Power pipe gallery inspection method based on unmanned aerial vehicle

By atomizing detection and defogging treatment of the power gallery images collected by the drone, combined with the maximum likelihood estimation theory, the drone inspection strategy is adjusted, and the inspection quality problems in high underground humidity and motion blur environments are solved. , improves the efficiency and accuracy of power pipeline inspection and reduces the probability of failure.

CN118967558AInactive Publication Date: 2024-11-15福建华电可门发电有限公司 +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410923162.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-10
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing drone power pipeline inspection technology is difficult to obtain accurate images in environments with high underground humidity and blurred motion, resulting in poor inspection quality and inability to adjust the inspection strategy in real time, increasing the probability of power failure.

Method used

By performing atomization detection and defogging treatment on the collected images, the most realistic images are obtained using sub-block local overlap histogram equalization and maximum likelihood estimation theory, and the drone patrol strategy is adjusted according to the patrol strategy adaptation information. To meet the needs of different power pipeline areas.

Benefits of technology

It improves the efficiency and accuracy of drone power pipeline inspection, reduces the probability of power failure, and ensures the stability of power supply.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967558A_ABST
    Figure CN118967558A_ABST
Patent Text Reader

Abstract

The invention discloses an electric power pipe gallery inspection method based on an unmanned aerial vehicle, relates to the technical field of pipe gallery inspection processing, and is used for solving the problem of low inspection efficiency of an electric power pipe gallery. The method comprises the following steps: collecting an image of a corresponding inspection area in an inspection process, carrying out atomization detection on the collected image, screening out an atomized image, carrying out defogging operation on the screened atomized image by using sub-block local overlapping histogram equalization, and obtaining a defogged image; the method comprises the following steps of: acquiring an image of an unmanned aerial vehicle, performing defuzzification processing on the image by using a maximum likelihood estimation theory to obtain a most real image, and finally, acquiring inspection strategy adaptation information generated by the unmanned aerial vehicle in the inspection process of the electric power pipe gallery, and determining adaptation conditions of the unmanned aerial vehicle with different electric power pipe gallery areas in the inspection process of the electric power pipe gallery. Different strategies are used for regulation and control of electric power pipe gallery inspection, so that the efficiency of unmanned aerial vehicle electric power pipe gallery inspection is improved, and the probability of occurrence of electric power faults is reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of power pipeline corridor inspection, and in particular to a power pipeline corridor inspection method based on an unmanned aerial vehicle. Background Art

[0002] A power pipeline corridor is a tunnel or pipeline specially designed to accommodate cables and other power infrastructure. It is mainly used for power transmission and distribution. These pipeline corridors are usually located underground in cities and may sometimes be located in other areas, such as industrial facilities. The main purpose of the power pipeline corridor is to protect the cables from the external environment and facilitate the maintenance and management of the power system. Burying the transmission cables deep in the power pipeline corridor can not only improve the city's image and achieve rational use of land resources, but also ensure the stable transmission of the power system.

[0003] The prior art has the following deficiencies:

[0004] With the construction of a large number of lines, the area occupied by the pipeline corridor is getting larger and larger, the landforms involved are also complex and changeable, and the cable layout inside the power pipeline corridor is also intricate. Manual inspections cannot meet specific inspection needs. When conducting power inspections, drone inspections are gradually used in combination with manual inspection communities to monitor power pipeline corridors. However, due to the humid underground environment and motion blur during the shooting process, the accuracy of drone image acquisition of the power pipeline corridor is affected, resulting in substandard inspection quality of the power pipeline corridor. Some areas are prone to power or image acquisition problems. The historical inspection strategy allows drones to use more resources to inspect this area, but if the inspection is carried out according to this strategy, the inspection quality of other areas will be reduced accordingly during the latest inspection process, and the corresponding inspection strategy cannot be adjusted in real time, which increases the number of omissions in the power inspection process and has an adverse impact on subsequent power supply.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a power pipeline corridor inspection method based on unmanned aerial vehicle (UAV). The method collects images of corresponding inspection areas during the inspection process, performs fog detection on the collected images, and screens out fogged images. The screened out fogged images are defogged using sub-block local overlapping histogram equalization, and the images are defuzzified using maximum likelihood estimation theory to obtain the most realistic image. Finally, the inspection strategy adaptation information generated by the UAV during the inspection of the power pipeline corridor is obtained, the adaptation of the UAV to different power pipeline corridor areas during the inspection of the power pipeline corridor is determined, and different strategies are used to regulate the power pipeline corridor inspection to solve the shortcomings of the background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a power pipeline corridor inspection method based on a drone, comprising the following steps:

[0008] Develop drone inspection strategies and routes based on the geographical location, structural complexity, and key inspection points of the pipeline corridor, and collect images of the corresponding inspection areas during the inspection process;

[0009] Perform fog detection on the collected images, use the statistical features of the images in combination with the Canny edge detector to evaluate the image clarity and filter out the foggy images, and use sub-block local overlapping histogram equalization to perform defogging operations on the filtered foggy images;

[0010] Use the maximum likelihood estimation theory to deblur the image to obtain the most realistic image, and then perform image defect detection after obtaining the most realistic image;

[0011] Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power corridor, determine the adaptation of the inspection strategies in different areas and different power corridor areas during the inspection of the power corridor by the drone, and use different strategies to regulate the power corridor inspection according to the obtained adaptation results.

[0012] Preferably, the collected image is subjected to fog detection, and the statistical features of the image are combined with the Canny edge detector to evaluate the image clarity and filter out the fogged image, including:

[0013] Preprocess the image and convert it from RGB color space to HSV color space;

[0014] Calculate the average brightness, image contrast and average saturation of the image in the statistical features of the image, and perform preliminary fog detection of the image based on the average brightness, image contrast and average saturation;

[0015] After preliminary fog detection, the image is converted from RGB color space to grayscale color space, the horizontal and vertical gradients of the image grayscale are calculated, the edge with the largest gradient is retained, and the number of edge pixels is calculated as the number of edges detected in the image. The processed image is compared with the edge detection results of a clear and fog-free image to screen out the fogged image.

[0016] Preferably, the filtered fogged image is defogged using sub-block local overlapping histogram equalization, and the specific steps include:

[0017] Calculate the histogram of the input fogged image, and calculate the gray value change table according to the histogram of the input image;

[0018] Use the grayscale change table to perform a lookup table transformation operation, and traverse each pixel point, put the grayscale value of the original image into the exchange table, and obtain the new grayscale value of the pixel point;

[0019] The sub-block size and overlapping area are determined using sub-block local overlapping histogram equalization, and a sliding window is used to traverse the entire image and reconstruct the image from all processed sub-blocks.

[0020] Preferably, the image is deblurred using the maximum likelihood estimation theory to obtain the most realistic image, and the specific steps are as follows:

[0021] Establishing an image degradation model, establishing the image degradation model through the influence of a blur model or through the influence of noise, and estimating the original form of the image according to the image degradation model;

[0022] In the process of image deblurring, maximum likelihood estimation is used to find the most similar parameter estimate of the blurred image as the most realistic image.

[0023] Preferably, the inspection strategy adaptation information generated by the drone during the inspection of the power pipe gallery is obtained to determine the adaptation of the inspection strategies of different areas and different power pipe gallery areas during the inspection of the power pipe gallery by the drone. The specific steps are as follows:

[0024] Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power pipeline corridor. The inspection strategy adaptation information includes data processing quality information and inspection performance impact information;

[0025] Data processing quality information includes the signal-to-noise structure similarity index, and inspection performance impact information includes the inspection adaptation stability index:

[0026] The signal-to-noise structure similarity index and the inspection adaptation stability index are combined to generate the inspection control coefficient;

[0027] The signal-to-noise structure similarity index and the inspection adaptation stability index are proportional to the inspection control coefficient.

[0028] Preferably, different strategies are used to regulate the inspection of the power pipeline corridor according to the obtained adaptation results, and the specific steps include:

[0029] Compare the generated inspection control coefficient with the inspection strategy control threshold;

[0030] If the inspection control coefficient is greater than the inspection strategy control threshold, a inspection measurement stability signal is generated without additional maintenance or processing;

[0031] If the inspection control coefficient is less than or equal to the inspection strategy control threshold, an inspection measurement abnormality signal is generated to adjust and optimize the UAV inspection strategy.

[0032] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0033] The present invention collects images of corresponding inspection areas during the inspection process, performs fog detection on the collected images, and screens out fogged images, performs defogging operation on the screened out fogged images using sub-block local overlapping histogram equalization, and uses maximum likelihood estimation theory to defuzzify the images to obtain the most realistic images. Finally, the inspection strategy adaptation information generated by the UAV during the inspection of the power corridor is obtained, the adaptation of the UAV to different power corridor areas during the inspection of the power corridor is determined, and different strategies are used to regulate the power corridor inspection, thereby improving the efficiency of the UAV power corridor inspection and reducing the probability of power failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0035] Figure 1 The present invention is a flow chart of a method for inspecting a power pipeline corridor based on a drone. DETAILED DESCRIPTION

[0036] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0037] Embodiment 1: The present invention provides Figure 1 A method for inspecting a power pipeline corridor based on a drone is shown, and the method comprises the following steps:

[0038] Develop drone inspection strategies and routes based on the geographical location, structural complexity, and key inspection points of the pipeline corridor, and collect images of the corresponding inspection areas during the inspection process;

[0039] Perform fog detection on the collected images, use the statistical features of the images in combination with the Canny edge detector to evaluate the image clarity and filter out the foggy images, and use sub-block local overlapping histogram equalization to perform defogging operations on the filtered foggy images;

[0040] Use the maximum likelihood estimation theory to deblur the image to obtain the most realistic image, and then perform image defect detection after obtaining the most realistic image;

[0041] Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power corridor, determine the adaptation of the inspection strategies in different areas and different power corridor areas during the inspection of the power corridor by the drone, and use different strategies to regulate the power corridor inspection according to the obtained adaptation results.

[0042] Develop detailed inspection strategies and routes based on the geographical location, structural complexity and key inspection points of the pipeline corridor, consider the endurance and load limit of the drone, arrange necessary strategies for charging or replacing batteries, select suitable drone models, and equip them with high-resolution cameras, infrared sensors, lidar, etc. to ensure that the drone can capture detailed images and other relevant data. At the same time, equip the drone with anti-fogging lenses and technology to adapt to flying in narrow spaces;

[0043] Perform a thorough pre-flight check of the drone, including battery charge, sensor functionality, and flight control system, to ensure all equipment is in top condition;

[0044] The drone is started at the entrance of the power corridor, automatically takes off according to the preset flight plan, and uses GPS or other positioning technology to accurately locate the starting point. Before entering the power corridor, the drone begins to collect surrounding environmental data, such as temperature and humidity, to provide environmental reference for subsequent flights;

[0045] The drone automatically follows the predetermined route based on the navigation system and obstacle avoidance technology. During this process, the drone will adjust its flight altitude and speed to adapt to the narrow and complex environment in the tunnel.

[0046] Using various sensors and cameras carried by drones, we continuously collect videos, images and other sensor data inside the tunnel, process the relevant data of the image type, and use sub-block local overlapping histogram equalization to defog the image. The specific steps are as follows:

[0047] Preprocess the image, load the image collected in the power pipeline corridor area that needs to be defogged, and select the images of the power pipeline corridor that needs to be defogged based on the visual features, quality indicators and requirements related to the specific task of the image. The image is determined based on the image statistics. The impact of fogging, for example, fogged images usually have lower contrast and higher brightness. The statistical features of the image, such as brightness, contrast and color saturation, are used to evaluate the degree of fogging in the image, as follows:

[0048] Preprocess the image to ensure that the input image is in the appropriate format and color space. For most image analysis, converting the image from RGB color space to HSV or YCbCr color space is more helpful in separating brightness and color information. In HSV color space, H stands for hue, S stands for saturation, and V stands for value (brightness). In YCbCr color space, Y stands for brightness, and Cb and Cr stand for chrominance components.

[0049] Brightness is a key indicator in fog image analysis because fog increases the overall brightness of the image. The average brightness is calculated using the expression: Among them, V i is the brightness value of the i-th pixel in the HSV color space, and N is the total number of pixels in the image;

[0050] After the image is fogged, the contrast of the image is usually reduced, making the image look flatter and blurrier. The image standard deviation is calculated to evaluate the image contrast. The calculation expression is: Among them, σ is the standard deviation of brightness, which reflects the degree of change of the image brightness value relative to the average brightness; L avg is the average brightness, V i is the brightness value of the i-th pixel in the HSV color space, N is the total number of pixels in the image,

[0051] The color saturation of the image under the influence of fog may be reduced, because the image will scatter light and reduce the depth of color after fogging, so the average saturation of the image is calculated to determine the average saturation: Among them, S i is the saturation value of the i-th pixel in the HSV color space, and N is the total number of pixels in the image;

[0052] Combining the above indicators, a preliminary fog detection scheme can be formed. According to the needs of specific applications, thresholds can be set to decide whether the image needs to be dehazed. The decision criteria can be: if the average brightness is higher than a certain threshold and the image contrast is lower than another threshold, and the average saturation is lower than a third threshold, it is judged that there is fog in the image, and the image is marked and screened for subsequent processing.

[0053] Edge detection is a technique commonly used in image analysis, especially for evaluating image clarity and determining the degree of image fogging. The edges of a fogged image will become blurred, resulting in a reduction in the number of identifiable edges in the edge detection results. The following are the specific steps for using the edge detection algorithm Canny edge detector to evaluate image clarity and secondary detection to determine the fogged image:

[0054] Load the image to be analyzed (the image that passed the preliminary haze detection scheme) and convert the image from RGB color space to grayscale color space, because edge detection is usually performed on grayscale images;

[0055] Use the Sobel operator to calculate the horizontal and vertical gradients of the image grayscale. The horizontal gradient is: Vertical Gradient: The edge strength G is calculated by the following formula: Where A is the image area, * represents the convolution operation, G x and G y Respectively represent the image grayscale value detected by horizontal edge and the image grayscale value detected by vertical edge in the grayscale image;

[0056] The Canny edge detector is a multi-step process that first preprocesses the image, applies a Gaussian filter to smooth the image to reduce image noise, obtains the gradient strength and direction of the smoothed image using the Sobel operator, eliminates non-edge pixels, and retains only the edges with the largest gradients. Two thresholds (a high threshold and a low threshold) are used to determine the real edges and potential edges. By suppressing isolated weak edges, only weak edges connected to strong edges are retained.

[0057] Count the number of edges detected in the image by calculating the number of edge pixels. A decrease in the number of edges usually indicates that the image has lost clarity due to fogging. Compare the edge detection results of the processed image with those of the clear (non-fogging) image to evaluate the impact of fogging. Based on the changes in the number of edges and clarity, decide whether the image needs further dehazing. This process can be automated using image processing software or programming tools (such as OpenCV) to effectively evaluate and filter out fogged images.

[0058] The filtered fogged image is defogged using sub-block local overlapping histogram equalization. The specific steps are as follows:

[0059] First, the image is preprocessed, the histogram of the input fogged image is calculated, and then the gray value change table is calculated according to the histogram of the input image;

[0060] Perform a lookup table transformation, and use the grayscale change table to perform a lookup table transformation operation on the preprocessed histogram. Specifically, the operation is to traverse each pixel point and put the grayscale value of the original image into the exchange table to obtain the new grayscale value of the point;

[0061] The local overlapping histogram equalization method of sub-blocks is used to improve the local contrast of the image. By allowing overlap between sub-blocks, the block effect (i.e., obvious block boundaries) is effectively reduced, and the overall visual quality of the image is improved. The specific steps are as follows:

[0062] Determine the sub-block size and overlapping area. For example, set the image size to 3*3 pixels and divide it into 9 sub-blocks. Determine the overlapping area between sub-blocks, that is, the number of overlapping pixels. Obtain the transformation function of the required sub-block by weighting adjacent sub-blocks. The weighted weight parameter is determined by the fixed step length of moving the sub-block. The fixed step length is half of the sub-block length, and the sub-block is moved by half of the sub-block length each time.

[0063] Assume that the four adjacent small blocks are taken as a whole, where the histogram transformation function of each block is: T 1 (r k ), T 2 (r k ), T 3 (r k ), T 4 (r k ), the 3*3 pixel image is divided into 9 blocks, and different blocks are distinguished by labels a, b, c, d, e, f, g, h, i. The change function of area e is composed of the sliding superposition of four sub-blocks;

[0064] The histogram change functions of each block are: In the formula, n represents the total number of pixels to be processed. represents the number of pixels with grayscale value j in the template area a, k represents the total number of grayscale values, and r j Represents the gray level with gray value j. Similarly, In the above figure, a is replaced by b, c, d, e, f, g, h, i, and P represents the equalization processing function;

[0065] The histogram change function of each block is summarized to obtain the change function of area e The formula is:

[0066] Use a sliding window to traverse the entire image, and the window moves less than the size of a sub-block each time (based on the amount of overlap), ensuring that each pixel is covered by multiple sub-blocks;

[0067] Reconstruct the entire image using all processed sub-blocks to ensure that all overlapping parts are smoothly blended together. Post-processing steps such as sharpening can be performed to further improve image clarity.

[0068] The sub-block local overlapping histogram equalization method can effectively improve the visibility of images in poor visual environments. It is particularly suitable for dealing with situations such as fogging that cause visual degradation. By reducing the block effect, this method significantly improves the visual quality of the image and is suitable for complex environmental monitoring and image acquisition and analysis.

[0069] When processing the collected color images, especially in the foggy power pipeline images, the HSI (hue, saturation, brightness) model can more effectively separate the color hue and brightness information, and perform color correction and image quality enhancement more accurately. The specific steps are as follows:

[0070] Convert from RGB to HIS, calculate the brightness of the captured image, the brightness component I is the average value of the three color (R, G, B) components: I = (R + G + B) / 3, calculate the saturation, only when I is 0, the saturation is 0, in other cases, the saturation S is calculated according to the following formula:

[0071] The calculation of hue H depends on the relative size of RGB values, and the formula is: It should be noted that each pixel of a color image has three channels, which represent the three components of R (Red), G (Green), and B (Blue), and each channel has an 8-bit depth. The value range of each channel is 0-255. These three color components can also be used to represent different levels of gray. The above R, G, and B components represent the gray values ​​of each pixel in the corresponding channel.

[0072] After converting to the HSI model, limited contrast adaptive histogram equalization can be used for brightness (I) component to enhance the image quality. Only the brightness component is enhanced to improve the image contrast without affecting the hue and saturation, increase the local contrast of the image, and reduce the noise amplification in the bright or dark areas.

[0073] The processed brightness component is recombined with the unchanged hue (H) and saturation (S) components to generate a new HSI image, and an inverse transformation is used to convert the HSI image back to the RGB model for display on a regular display device.

[0074] When a drone inspects a power pipeline corridor, there will be a relative displacement between the camera on the drone and the object being photographed. This relative displacement will cause the captured image to be blurry, i.e. motion blur.

[0075] In the image deblurring process, the maximum likelihood estimation (MLE) theory is used to deblur the image. This method relies on a statistical model to estimate the original form of the image, that is, through a model of image degradation (such as blur caused by camera shake, focus error, etc.) for analysis. The specific steps of deblurring are: establish a statistical model. Before deblurring, it is necessary to define an image degradation model. The image degradation model can be expressed as: g = Hf + n, where g is the blurred image collected and observed, f is the unknown original real image, H is the blur model, and n is Gaussian noise;

[0076] Maximum likelihood estimation is to find the parameter estimation that makes the observed blurred image most likely. In image deblurring, the expression formula is as follows: In the formula, argmax f represents the value of the original real image f that makes the probability density function of the blurred image g reach the maximum value with respect to the original real image f. p(g|f) is the probability density function of observing the blurred image g given the real image f. When the blurred image g is given, f is unknown. Find the real image f that is most likely to make the observed blurred image g.

[0077] In the problem of image deblurring, the image degradation model is established through the following two factors p(g|f), that is, the image degradation model is established through the influence of the blur model H, that is, the influence of the blur operation on the real image is described, which can be represented by the blur kernel h of the blur model. The blur model H can be represented by a convolution operation, that is, g=Hf+n=h*f+n, where * represents the convolution operation;

[0078] The image degradation model is established through the influence of noise n, that is, describing the uncertainty introduced by noise in the blurred image g. It is usually assumed that the noise obeys the Gaussian distribution N, and its probability density function can be expressed as: p(n) = N(0, σ 2 ), where σ 2 is the variance of the noise;

[0079] The probability density function after modeling is in the form of: p(g|f)=p(g|Hf)*p(n), where p(g|Hf) represents the influence of the blur model H on the real image f when the blur image g is given, and p(n) is the probability density function of the noise. Through maximum likelihood estimation, we can determine the most likely real image f under the influence of the blur model H and the image degradation model of the noise n.

[0080] After obtaining real images, automated or manual defect detection can be performed, which includes identifying and locating cracks, corrosion, leakage, insulation damage and other problems in the pipeline corridor. Clear images help engineers and technicians perform preventive maintenance. By detecting potential problems early, repairs can be arranged in advance to prevent small problems from turning into major failures, reducing unplanned downtime and maintenance costs. Clear images can be used to train and improve machine learning models.

[0081] Through large amounts of high-quality data, machine learning algorithms can more accurately identify and classify various types of pipeline corridor defects, further improving the accuracy and efficiency of automatic detection.

[0082] During the inspection of the power corridor, drones will inspect different areas for different lengths of time according to historical inspection data. For example, some areas are prone to power or image acquisition problems, so all drones will use more resources to inspect this area. However, if the inspection is carried out according to the set strategy, the inspection quality of other areas will be reduced during the latest inspection. When problems occur in other areas, the corresponding inspection strategy cannot be adjusted in time, which may cause some omissions in the power inspection. Therefore, an inspection effect feedback mechanism is established to monitor and evaluate the inspection effect in real time, and adjust the inspection strategy in time when problems are found;

[0083] During the inspection of the power corridor by drone, the images collected and transmitted by the drone are analyzed to determine the data collection of different areas by the drone during the inspection of the power corridor, determine the areas that need to be inspected, adjust the flight route of the drone, increase the inspection frequency of key areas, and then adjust the inspection strategy or method of the drone in real time, increase the inspection time of power areas where problems such as power failure, image acquisition, data measurement, etc. occur, and adjust the drone inspection strategy in real time. The specific steps are as follows:

[0084] Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power corridor, analyze the inspection strategy adaptation information, determine whether the inspection strategies for different areas of the drone are adapted to different power corridor areas during the inspection of the power corridor, control the drone according to the analysis results, and adjust different inspection strategies. The inspection strategy adaptation information includes data processing quality information and inspection performance impact information;

[0085] The data processing quality information includes the signal-noise structure similarity index, and the inspection performance impact information includes the inspection adaptation stability index. After collection, the signal-noise structure similarity index and the inspection adaptation stability index are calibrated as XZJ and XJS respectively.

[0086] The signal-to-noise structure similarity index is used to measure and evaluate the quality of images obtained during the UAV power pipeline inspection process. It reflects the signal-to-noise ratio and structural similarity characteristics of the images collected in the inspection area. It is used to ensure that the collected images are of high quality, provide reliable data support for fault detection and maintenance decisions, optimize inspection strategies, and improve inspection efficiency and effectiveness.

[0087] High-quality images are the basis for identifying and detecting problems in power pipe corridors (such as cracks, corrosion, leakage, etc.). The signal-to-noise structure similarity index can ensure that the collected images have sufficient signal-to-noise ratio and structural clarity. By automatically evaluating and screening images with high signal-to-noise structure similarity index, the time and workload of manual inspection can be reduced and the overall efficiency of inspection can be improved. The signal-to-noise structure similarity index can be used to monitor and adjust the inspection parameters of drones (such as flight altitude, shooting angle, sensor settings, etc.) in real time to ensure that high-quality images can be obtained under various environmental conditions. During the inspection process, if the signal-to-noise structure similarity index value is low, the flight parameters of the drone can be automatically adjusted or secondary shooting can be performed to obtain higher quality images. The signal-to-noise structure similarity index has an impact on the following aspects:

[0088] Inspection efficiency: Image quality with high signal-to-noise structure similarity index values ​​can reduce repeated inspections of the same area and improve inspection efficiency. The inspection team can rely on high-quality images for analysis and decision-making more confidently. By monitoring the signal-to-noise structure similarity index, the flight parameters of the drone (such as altitude, speed, angle) can be dynamically adjusted to optimize resource utilization and ensure the best image is obtained in the shortest time;

[0089] Maintenance decision-making: High-quality images help to detect potential problems in advance, support preventive maintenance strategies, and reduce the risk of sudden failures and maintenance costs. By providing images with high signal-to-noise structural similarity index, the inspection team can more accurately assess the status of the pipeline corridor and develop more scientific maintenance plans and emergency plans.

[0090] Data storage and management: The signal-to-noise structure similarity index can be used as a standard for image quality screening, automatically screening out high-quality images for storage and further analysis, saving storage space and processing time. Images have higher reference value in data comparison and trend analysis, which is helpful for long-term monitoring and analysis of the status changes of power pipeline corridors.

[0091] The logic for obtaining the signal-noise structure similarity index is as follows:

[0092] The image collected during the inspection of the power pipeline corridor is obtained as the original image I, the processed image K after the original image is processed is obtained, and the image error value MW is calculated. The calculation expression is: Where I(i, j) is the pixel value of the original image at position (i, j), K(i, j) is the pixel value of the processed image at position (i, j), m and n are the width and height of the image, and the image signal-to-noise value SN is calculated as follows: In the formula, MAX I is the maximum pixel value of the original image I, and MW is the image error value;

[0093] Get the original image I and the processed image K, and calculate the local mean μ of the original image and the processed image I , the calculation expression is: Where N is the number of pixels in the image area, I u , K u Represents the pixel value of the original image I and the processed image K at the u-th position of the image, and calculates the local difference of the image. The calculation expression of the local difference of the original image is: The local difference calculation expression of the processed image is: Calculate the local similarity value of the image, the calculation expression is:

[0094] Calculate the signal-to-noise structure similarity index XZJ, the calculation expression is:

[0095] It should be noted that the maximum pixel value of the original image is determined by the number of pixel bits in the image. For an 8-bit image, the maximum pixel value is usually 255. For a grayscale image, each pixel is a single intensity value. For a color image, each pixel is a total value containing red, green, and blue (RGB) values.

[0096] During the inspection process, the UAV provides a comprehensive evaluation of the inspection quality by combining flight performance parameters and environmental monitoring parameters, which helps to optimize inspection strategies, improve data quality, ensure flight safety and reliability, support scientific decision-making, and promote technology improvements and operational training.

[0097] The inspection adaptation stability index indicates the ability of the UAV to adapt to different environmental conditions and flight stability during the inspection process. By analyzing the inspection adaptation stability index, it can be ensured that the UAV can operate efficiently and stably under various environmental conditions, provide high-quality inspection data, provide a quantitative evaluation of the UAV flight performance, and also reflect the impact of environmental factors on the inspection quality. This is for optimizing UAV inspection strategies, improving inspection efficiency and effectiveness, and ensuring the safe operation of power pipe corridors. The inspection adaptation stability index affects the following aspects:

[0098] Inspection efficiency: By real-time monitoring and adjusting flight parameters (such as altitude, speed, and attitude), the inspection adaptation stability index can help the drone fly on the optimal path, avoiding unnecessary repetition and detours, thereby improving inspection efficiency. While maintaining a high inspection adaptation stability index, the drone can complete the inspection task more quickly and efficiently, reducing the total time required for inspection;

[0099] Data quality: A high inspection adaptability stability index means that the drone remains stable during flight, which helps to obtain clear, unambiguous image data, which is crucial for identifying and analyzing potential problems in power pipeline corridors. By adapting to different environmental conditions, the inspection adaptability stability index ensures that the drone can collect high-quality images and sensor data in various situations, ensuring the integrity and accuracy of the data.

[0100] The inspection adaptation stability index has an important impact on the efficiency, data quality, safety, maintenance decisions, operation training, environmental adaptability, decision support and economic benefits of the drone power corridor inspection process. By monitoring and optimizing the inspection adaptation stability index, the overall performance and effectiveness of the drone inspection system can be significantly improved, ensuring the high quality and efficient completion of the inspection tasks.

[0101] The logic for obtaining the inspection adaptation stability index is as follows:

[0102] Get the pre-planned flight path of the drone and record the GPS coordinates (x pre ,y pre , z pre ), during the flight of the drone, the GPS coordinates (x, y, z) of each key point on the actual flight path are recorded in real time, and the distance deviation between the actual path and the preset path at each key point is calculated. The calculation expression is: In the formula, x t ,y t 、z t Respectively represent the actual x, y, and z coordinates at point t. t,pre 、x t,pre 、x t,pre They represent the pre-planned x, y, and z coordinates at point t, respectively, and the average path deviation FPD is calculated. The calculation expression is: In the formula, M represents the number of key points;

[0103] The acceleration of the drone on the x, y, and z axes at the collection point is obtained as a x 、a y 、a z , and obtain the corresponding roll angle (Roll, φ), pitch angle (Pitch, θ) and yaw angle (Yaw, ψ), calculate the flight error value, the calculation expression is: In the formula, a x,c 、a y,c 、a z,c Respectively represent the acceleration of the drone on the x, y, and z axes at the collection point C, v is a positive integer, and the attitude angle difference ZT is calculated. The calculation expression is: In the formula, φ c ,θ c , c Respectively represent the roll angle, pitch angle and yaw angle at the acquisition point C;

[0104] Get the light intensity L at the collection point and establish the light intensity set: L c ={L 1 , L 2 , ..., L v}, calculate the light intensity difference between adjacent collection points as the rate of change: ΔL = L c+1 -L c , establish the variable rate set as: ΔL c ={ΔL 1 , ΔL 2 , ..., ΔL v-1}, calculate the rate of change of light intensity, The inspection adaptation stability index XJS is calculated and the calculation expression is:

[0105] It should be noted that the key points on the UAV's flight path are determined according to actual needs, such as the turns in the power pipeline corridor; flight stability describes the UAV's ability to maintain stability during flight, which is usually measured by the UAV's attitude changes (such as tilt angle) and vibration. The accelerometer and gyroscope on the UAV are used to collect the UAV's acceleration and angular velocity data in real time, and the UAV's attitude angle (roll angle, pitch angle and yaw angle) is calculated based on the gyroscope data.

[0106] The obtained signal-to-noise structure similarity index and patrol adaptation stability index are normalized and calculated simultaneously to obtain the patrol control coefficient, which is calibrated as K x , based on the formula: In the formula, XZJ is the signal-noise structure similarity index, XJS is the inspection adaptation stability index, α and β are the preset proportional coefficients of the signal-noise structure similarity index and the inspection adaptation stability index, respectively, and α and β are both greater than 0;

[0107] It should be noted that the formula is obtained by technicians in this field by collecting multiple groups of sample data and setting corresponding preset proportional coefficients for each group of sample data. The specific calculation process is: substituting the obtained signal-to-noise structure similarity index and inspection adaptation stability index into the formula to form an equation group, and then screening and averaging the calculated preset proportional coefficients to obtain the preset proportional coefficients. Since each group of sample data is different, the specific values ​​of the preset proportional coefficients under different sample data are also different.

[0108] It can be seen from the formula that the larger the signal-to-noise structure similarity index and the larger the inspection adaptation stability index, that is, the larger the value of the inspection control coefficient, the better the performance of the drone in collecting and measuring the relevant data of the power corridor during the inspection of the power corridor, and the data measured during the inspection of the power corridor has higher credibility and reference value;

[0109] The smaller the signal-to-noise structure similarity index and the smaller the inspection adaptation stability index, that is, the smaller the performance value of the inspection control coefficient, the worse the UAV's performance in collecting and measuring power pipeline corridor-related data during the inspection process. It is necessary to improve flight stability and image quality processing capabilities, and further optimize UAV flight parameters or image processing algorithms to improve inspection results.

[0110] The smaller the signal-to-noise structure similarity index is, the lower the signal-to-noise ratio of the image is, the poorer the structural similarity is, the poorer the image quality is, the more noise there is, the more blurry the image is, and the details and potential problems of the power corridor cannot be clearly displayed. The smaller the inspection adaptation stability index is, the larger the flight path deviation is, the poorer the flight stability is, and the lower the environmental adaptability is. The drone may experience greater vibrations during flight, deviate from the preset path, and it is difficult to collect data stably.

[0111] Due to poor image quality and flight stability, the data collection efficiency during the inspection process is low, and more time and resources may be required to complete the inspection task. There may be a high risk of omissions, and potential faults and problems in the power pipeline corridor cannot be discovered in time.

[0112] By normalizing the signal-to-noise structure similarity index and the inspection adaptation stability index and jointly calculating the inspection control coefficient, a comprehensive indicator can be obtained to quantify the overall performance of the UAV during the inspection process. The smaller the inspection control coefficient value, the worse the inspection performance of the UAV, and further optimization is needed to improve the inspection efficiency and data quality.

[0113] The generated inspection control coefficient is compared with the inspection strategy control threshold, and different signals are generated according to the comparison results, that is, an abnormal inspection measurement signal and a stable inspection measurement signal are generated;

[0114] After obtaining the generated inspection control coefficient, the generated inspection control coefficient is compared with the inspection strategy control threshold. If the inspection control coefficient is greater than the inspection strategy control threshold, a patrol measurement stability signal is generated, indicating that under the current circumstances, the drone performs well in the collection and measurement of relevant data during the inspection of the power corridor. The collected data has high credibility and reference value, and can be used as a basis for decision-making. The data collection efficiency is high, and the inspection task can be completed within the specified time, reducing unnecessary repeated inspections, improving resource utilization efficiency, and timely discovering potential faults and problems in the power corridor, reducing the probability of omission risks. Through high-quality data measurement, collection and analysis, preventive maintenance can be carried out, reducing the occurrence of sudden failures, and improving the safety and reliability of the power corridor.

[0115] The high inspection control coefficient verifies the effectiveness of the current inspection strategy. Inspections can continue to be performed according to the existing strategy without frequent adjustments to flight parameters and image processing algorithms, reducing the need for adjustments during the inspection process and maintaining an efficient and stable inspection process.

[0116] If the inspection control coefficient is less than or equal to the inspection strategy control threshold, an inspection measurement abnormality signal is generated, indicating that the UAV performs poorly in the current inspection process, the data collection and measurement quality is low, the inspection efficiency is low, and potential problems cannot be discovered in time. It is necessary to adjust and optimize the UAV inspection strategy to improve the inspection efficiency and strengthen fault detection and prevention to ensure the safe and efficient operation of the power pipeline corridor. That is, after the inspection measurement abnormality signal is generated, the following measures can be taken:

[0117] Retransmission: Dynamically adjust the flight altitude, speed and attitude of the drone based on real-time monitored environmental parameters (such as wind speed, humidity, temperature, etc.).

[0118] Adjust the inspection strategy. According to the real-time environmental monitoring data, dynamically adjust the inspection strategy to ensure efficient inspection under different environmental conditions. Set environmental parameter thresholds. When the environmental parameters exceed the set range, take preventive measures, such as adjusting the flight altitude or speed, or suspending the inspection.

[0119] Key point inspections plan the optimal flight path based on the layout of the power corridor and the location of key equipment to ensure focused inspections in key areas.

[0120] For some power corridor areas prone to problems, the following strategies can be optimized when drones conduct power corridor inspections to ensure the comprehensiveness and accuracy of the inspections, timely detect potential problems and take appropriate preventive measures. For areas prone to problems, such as connections, turns, pipeline intersections, etc., increase the inspection frequency to ensure that these areas receive more attention. For example, real-time monitoring of key areas can be carried out, and drones can be configured for scheduled or real-time flight missions to ensure that abnormal situations are discovered at the first time. When the drone passes by, more time is spent on inspections and related data measurements and image collection. For example, key areas can be photographed from multiple angles to ensure full coverage and avoid blind spots.

[0121] By taking the above measures, the quality and credibility of inspection data can be improved, resource utilization can be optimized, inspection costs can be reduced, and overall inspection efficiency can be improved.

[0122] It should be noted that the relevant threshold information in this embodiment is pre-set by professionals and will not be explained in detail here.

[0123] The present invention collects images of corresponding inspection areas during the inspection process, performs fog detection on the collected images, and screens out fogged images, performs defogging operation on the screened out fogged images using sub-block local overlapping histogram equalization, and uses maximum likelihood estimation theory to defuzzify the images to obtain the most realistic images. Finally, the inspection strategy adaptation information generated by the UAV during the inspection of the power corridor is obtained, the adaptation of the UAV to different power corridor areas during the inspection of the power corridor is determined, and different strategies are used to regulate the power corridor inspection, thereby improving the efficiency of the UAV power corridor inspection and reducing the probability of power failure.

[0124] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0125] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0126] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0127] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A power pipeline inspection method based on drones. It is characterized in that The steps include: Develop drone inspection strategies and routes based on the geographical location, structural complexity, and key inspection points of the pipeline corridor, and collect images of the corresponding inspection areas during the inspection process; Perform fog detection on the collected images, use the statistical features of the images in combination with the Canny edge detector to evaluate the image clarity and filter out the foggy images, and use sub-block local overlapping histogram equalization to perform defogging operations on the filtered foggy images; Use the maximum likelihood estimation theory to deblur the image to obtain the most realistic image, and then perform image defect detection after obtaining the most realistic image; Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power corridor, determine the adaptation of the inspection strategies in different areas and different power corridor areas during the inspection of the power corridor by the drone, and use different strategies to regulate the power corridor inspection according to the obtained adaptation results.

2. According to the method for inspecting a power pipeline corridor based on a drone according to claim 1, It is characterized in that Perform fog detection on the collected images, use the statistical features of the image combined with the Canny edge detector to evaluate the image clarity and filter out foggy images, including: Preprocess the image and convert it from RGB color space to HSV color space; Calculate the average brightness, image contrast and average saturation of the image in the statistical features of the image, and perform preliminary fog detection of the image based on the average brightness, image contrast and average saturation; After preliminary fog detection, the image is converted from RGB color space to grayscale color space, the horizontal and vertical gradients of the image grayscale are calculated, the edge with the largest gradient is retained, and the number of edge pixels is calculated as the number of edges detected in the image. The processed image is compared with the edge detection results of a clear and fog-free image to screen out the fogged image.

3. According to the method for inspecting a power pipeline corridor based on a drone as described in claim 2, It is characterized in that The filtered fogged image is defogged using sub-block local overlapping histogram equalization. The specific steps include: Calculate the histogram of the input fogged image, and calculate the gray value change table according to the histogram of the input image; Use the grayscale change table to perform a lookup table transformation operation, and traverse each pixel point, put the grayscale value of the original image into the exchange table, and obtain the new grayscale value of the pixel point; The sub-block size and overlapping area are determined using sub-block local overlapping histogram equalization, and a sliding window is used to traverse the entire image and reconstruct the image from all processed sub-blocks.

4. According to the method for inspecting a power pipeline corridor based on a drone as described in claim 3, It is characterized in that Use the maximum likelihood estimation theory to deblur the image to get the most realistic image. The specific steps are as follows: Establishing an image degradation model, establishing the image degradation model through the influence of a blur model or through the influence of noise, and estimating the original form of the image according to the image degradation model; In the process of image deblurring, maximum likelihood estimation is used to find the most similar parameter estimate of the blurred image as the most realistic image.

5. According to the method for inspecting a power pipeline corridor based on a drone as described in claim 4, It is characterized in that Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power corridor, and determine the adaptation of the inspection strategies of different areas and different power corridor areas during the inspection of the power corridor by the drone. The specific steps are as follows: Obtain the inspection strategy adaptation information generated by the drone during the inspection of the power pipeline corridor. The inspection strategy adaptation information includes data processing quality information and inspection performance impact information; Data processing quality information includes the signal-to-noise structure similarity index, and inspection performance impact information includes the inspection adaptation stability index: The signal-to-noise structure similarity index and the inspection adaptation stability index are combined to generate the inspection control coefficient; The signal-to-noise structure similarity index and the inspection adaptation stability index are proportional to the inspection control coefficient.

6. According to the method of claim 5, a power pipeline corridor inspection method based on drones, It is characterized in that According to the adaptation results, different strategies are used to regulate the inspection of the power pipeline corridor. The specific steps include: Compare the generated inspection control coefficient with the inspection strategy control threshold; If the inspection control coefficient is greater than the inspection strategy control threshold, a inspection measurement stability signal is generated without additional maintenance or processing; If the inspection control coefficient is less than or equal to the inspection strategy control threshold, an inspection measurement abnormality signal is generated to adjust and optimize the UAV inspection strategy.