A dike seepage automatic monitoring and early warning system and method

Abstract

The present invention relates to the technical field of levee seepage monitoring, specifically to an automatic monitoring and early warning system and method for levee seepage, including an image matching unit, an early warning classification unit, a safety detection unit, a data acquisition unit, an abnormality detection unit and an abnormality alarm unit, and also includes: an image acquisition unit, the image acquisition unit is used to set an observation point at a target area, and an image acquisition device with a fixed position is set at the observation point, and the image acquisition device is used to perform image acquisition on a target dam body in the target area according to a preset acquisition cycle to obtain a target dam body image, and the target dam body image is transmitted to an image preprocessing unit. The present invention monitors the state of the target dam body in real time through image acquisition and processing, feature point extraction, and image matching units. This real-time performance can effectively detect potential seepage or structural abnormalities, and improve the monitoring accuracy and timely response capability of levee safety.

Description

Technical Field

[0001] The present invention relates to the technical field of dike seepage monitoring, and in particular to an automatic dike seepage monitoring and early warning system and method. Background Art

[0002] Levee seepage refers to the phenomenon in which groundwater infiltrates or passes through the materials of the dam or levee due to a water pressure difference caused by the groundwater level being higher than or in contact with the water level downstream of the dam. This seepage may cause leakage in the levee structure. If not handled in time, it may affect the stability and safety of the levee and even cause a dam breach.

[0003] Traditional systems usually rely on manual inspections and observations, which leads to delays and uncertainties in monitoring responses. The accuracy of manual detection is limited by the operator's experience and attention, and real-time, continuous monitoring cannot be achieved. Abnormal detection and alarms in traditional systems usually rely on the operator's experience and judgment, which are easily affected by subjective factors, and it is difficult to accurately identify small or gradual abnormalities. In addition, the monitoring coverage of traditional systems is limited, and they usually rely on limited observation points and equipment. They cannot fully cover large or complex embankment structures, and there is a risk of blind spots and missed detections. Summary of the invention

[0004] The purpose of the present invention is to address the problems existing in the background technology and to propose an automatic monitoring and early warning system and method for embankment seepage.

[0005] The technical solution of the present invention is: an automatic monitoring and early warning system for dike seepage, comprising an image matching unit, an early warning classification unit, a safety detection unit, a data acquisition unit, an abnormality detection unit and an abnormality alarm unit, and also comprising:

[0006] An image acquisition unit, wherein the image acquisition unit is used to set an observation point at the target area, set an image acquisition device with a fixed position at the observation point, and perform image acquisition of a target dam body in the target area according to a preset acquisition cycle by the image acquisition device to obtain an image of the target dam body, and transmit the image of the target dam body to an image preprocessing unit;

[0007] An image preprocessing unit, which receives the target dam body image transmitted by the image acquisition unit, and denoises the target dam body image by using the neighborhood grayscale features of the pixel points in the target dam body image to obtain a target dam body standard image, and transmits the target dam body standard image to a feature point extraction unit;

[0008] A feature point extraction unit receives the target dam body standard image transmitted by the image preprocessing unit, and extracts feature points from the target dam body standard image and the reference image based on the SIFT algorithm to obtain a first feature point set corresponding to the target dam body standard image and a second feature point set corresponding to the reference image, and transmits the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image to the image matching unit.

[0009] Preferably, the image matching unit receives the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image transmitted by the feature point extraction unit, and extracts the first feature descriptor set of the first feature point set and the second feature descriptor set of the second feature point set, calculates the similarity between the first feature descriptor set and the second feature descriptor set based on the improved Euclidean distance, and compares the similarity with a preset similarity threshold. If it is greater than the preset similarity threshold, the first feature point corresponding to the first feature descriptor and the second feature point corresponding to the second feature descriptor are added to a matching point set, and the number of the matching point sets is counted and normalized to obtain the matching degree between the target dam body standard image and the reference image, and the matching degree between the target dam body standard image and the reference image is transmitted to the early warning classification unit.

[0010] Preferably, the warning classification unit receives the matching degree between the target dam body standard image and the reference image transmitted by the image matching unit, and matches the matching degree between the target dam body standard image and the reference image with the warning matching table pre-stored in the database to obtain the warning level corresponding to the target dam body standard image. If the warning level is lower than the preset warning level threshold, a first warning signal is generated and the first warning signal is transmitted to the safety detection unit.

[0011] Preferably, the data acquisition unit is used to set a vibration sensor at the target dam body, obtain a vibration signal of the target dam body through the vibration sensor, and transmit the vibration signal of the target dam body to the abnormality detection unit.

[0012] Preferably, the abnormality detection unit receives the vibration signal of the target dam body transmitted by the data acquisition unit, and performs statistics on the frequency values ​​of the vibration signal of the target dam body at each time point to obtain a time-frequency diagram corresponding to the vibration signal of the target dam body, and detects the time-frequency diagram based on preset upper and lower limits of the vibration frequency to obtain a first detection result. If the first detection result indicates that the vibration signal of the target dam body is abnormal, the vibration signal of the target dam body is detected for abnormality through a trained abnormality detection model to obtain a second detection result, and a second warning signal is generated based on the second detection result, and the second warning signal is transmitted to the safety detection unit.

[0013] Preferably, the safety detection unit receives the first warning signal transmitted by the warning classification unit and the second warning signal transmitted by the abnormality detection unit, and averagely weights the first warning signal and the second warning signal to obtain a third warning signal, and matches the third warning signal with the abnormal alarm table pre-stored in the database to obtain the abnormal alarm corresponding to the target dam body, and transmits the abnormal alarm corresponding to the target dam body to the abnormal alarm unit.

[0014] Preferably, the abnormal alarm unit receives the abnormal alarm corresponding to the target dam body transmitted by the safety detection unit, and displays and alarms the abnormal alarm corresponding to the target dam body through the abnormal alarm device.

[0015] Preferably, the target dam body image is denoised using the neighborhood grayscale features of the pixel points in the target dam body image to obtain a target dam body standard image, comprising the following steps:

[0016] A1. Calculate the spatial domain function of the target dam image. The spatial domain function is as follows:

[0017] ;

[0018] in, Represents the spatial domain function of the target dam image, Indicates that the gray value in the target dam image is The coordinates of the pixel point, Indicates that the gray value in the target dam image is The coordinates of the pixel point, represents the Gaussian filter kernel in the spatial domain;

[0019] A2. Calculate the grayscale domain function of the target dam image. The grayscale domain function is as follows:

[0020] ;

[0021] in, Represents the gray domain function of the target dam image, Represents the Gaussian filter kernel in the spatial domain and grayscale domain;

[0022] A3. Calculate the filter function of the target dam image based on the spatial domain function and the gray domain function of the target dam image. The filter function is as follows:

[0023] ;

[0024] in, Represents the filter function of the target dam image, Indicates the filter window size;

[0025] A4, filtering the target dam body image based on the spatial domain function, gray domain function and filter function of the target dam body image to obtain a target dam body standard image, wherein the target dam body standard image is expressed as follows;

[0026] ;

[0027] in, represents the standard image of the target dam, Indicates the size of the target dam image, Indicates that the gray value in the target dam image is Pixels, Indicates that the gray value in the target dam image is Pixels.

[0028] Preferably, extracting a first feature descriptor set of the first feature point set and a second feature descriptor set of the second feature point set comprises the following steps:

[0029] B1. Calculate the pixel gradients of the first feature point set and the second feature point set in the Gaussian scale space. The pixel gradient calculation formula is as follows:

[0030] ;

[0031] in, represents the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, represents the difference in the horizontal direction of the first-order Gaussian scale space, represents the difference in the vertical direction of the first-order Gaussian scale space, and , represents the SOBEL operator along the horizontal direction of the first-order Gaussian scale space, represents the Gaussian scale space, and , represents the SOBEL operator in the first-order Gaussian scale space along the vertical direction;

[0032] B2. Calculate the final gradient amplitude and gradient direction of the first feature point set and the second feature point set based on the pixel gradients of the first feature point set and the second feature point set in the Gaussian scale space. The final gradient amplitude calculation formula is as follows:

[0033] ;

[0034] in, Represents the final pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, Represents the horizontal difference in pixel gradient between the first feature point set and the second feature point set in Gaussian scale space, Represents the difference in the vertical direction of the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space;

[0035] The final gradient direction calculation formula is as follows:

[0036] ;

[0037] in, Represents the final gradient direction of the first feature point set and the second feature point set;

[0038] B3. Combine the final pixel gradients and gradient directions of the first feature point set and the second feature point set to obtain a first feature descriptor set of the first feature point set and a second feature descriptor set of the second feature point set.

[0039] Preferably, the improved Euclidean distance calculation formula is as follows:

[0040] ;

[0041] in, Indicates the first feature point and the second feature point The improved Euclidean distance between Indicates the first feature point and the second feature point The position transformation error between , Indicates the first feature point The coordinates of Indicates the second feature point The coordinates of represents the similarity transformation model, represents the similarity model transformation parameters, Indicates the first feature point and the second feature point The main direction error between , represents the absolute value function, Indicates the first feature point and the second feature point The direction between represents the peak position of the main direction difference histogram, Represents the Euclidean distance between two feature points.

[0042] Preferably, the statistical formula for the frequency value of the vibration signal of the target dam body at each time point is as follows:

[0043] ;

[0044] in, Represents the frequency value of the vibration signal of the target dam at time point t, represents a window function, Represents the translation transformation parameter value of the window function, Represents the scaling transformation parameter value of the window function, Represents the vibration signal of the target dam body at time point t;

[0045] The first detection result is expressed as follows:

[0046] ;

[0047] in, Indicates the first test result, Indicates normal vibration signal, Indicates abnormal vibration signal, Indicates the lower limit of vibration frequency, Indicates the upper limit of vibration frequency.

[0048] The technical solution of the present invention is a method for automatic monitoring and early warning of dike seepage, which is applicable to the automatic monitoring and early warning system for dike seepage, and comprises the following steps:

[0049] S1. Setting observation points in the target area and installing fixed-position image acquisition equipment to regularly acquire images of the target dam body in the target area according to a preset acquisition cycle to obtain images of the target dam body;

[0050] S2, using the neighborhood grayscale features of the pixel points to perform denoising on the image, obtain a standard image of the target dam, and use the SIFT algorithm to extract a feature point set of the standard image of the target dam and the reference image;

[0051] S3, extracting feature descriptors of the feature point set, calculating the similarity between the feature descriptors, comparing the similarity with a preset similarity threshold, and determining the matching degree between the target dam body standard image and the reference image;

[0052] S4, comparing the matching degree with a preset warning matching table to determine the warning level of the target dam body standard image, and generating a first warning signal if the warning level is lower than a preset threshold;

[0053] S5, collecting vibration signals of the target dam body, analyzing and processing the vibration signals, detecting whether there is abnormal vibration, and generating a second warning signal;

[0054] S6. Perform weighted average processing on the first warning signal and the second warning signal to generate a third warning signal, and match the third warning signal with the abnormal alarm table to determine whether there is an abnormality in the target dam body, and display and alarm through the abnormal alarm device based on the third warning signal.

[0055] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0056] 1. The present invention monitors the state of the target dam in real time through image acquisition and processing, feature point extraction, and image matching units. This real-time performance can effectively detect potential seepage or structural abnormalities, improve the monitoring accuracy and timely response capabilities for embankment safety, and also integrates the detection and analysis of vibration signals. This multi-source data fusion can improve the accuracy and reliability of anomaly detection, reduce the false alarm rate, and ensure that truly important abnormalities can be identified and processed in a timely manner.

[0057] 2. The present invention can accurately classify and determine the severity of abnormalities according to the preset warning matching table and abnormality detection results. Such classification and level setting enable the monitoring system to take different emergency response measures according to different warning levels, and protect the safety of the embankment and the surrounding environment in a targeted manner. It can not only generate early warning signals, but also notify relevant personnel or management systems of abnormal alarms in real time through display and alarm devices. This real-time alarm and response mechanism helps to quickly take emergency measures and minimize potential disaster risks and losses. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 A schematic diagram of the overall system architecture in an embodiment of the present invention;

[0059] Figure 2 A schematic flow chart of the steps of an overall method in an embodiment of the present invention.

[0060] Figure numerals: 1. Image acquisition unit; 2. Image preprocessing unit; 3. Feature point extraction unit; 4. Image matching unit; 5. Warning classification unit; 6. Safety detection unit; 7. Data acquisition unit; 8. Abnormal detection unit; 9. Abnormal alarm unit. DETAILED DESCRIPTION

[0061] Embodiment 1, as Figure 1 As shown, the automatic monitoring and early warning system for levee seepage proposed by the present invention includes an image matching unit 4, an early warning classification unit 5, a safety detection unit 6, a data acquisition unit 7, an abnormality detection unit 8 and an abnormality alarm unit 9, and also includes:

[0062] An image acquisition unit 1 is used to set an observation point at the target area, set an image acquisition device with a fixed position at the observation point, and acquire an image of a target dam body in the target area through the image acquisition device according to a preset acquisition cycle to obtain an image of the target dam body, and transmit the image of the target dam body to the image preprocessing unit 2;

[0063] The image preprocessing unit 2 receives the target dam body image transmitted by the image acquisition unit 1, and uses the neighborhood grayscale features of the pixel points in the target dam body image to denoise the target dam body image to obtain a target dam body standard image, and transmits the target dam body standard image to the feature point extraction unit 3;

[0064] The feature point extraction unit 3 receives the target dam body standard image transmitted by the image preprocessing unit 2, and extracts feature points of the target dam body standard image and the reference image based on the SIFT algorithm to obtain a first feature point set corresponding to the target dam body standard image and a second feature point set corresponding to the reference image, and transmits the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image to the image matching unit 4.

[0065] In the present invention, the image acquisition device is a device installed on a fixed position at an observation point, and is used to capture the image of the target dam body in the target area in real time or periodically. The image acquisition device can be a video camera, a camera or other dedicated visual sensor device, and is used to capture the visual information of the target dam body; the neighborhood grayscale feature is to analyze and process the grayscale values ​​of each pixel point in the target dam body image and its surrounding pixels; the SIFT algorithm is a classic image feature extraction algorithm, and is used to detect and describe key points in an image. These feature points have scale invariance and rotation invariance, and can remain stable even when the image scale and viewing angle change; the reference image is a benchmark image used for comparison or matching. The reference image can be a frame of the target dam body image previously collected, or a standard image predetermined for comparison. By matching feature points with the reference image, the state change of the target dam body can be evaluated or specific change points can be detected.

[0066] In an optional embodiment, the image matching unit 4 receives the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image transmitted by the feature point extraction unit 3, and extracts the first feature descriptor set of the first feature point set and the second feature descriptor set of the second feature point set, calculates the similarity between the first feature descriptor set and the second feature descriptor set based on the improved Euclidean distance, and compares the similarity with a preset similarity threshold. If it is greater than the preset similarity threshold, the first feature point corresponding to the first feature descriptor and the second feature point corresponding to the second feature descriptor are added to the matching point set, and the number of matching point sets is counted and the number of matching point sets is normalized to obtain the matching degree between the target dam body standard image and the reference image, and the matching degree between the target dam body standard image and the reference image is transmitted to the early warning classification unit 5.

[0067] It should be noted that the feature point set is a collection of significant feature points extracted from the image; the feature descriptor set is a set of vectors that mathematically describe the local features of the area around each feature point; the Euclidean distance is a common method used in mathematics to measure the distance or difference between two vectors; the similarity threshold is a pre-set parameter used to determine whether the feature descriptors are similar enough to match; the matching point set contains feature point pairs that meet the similarity threshold conditions; normalization is the process of mapping data of different measurement scales into a unified standard range.

[0068] In an optional embodiment, the warning classification unit 5 receives the matching degree between the target dam body standard image and the reference image transmitted by the image matching unit 4, and matches the matching degree between the target dam body standard image and the reference image with the warning matching table pre-stored in the database to obtain the warning level corresponding to the target dam body standard image. If the warning level is lower than the preset warning level threshold, a first warning signal is generated and the first warning signal is transmitted to the safety detection unit 6.

[0069] It should be noted that the warning matching table pre-stores the relationship between different warning levels and matching degrees, which is used to determine the warning level corresponding to the target dam standard image; when the system determines that the warning level of the target dam standard image is lower than the preset threshold, this signal will be generated, indicating that there may be potential safety risks or abnormal conditions.

[0070] In an optional embodiment, the data acquisition unit 7 is used to set a vibration sensor at the target dam body, obtain a vibration signal of the target dam body through the vibration sensor, and transmit the vibration signal of the target dam body to the abnormality detection unit 8.

[0071] In an optional embodiment, the abnormality detection unit 8 receives the vibration signal of the target dam body transmitted by the data acquisition unit 7, and counts the frequency values ​​of the vibration signal of the target dam body at each time point to obtain a time-frequency diagram corresponding to the vibration signal of the target dam body, and detects the time-frequency diagram based on preset upper and lower limits of the vibration frequency to obtain a first detection result. If the first detection result indicates that the vibration signal of the target dam body is abnormal, the vibration signal of the target dam body is detected for abnormality through a trained abnormality detection model to obtain a second detection result, and a second warning signal is generated based on the second detection result, and the second warning signal is transmitted to the safety detection unit 6.

[0072] It should be noted that the time-frequency diagram performs windowed analysis on the vibration signal in time, and then performs Fourier transform or wavelet transform on the signal in each time window to obtain the frequency component of the signal in that time period. In this way, the characteristics of the signal in time and frequency can be displayed simultaneously in one image.

[0073] In an optional embodiment, the safety detection unit 6 receives the first warning signal transmitted by the warning classification unit 5 and the second warning signal transmitted by the abnormality detection unit 8, and averagely weights the first warning signal and the second warning signal to obtain a third warning signal, and matches the third warning signal with the abnormal alarm table pre-stored in the database to obtain the corresponding abnormal alarm of the target dam body, and transmits the corresponding abnormal alarm of the target dam body to the abnormal alarm unit 9.

[0074] In an optional embodiment, the abnormal alarm unit 9 receives the abnormal alarm corresponding to the target dam body transmitted by the safety detection unit 6, and displays and alarms the abnormal alarm corresponding to the target dam body through the abnormal alarm device.

[0075] Embodiment 2, an automatic monitoring and early warning system for embankment seepage proposed by the present invention, compared with embodiment 1, this embodiment further includes: using the neighborhood grayscale features of the pixel points in the target dam body image to denoise the target dam body image to obtain a target dam body standard image, including the following steps:

[0076] A1. Calculate the spatial domain function of the target dam image. The spatial domain function is as follows:

[0077] ;

[0078] in, Represents the spatial domain function of the target dam image, Indicates that the gray value in the target dam image is The coordinates of the pixel point, Indicates that the gray value in the target dam image is The coordinates of the pixel point, represents the Gaussian filter kernel in the spatial domain;

[0079] A2. Calculate the grayscale domain function of the target dam image. The grayscale domain function is as follows:

[0080] ;

[0081] in, Represents the gray domain function of the target dam image, Represents the Gaussian filter kernel in the spatial domain and grayscale domain;

[0082] A3. Calculate the filter function of the target dam image based on the spatial domain function and gray domain function of the target dam image. The filter function is as follows:

[0083] ;

[0084] in, Represents the filter function of the target dam image, Indicates the filter window size;

[0085] A4. Filter the target dam image based on the spatial domain function, gray domain function and filter function of the target dam image to obtain a target dam standard image. The expression of the target dam standard image is as follows;

[0086] ;

[0087] in, represents the standard image of the target dam, Indicates the size of the target dam image, Indicates that the gray value in the target dam image is Pixels, Indicates that the gray value in the target dam image is Pixels.

[0088] In this embodiment, the spatial domain function represents the spatial domain function of the target dam image, which generally refers to a function that describes the characteristics of each pixel in the image and its surrounding neighborhood; the grayscale domain function represents the grayscale domain function of the target dam image, which is used to describe the distribution and density of different grayscale levels in the image; the spatial domain Gaussian filter kernel represents a filter kernel used to smooth the image in the spatial domain; the filter function represents a function calculated based on the spatial domain function and the grayscale domain function, which is used to apply filtering operations in image processing; the filter window size is used to define the window size of the filter function, which generally represents the pixel neighborhood size considered when calculating the filtering effect.

[0089] In an optional embodiment, extracting a first feature descriptor set of a first feature point set and a second feature descriptor set of a second feature point set comprises the following steps:

[0090] B1. Calculate the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space. The pixel gradient calculation formula is as follows:

[0091] ;

[0092] in, represents the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, represents the difference in the horizontal direction of the first-order Gaussian scale space, represents the difference in the vertical direction of the first-order Gaussian scale space, and , represents the SOBEL operator along the horizontal direction of the first-order Gaussian scale space, represents the Gaussian scale space, and , represents the SOBEL operator in the first-order Gaussian scale space along the vertical direction;

[0093] B2. Calculate the final gradient amplitude and gradient direction of the first feature point set and the second feature point set based on the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space. The final gradient amplitude calculation formula is as follows:

[0094] ;

[0095] in, Represents the final pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, Represents the horizontal difference in pixel gradient between the first feature point set and the second feature point set in Gaussian scale space, Represents the difference in the vertical direction of the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space;

[0096] The final gradient direction calculation formula is as follows:

[0097] ;

[0098] in, Represents the final gradient direction of the first feature point set and the second feature point set;

[0099] B3. Combine the final pixel gradients and gradient directions of the first feature point set and the second feature point set to obtain a first feature descriptor set of the first feature point set and a second feature descriptor set of the second feature point set.

[0100] It should be noted that the Gaussian scale space is a set of images obtained by smoothing and scaling the images using Gaussian kernel functions of different scales; the pixel gradient represents the gradient of each pixel in the image, which refers to the direction and rate of change of the fastest changing grayscale value of the pixel.

[0101] In an optional embodiment, the improved Euclidean distance calculation formula is as follows:

[0102] ;

[0103] in, Indicates the first feature point and the second feature point The improved Euclidean distance between Indicates the first feature point and the second feature point The position transformation error between , Indicates the first feature point The coordinates of Indicates the second feature point The coordinates of represents the similarity transformation model, represents the similarity model transformation parameters, Indicates the first feature point and the second feature point The main direction error between , represents the absolute value function, Indicates the first feature point and the second feature point The direction between represents the peak position of the main direction difference histogram, Represents the Euclidean distance between two feature points.

[0104] In an optional embodiment, the statistical formula of the frequency value of the vibration signal of the target dam body at each time point is as follows:

[0105] ;

[0106] in, Represents the frequency value of the vibration signal of the target dam at time point t, represents a window function, Represents the translation transformation parameter value of the window function, Represents the scaling transformation parameter value of the window function, Represents the vibration signal of the target dam body at time point t;

[0107] The first test result is expressed as follows:

[0108] ;

[0109] in, Indicates the first test result, Indicates normal vibration signal, Indicates abnormal vibration signal, Indicates the lower limit of vibration frequency, Indicates the upper limit of vibration frequency.

[0110] Embodiment three, as Figure 2 As shown, the present invention proposes an automatic monitoring and early warning method for dike seepage, which is applicable to the automatic monitoring and early warning system for dike seepage, and comprises the following steps:

[0111] S1. Setting observation points in the target area and installing fixed-position image acquisition equipment to regularly acquire images of the target dam body in the target area according to a preset acquisition cycle to obtain images of the target dam body;

[0112] S2, using the neighborhood grayscale features of the pixel points to perform denoising on the image, obtain a standard image of the target dam, and use the SIFT algorithm to extract a feature point set of the standard image of the target dam and the reference image;

[0113] S3, extracting feature descriptors of the feature point set, calculating the similarity between the feature descriptors, comparing the similarity with a preset similarity threshold, and determining the matching degree between the target dam body standard image and the reference image;

[0114] S4, comparing the matching degree with a preset warning matching table to determine the warning level of the target dam body standard image, and generating a first warning signal if the warning level is lower than a preset threshold;

[0115] S5, collecting vibration signals of the target dam body, analyzing and processing the vibration signals, detecting whether there is abnormal vibration, and generating a second warning signal;

[0116] S6. Perform weighted average processing on the first warning signal and the second warning signal to generate a third warning signal, and match the third warning signal with the abnormal alarm table to determine whether there is an abnormality in the target dam body, and display and alarm through the abnormal alarm device based on the third warning signal.

[0117] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. An automatic monitoring and early warning system for embankment seepage, comprising an image acquisition unit (1), an image preprocessing unit (2), a feature point extraction unit (3), an image matching unit (4), an early warning classification unit (5), a safety detection unit (6), a data acquisition unit (7), an anomaly detection unit (8) and an anomaly alarm unit (9), characterized in that: The image acquisition unit (1) is used to set an observation point at the target area, set an image acquisition device with a fixed position at the observation point, and use the image acquisition device to acquire images of a target dam body in the target area according to a preset acquisition cycle to obtain an image of the target dam body, and transmit the image of the target dam body to the image preprocessing unit (2); The image preprocessing unit (2) receives the target dam body image transmitted by the image acquisition unit (1), and uses the neighborhood grayscale features of the pixel points in the target dam body image to denoise the target dam body image, then obtains a target dam body standard image, and transmits the target dam body standard image to the feature point extraction unit (3); Denoising the target dam image based on the neighborhood grayscale features of the pixel points in the target dam image to obtain a target dam standard image includes the following steps: A1. Calculate the spatial domain function of the target dam image. The spatial domain function is as follows: ； in, Represents the spatial domain function of the target dam image, Indicates that the gray value in the target dam image is The coordinates of the pixel point, Indicates that the gray value in the target dam image is The coordinates of the pixel point, represents the Gaussian filter kernel in the spatial domain; A2. Calculate the grayscale domain function of the target dam image. The grayscale domain function is as follows: ； in, Represents the gray domain function of the target dam image, Represents the Gaussian filter kernel in the spatial domain and grayscale domain; A3. Calculate the filter function of the target dam image based on the spatial domain function and the gray domain function of the target dam image. The filter function is as follows: ； in, Represents the filter function of the target dam image, Indicates the filter window size; A4, filtering the target dam body image based on the spatial domain function, gray domain function and filter function of the target dam body image to obtain a target dam body standard image, wherein the target dam body standard image is expressed as follows; ； in, represents the standard image of the target dam, Indicates the size of the target dam image, Indicates that the gray value in the target dam image is Pixels, Indicates that the gray value in the target dam image is Pixels of The feature point extraction unit (3) receives the target dam body standard image transmitted by the image preprocessing unit (2), and extracts feature points from the target dam body standard image and the reference image based on the SIFT algorithm to obtain a first feature point set corresponding to the target dam body standard image and a second feature point set corresponding to the reference image, and transmits the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image to the image matching unit (4); The image matching unit (4) receives the first feature point set corresponding to the target dam body standard image and the second feature point set corresponding to the reference image transmitted by the feature point extraction unit (3), and extracts a first feature descriptor set of the first feature point set and a second feature descriptor set of the second feature point set; The first feature descriptor set of the first feature point set and the second feature descriptor set of the second feature point set include the following steps: B1. Calculate the pixel gradients of the first feature point set and the second feature point set in the Gaussian scale space. The pixel gradient calculation formula is as follows: ； in, represents the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, represents the difference in the horizontal direction of the first-order Gaussian scale space, represents the difference in the vertical direction of the first-order Gaussian scale space, and , represents the SOBEL operator along the horizontal direction of the first-order Gaussian scale space, represents the Gaussian scale space, and , represents the SOBEL operator in the first-order Gaussian scale space along the vertical direction; B2. Calculate the final gradient amplitude and gradient direction of the first feature point set and the second feature point set based on the pixel gradients of the first feature point set and the second feature point set in the Gaussian scale space. The final gradient amplitude calculation formula is as follows: ； in, Represents the final pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space, Represents the horizontal difference in pixel gradient between the first feature point set and the second feature point set in Gaussian scale space, Represents the difference in the vertical direction of the pixel gradient of the first feature point set and the second feature point set in the Gaussian scale space; The final gradient direction calculation formula is as follows: ； in, Represents the final gradient direction of the first feature point set and the second feature point set; B3, combining the final pixel gradients and gradient directions of the first feature point set and the second feature point set to obtain a first feature descriptor set of the first feature point set and a second feature descriptor set of the second feature point set; Afterwards, the similarity between the first feature descriptor set and the second feature descriptor set is calculated based on the improved Euclidean distance, and the similarity is compared with a preset similarity threshold. If the similarity is greater than the preset similarity threshold, the first feature point corresponding to the first feature descriptor and the second feature point corresponding to the second feature descriptor are added to a matching point set, and the number of the matching point sets is counted and normalized to obtain the matching degree between the target dam body standard image and the reference image, and the matching degree between the target dam body standard image and the reference image is transmitted to the early warning classification unit (5); The improved Euclidean distance calculation formula is as follows: ； in, Indicates the first feature point and the second feature point The improved Euclidean distance between Indicates the first feature point and the second feature point The position transformation error between , Indicates the first feature point The coordinates of Indicates the second feature point The coordinates of represents the similarity transformation model, represents the similarity model transformation parameters, Indicates the first feature point and the second feature point The main direction error between , represents the absolute value function, Indicates the first feature point and the second feature point The direction between represents the peak position of the main direction difference histogram, Represents the Euclidean distance between two feature points; The warning classification unit (5) receives the matching degree between the target dam body standard image and the reference image transmitted by the image matching unit (4), and matches the matching degree between the target dam body standard image and the reference image with a warning matching table pre-stored in a database to obtain a warning level corresponding to the target dam body standard image; if the warning level is lower than a preset warning level threshold, a first warning signal is generated, and the first warning signal is transmitted to the safety detection unit (6); The data acquisition unit (7) is used to set a vibration sensor at the target dam body, obtain a vibration signal of the target dam body through the vibration sensor, and transmit the vibration signal of the target dam body to the abnormality detection unit (8); The abnormality detection unit (8) receives the vibration signal of the target dam body transmitted by the data acquisition unit (7), and performs statistics on the frequency values ​​of the vibration signal of the target dam body at various time points to obtain a time-frequency diagram corresponding to the vibration signal of the target dam body, and detects the time-frequency diagram based on preset upper and lower limits of the vibration frequency to obtain a first detection result. If the first detection result indicates that the vibration signal of the target dam body is abnormal, the vibration signal of the target dam body is detected for abnormality using a trained abnormality detection model to obtain a second detection result, and a second warning signal is generated based on the second detection result, and the second warning signal is transmitted to the safety detection unit (6); The statistical formula for the frequency value of the vibration signal of the target dam body at each time point is as follows: ； in, Represents the frequency value of the vibration signal of the target dam at time point t, represents a window function, Represents the translation transformation parameter value of the window function, Represents the scaling transformation parameter value of the window function, Represents the vibration signal of the target dam body at time point t; The first detection result is expressed as follows: ； in, Indicates the first test result, Indicates normal vibration signal, Indicates abnormal vibration signal, Indicates the lower limit of vibration frequency, Indicates the upper limit of vibration frequency.

2. The automatic monitoring and early warning system for levee seepage according to claim 1 is characterized in that: The safety detection unit (6) receives the first warning signal transmitted by the warning classification unit (5) and the second warning signal transmitted by the abnormality detection unit (8), performs average weighting on the first warning signal and the second warning signal to obtain a third warning signal, matches the third warning signal with an abnormality alarm table pre-stored in a database to obtain an abnormality alarm corresponding to the target dam body, and transmits the abnormality alarm corresponding to the target dam body to the abnormality alarm unit (9); The abnormal alarm unit (9) receives the abnormal alarm corresponding to the target dam body transmitted by the safety detection unit (6), and displays and alarms the abnormal alarm corresponding to the target dam body through the abnormal alarm device.

3. A dike seepage automatic monitoring and early warning method, which is applicable to a dike seepage automatic monitoring and early warning system according to any one of claims 1 to 2, characterized in that: The following steps are involved: S1. Setting observation points in the target area and installing fixed-position image acquisition equipment to regularly acquire images of the target dam body in the target area according to a preset acquisition cycle to obtain images of the target dam body; S2, using the neighborhood grayscale features of the pixel points to perform denoising on the image, obtain a standard image of the target dam, and use the SIFT algorithm to extract a feature point set of the standard image of the target dam and the reference image; S3, extracting feature descriptors of the feature point set, calculating the similarity between the feature descriptors, comparing the similarity with a preset similarity threshold, and determining the matching degree between the target dam body standard image and the reference image; S4, comparing the matching degree with a preset warning matching table to determine the warning level of the target dam body standard image, and generating a first warning signal if the warning level is lower than a preset threshold; S5, collecting vibration signals of the target dam body, analyzing and processing the vibration signals, detecting whether there is abnormal vibration, and generating a second warning signal; S6. Perform weighted average processing on the first warning signal and the second warning signal to generate a third warning signal, and match the third warning signal with the abnormal alarm table to determine whether there is an abnormality in the target dam body, and display and alarm through the abnormal alarm device based on the third warning signal.

Images