is a disaster level, it is determined that water gushing and mud bursting occur in the tunnel.

8. The method according to any one of claims 1 to 7, wherein The method further comprises: When water or mud burst occurs in the tunnel, a danger alarm is issued.

9. A storage medium, characterized in that: It comprises a series of instructions for executing the method steps according to any one of claims 1-8.

10. A tunnel water and mud burst detection system, characterized in that: Executing the method according to any one of claims 1 to 8, the system comprises: a video acquisition module, a video image processing unit, a key frame extraction module, and a target detection module connected in sequence; The video acquisition module is used to obtain the video image corresponding to the target lens; The video image processing unit is used to pre-process the video image to obtain a target video image; The key frame extraction module is used to extract key frames based on the target video image; The target detection module is used to determine whether water gushing or mud bursting occurs in the tunnel based on the key frames and the target detection model.

Images