Visual monitoring method for tension stringing construction of stretch pay-off electric engineering vehicle

Abstract

The invention discloses a visual monitoring method for tension stringing construction of a tension pay-off electric engineering vehicle. The method comprises the step that a video acquisition device on a vehicle-mounted tension device performs real-time acquisition on the stringing construction process of a high-voltage wire iron tower. According to the method, the images in the construction process are collected, and the construction images are denoised, enhanced and segmented, so that the iron tower frame power transmission line area can be better identified and positioned, and when the images are enhanced, the images are corrected, so that the images are in a fixed histogram state, the subsequent image analysis is in the same standard, and the accuracy of image analysis is improved. Based on a neural network model and through a mode of establishing coordinates in an image, an iron tower frame inclination angle, a height, a power transmission line-to-ground included angle and power transmission line swing amplitude data are obtained, a construction parameter set is formed and input into a power transmission line risk prediction model, so that a power transmission line erection risk value can be obtained, and an alarm is automatically given. And quickly arranging power transmission line construction rectification.

Description

Technical Field

[0001] The invention generally relates to the technical field of construction monitoring, and in particular to a method for visualizing the construction of tension wire stringing by a tensioning and paying-out electric power engineering vehicle. Background Art

[0002] With the rapid growth of electricity demand, my country's power industry has also made great progress. The contradiction between transmission capacity and line tension has begun to ease with the emergence of split conductors. Building energy-saving and consumption-reducing equipment and improving the transmission capacity of unit line corridors are of great significance to building a "resource-saving and environmentally friendly" power grid.

[0003] During the construction of high-voltage line towers, the tension of the transmission lines is related to parameters such as the tower inclination, height, angle between the transmission line and the ground, and the swing amplitude of the transmission line. However, most systems in real life cannot accurately obtain these parameters from images, and cannot accurately analyze whether there are risks in the laid transmission lines, making it difficult to meet the needs of staff. Summary of the invention

[0004] In order to solve the above technical problems, a visual monitoring method for tension line construction of a tensioning and wire-laying electric engineering vehicle is provided. This technical solution solves the problem proposed in the above background technology that during the construction of high-voltage line towers, the tension of the transmission line is related to parameters such as the inclination angle, height, angle between the transmission line and the ground, and the swing amplitude of the transmission line. However, most systems in real life cannot accurately obtain these parameters from images, and cannot accurately analyze whether there are risks in the laid transmission lines.

[0005] The present invention provides a visual monitoring method for tension wire construction of a tension wire electric engineering vehicle. The method comprises:

[0006] The video acquisition device on the vehicle-mounted tensioning equipment collects the high-voltage line tower stringing construction process in real time to obtain construction images;

[0007] Preprocessing the construction image, wherein the preprocessing includes image denoising, image enhancement and image segmentation;

[0008] Analyze the preprocessed image to obtain the tower inclination data, height data, transmission line-to-ground angle data, and transmission line swing amplitude data in the image, and form a real-time construction parameter set;

[0009] Retrieve the transmission line risk prediction model pre-set in the database, input the real-time construction parameter set into the transmission line risk prediction model, and obtain the transmission line erection risk value;

[0010] Based on the risk value of the transmission line installation, it is determined whether to issue an alarm to the construction personnel.

[0011] Preferably, the image denoising specifically comprises the following steps:

[0012] Decompose high-frequency and low-frequency information in construction images;

[0013] The pixels of low-frequency information are processed by local pixel grouping, and the unbiased estimation of the error is used to approximate the similarity between the local pixel block and the target pixel block, and a sample set of similar local pixel blocks is obtained;

[0014] Traverse each sample set obtained, use the principal component analysis algorithm to denoise it in turn, calculate the covariance matrix, obtain the orthogonal transformation matrix, and combine it with the eigenvalue matrix to remove the dimensions containing a small amount of information in the sample set to obtain the reconstructed low-frequency components;

[0015] Decompose high-frequency information into overlapping blocks of the same size, calculate the Euclidean distance to construct similar blocks into groups, use singular value decomposition to learn the adaptive learning dictionary of each group, calculate sparse coding through the split Bregman iterative algorithm combined with convex optimization algorithm, and reconstruct the high-frequency components using sparse coding and adaptive learning dictionary;

[0016] The inverse wavelet transform aggregates the high-frequency components and the low-frequency components to obtain the denoised image.

[0017] Preferably, the image enhancement specifically comprises the following steps:

[0018] Perform adaptive histogram equalization on the denoised image;

[0019] Determine the cumulative distribution function of the local area by calculating the grayscale histogram of the denoised image;

[0020] Based on the cumulative distribution function, the grayscale mapping relationship of the local area is adaptively adjusted to obtain an enhanced image.

[0021] Preferably, the image segmentation specifically comprises the following steps:

[0022] Segmenting the enhanced image into multiple superpixels using a superpixel segmentation algorithm;

[0023] Determine the similarity matrix between superpixels as the edge weight of the undirected graph, and construct a superpixel graph cut undirected graph;

[0024] Combined with the pre-set probability distribution of the iron tower transmission line area, the global optimal solution of the superpixel graph cut undirected graph is solved by the maximum flow minimum cut algorithm;

[0025] The segmentation result of the iron tower transmission line area is obtained to obtain the pre-processed image.

[0026] Preferably, the step of performing adaptive histogram equalization on the denoised image specifically comprises the following steps:

[0027] Perform edge detection on the image in the horizontal and vertical directions, and sum the detection results in the horizontal and vertical directions;

[0028] The barcode part of the image is processed based on the first morphological closing operation to obtain the connected area, and then the single line interference on the image is removed by the second morphological closing operation;

[0029] Filling holes in the image processed by the first morphological closing operation;

[0030] Calculate the minimum outer rectangle according to each white area in the image after filling the holes, and obtain information of each outer rectangle, wherein the rectangle information includes the coordinates of the four vertices of the rectangle, the tilt angle, and the number of white pixels in the rectangle area;

[0031] According to the calculated rectangular tilt angles, their distribution within the right angle range is counted, and the angles with more concentrated distribution are used to perform preliminary tilt angle correction.

[0032] Preferably, in the image after the preliminary tilt angle correction, the transmission line area of ​​the pylon is located in the 0 degree direction or the 90 degree direction, and the error is within the range of plus or minus 10 degrees;

[0033] Screen candidate rectangles of the iron tower transmission line area in the directions of 0 degrees and 90 degrees respectively to determine the final position and final angle of the iron tower transmission line area;

[0034] Perform a final rotation to obtain the final rectified image of the pylon transmission line area.

[0035] Preferably, the calculation formula of the minimum outer bounding rectangle is:

[0036] S min =min(S1,S2,.....S n )

[0037] In the formula, S min It is the smallest outer rectangle;

[0038] S1,S2,.....S n is the area of ​​a number of enclosing rectangles.

[0039] Preferably, the step of analyzing the preprocessed image to obtain the tower inclination data, height data, transmission line-to-ground angle data, and transmission line swing amplitude data in the image specifically comprises the following steps:

[0040] Mark the area of ​​the iron tower transmission line in rows;

[0041] The labeled tower transmission line areas are used as training sets, and the target detection model is trained until convergence.

[0042] The segmented iron tower transmission line area blocks are put into the trained target detection model to perform target detection on the iron tower transmission line area, and the coordinates of the iron tower transmission line area in each image block are obtained and output;

[0043] Based on the coordinates of the iron tower transmission line area, the iron tower inclination data, height data, transmission line-to-ground angle data and transmission line swing amplitude data in the iron tower transmission line area are obtained.

[0044] Preferably, the pre-set transmission line risk prediction model is:

[0045]

[0046] In the formula, S i is the similarity between the real-time construction parameter set and the i-th feature of the historical construction parameter set before the risk occurs, w ij is the jth characteristic index value of the ith characteristic of the real-time construction parameter set, v ij is the jth characteristic index value of the ith characteristic of the historical construction parameter set before the risk occurs, and n is the total number of ith characteristic index values.

[0047] Preferably, the step of determining whether to issue an alarm to construction personnel based on the risk value of the transmission line installation specifically includes the following steps:

[0048] Compare the risk value of transmission line installation with the pre-set threshold in the system;

[0049] If the risk value of the transmission line installation is greater than or equal to the preset threshold, it means that there is a high risk in the transmission line installation, and an alarm is issued to the construction personnel;

[0050] If the risk value of the transmission line installation is less than the preset threshold, it means that there is a small risk in the transmission line installation and there is no need to issue an alarm.

[0051] Compared with the prior art, the present invention provides a visual monitoring method for tension wire construction of a tension wire electric engineering vehicle, which has the following beneficial effects:

[0052] The present invention collects images during the construction process and removes noise in the construction images to improve the accuracy of subsequent identification of iron tower transmission lines. In order to enhance the contrast of the image, improve the image clarity and improve the effect of the iron tower transmission lines, the denoised image is enhanced and the image is divided into different areas to better identify and locate the iron tower transmission line area. When enhancing the image, the present invention also introduces correction of the image to make it in a fixed histogram state so that subsequent image analysis is in the same standard. Based on the neural network model and by establishing coordinates in the image, the inclination angle, height of the iron tower, the angle between the transmission line and the ground, and the swing amplitude data of the transmission line are obtained to form a construction parameter set, and the set of construction parameters is input into a transmission line risk prediction model set in advance by experts. The model is made by experts through a large amount of parameter data when sending transmission line risks, so that the risk value of the transmission line erection can be obtained, and an alarm can be automatically issued to quickly arrange the rectification of the transmission line construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The above and other features, advantages and aspects of the embodiments of the present invention will become more apparent with reference to the following detailed description in conjunction with the accompanying drawings. In the accompanying drawings, the same or similar reference numerals represent the same or similar elements, wherein:

[0054] Figure 1 It is a schematic diagram of the transmission line tension stringing construction monitoring method in the present invention;

[0055] Figure 2 A schematic diagram of the method for image denoising in the present invention;

[0056] Figure 3 Schematic diagram of the image enhancement method in the present invention;

[0057] Figure 4 It is a schematic diagram of the method for image segmentation in the present invention;

[0058] Figure 5 A schematic diagram of a method for performing adaptive histogram equalization processing on a denoised image in the present invention;

[0059] Figure 6 It is a schematic diagram of a method for analyzing a pre-processed image in the present invention;

[0060] Figure 7 It is a schematic diagram of the method for judging whether to send an alarm to the construction personnel in the present invention. DETAILED DESCRIPTION

[0061] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] In addition, the term "and / or" in this article is only a description of the association relationship between the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship.

[0063] Example 1

[0064] Please refer to Figure 1-Figure 6 As shown, the visual monitoring method for tensioning and laying out electric power engineering vehicle tensioning and laying out wire construction comprises:

[0065] S100, the video acquisition device on the vehicle-mounted tensioning device collects the high-voltage line tower stringing construction process in real time to obtain construction images;

[0066] S200, preprocessing the construction image, the preprocessing includes image denoising, image enhancement and image segmentation;

[0067] S300, analyzing the pre-processed image, obtaining the tower inclination data, height data, transmission line-to-ground angle data, and transmission line swing amplitude data in the image, and forming a real-time construction parameter set;

[0068] S400, calling a transmission line risk prediction model preset in a database, inputting a real-time construction parameter set into the transmission line risk prediction model, and obtaining a transmission line erection risk value;

[0069] S500: Based on the risk value of the transmission line installation, determine whether to issue an alarm to the construction personnel.

[0070] It can be understood by those skilled in the art that during the construction of high-voltage line towers, the transmission line tension is related to parameters such as the tower inclination, height, angle between the transmission line and the ground, and swing amplitude of the transmission line. By collecting images during the construction process and removing noise from the construction images, the accuracy of subsequent identification of the tower transmission line is improved. In order to enhance the contrast of the image and improve the image clarity to improve the effect of the tower transmission line, the denoised image is enhanced and the image is divided into different areas to better identify and locate the tower transmission line area. When enhancing the image, the present invention also introduces The image is corrected to be in a fixed histogram state so that subsequent image analysis is in the same standard. Based on the neural network model and by establishing coordinates in the image, the tower inclination, height, transmission line angle to the ground and transmission line swing amplitude data are obtained to form a set of construction parameters. The parameters are input into the transmission line risk prediction model set in advance by experts. The model is made by experts based on a large amount of parameter data when sending transmission line risks. The risk value of transmission line installation can be obtained, automatic alarms can be issued, and transmission line construction rectification can be quickly arranged to meet the needs of staff.

[0071] S210, image denoising specifically includes the following steps:

[0072] S211, decomposing high-frequency information and low-frequency information in the construction image;

[0073] S212, performing local pixel grouping processing on the pixel points of the low-frequency information, using an unbiased estimation of the error to approximately represent the similarity between the local pixel block and the target pixel block, and obtaining a sample set of similar local pixel blocks;

[0074] S213, traversing each sample set obtained, using the principal component analysis algorithm to perform denoising in turn, calculating the covariance matrix to obtain the orthogonal transformation matrix, combining the eigenvalue matrix to remove the dimensions containing a small amount of information in the sample set, and obtaining the reconstructed low-frequency components;

[0075] S214, decomposing the high-frequency information into overlapping blocks of the same size, calculating the Euclidean distance to construct similar blocks into groups, using singular value decomposition to learn an adaptive learning dictionary for each group, calculating sparse coding by a split Bregman iterative algorithm combined with a convex optimization algorithm, and reconstructing the high-frequency components by using sparse coding and an adaptive learning dictionary;

[0076] S215, performing inverse wavelet transform to aggregate high-frequency components and low-frequency components to obtain a denoised image.

[0077] It is understood by those skilled in the art that the present invention combines multi-scale analysis, local pixel grouping, principal component analysis (PCA), singular value decomposition (SVD), sparse coding and wavelet transform technology during denoising, which can effectively remove noise in the image while retaining the main structure and detail information of the image. Through local pixel grouping and adaptive learning dictionary, complex structures and textures in the image can be processed more accurately to improve the denoising effect.

[0078] S220, image enhancement specifically includes the following steps:

[0079] S221, performing adaptive histogram equalization processing on the denoised image;

[0080] S222, determining the cumulative distribution function of the local area by calculating the grayscale histogram of the denoised image;

[0081] S223. Based on the cumulative distribution function, adaptively adjust the grayscale mapping relationship of the local area to obtain an enhanced image.

[0082] It is understandable to those skilled in the art that the denoised image is not directly used for global adjustment in AHE or CLAHE, but is used to understand the overall distribution of the image grayscale. The image is divided into multiple small, overlapping local areas (usually referred to as "context windows" or "blocks"), and the cumulative distribution function (CDF) is calculated based on the grayscale histogram of each local area. The CDF represents the ratio of the number of pixels from the minimum grayscale value to the current grayscale value to the total number of pixels. The CDF of each local area is used to remap the grayscale value in the area. The goal is to stretch the grayscale range in the local area so that the contrast is enhanced, and after all local areas are processed, they are recombined into a complete image. Since each local area is independently grayscale mapped, the contrast of the entire image is adaptively enhanced.

[0083] S230, image segmentation specifically includes the following steps:

[0084] S231, segmenting the enhanced image into multiple superpixels using a superpixel segmentation algorithm;

[0085] S232, determining the similarity matrix between superpixels as the edge weight of the undirected graph, and constructing a superpixel graph cut undirected graph;

[0086] S233, combining the pre-set probability distribution of the iron tower transmission line area, solving the global optimal solution of the superpixel graph cut undirected graph by using the maximum flow minimum cut algorithm;

[0087] S234, obtaining the segmentation result of the iron tower transmission line area and obtaining a preprocessed image.

[0088] It can be understood by those skilled in the art that the present invention is used to segment the pylon and transmission line area from the enhanced image through superpixel segmentation, graph cut algorithm and pre-set probability distribution, and the enhanced image is segmented into multiple superpixels using superpixel segmentation algorithm (such as SLIC, LSC, ERS, etc.). These superpixels are a set of pixels with similar features such as color and texture in the image. They serve as the basic unit for subsequent processing, and each superpixel is regarded as a node of an undirected graph to determine the similarity matrix between superpixels, which is usually based on the similarity measurement of features such as color, texture, and position. The value of the similarity matrix will be used as the edge weight of the undirected graph to construct an undirected graph, in which the node represents the superpixel, the edge represents the connection between superpixels, and the edge weight represents the similarity; a probability distribution is set based on the prior knowledge of the pylon and transmission line area. This distribution may be based on features such as shape, color, and position, and is used to guide the decision of the graph cut algorithm during the segmentation process.

[0089] Use the maximum flow minimum cut algorithm (such as Ford-Fulkerson algorithm, Edmonds-Karp algorithm, etc.) to solve the global optimal solution on the undirected graph. This optimal solution corresponds to a cut that divides the undirected graph into two subsets so that the sum of the edge weights between the two subsets (that is, the capacity of the cut) is minimized.

[0090] According to the output of the maximum flow minimum cut algorithm, the subset to which each superpixel belongs (i.e., the pylon transmission line area or the background) is determined, and the superpixels belonging to the pylon transmission line area are merged to obtain the final segmentation result. Based on the segmentation result, the pylon and transmission line areas are extracted.

[0091] S2210, performing adaptive histogram equalization processing on the denoised image specifically includes the following steps:

[0092] S2211, performing edge detection on the image in the horizontal and vertical directions, and summing the detection results in the horizontal and vertical directions;

[0093] S2212, processing the barcode portion of the image based on the first morphological closing operation to obtain a connected area, and then using the second morphological closing operation to remove single line interference on the image;

[0094] S2213, filling holes in the image processed by the first morphological closing operation;

[0095] S2214, calculating the minimum outer rectangle according to each white area in the image after the holes are filled, and obtaining information of each outer rectangle, the rectangle information including the coordinates of the four vertices of the rectangle, the tilt angle and the number of white pixels in the rectangular area;

[0096] S2215, counting the distribution of the calculated rectangular tilt angles within the right angle range, and using the angles with relatively concentrated distribution to perform preliminary tilt angle correction;

[0097] S2216. In the image after the preliminary tilt angle correction, the transmission line area of ​​the iron tower is located in the 0 degree direction or the 90 degree direction, and the error is within the range of plus or minus 10 degrees;

[0098] S2217, screening candidate rectangles of the iron tower transmission line area in the directions of 0 degrees and 90 degrees respectively, and determining the final position and final angle of the iron tower transmission line area;

[0099] S2218. Perform a final rotation to obtain the final corrected image of the tower transmission line area.

[0100] The calculation formula for the minimum enclosing rectangle is:

[0101] S min =min(S1,S2,.....S n )

[0102] In the formula, S min It is the smallest outer rectangle;

[0103] S1,S2,.....S n is the area of ​​a number of enclosing rectangles.

[0104] Analyzing the preprocessed image to obtain the tower inclination data, height data, transmission line-to-ground angle data, and transmission line swing amplitude data in the image specifically includes the following steps:

[0105] S301, marking the area of ​​the iron tower transmission line in units of lines;

[0106] S302, forming a training set with the marked iron tower transmission line area, and training the target detection model until convergence;

[0107] S303, putting the segmented iron tower transmission line area blocks into the trained target detection model to perform target detection on the iron tower transmission line area, and obtaining the coordinates of the iron tower transmission line area in each image block and outputting them;

[0108] S304, based on the coordinates of the iron tower transmission line area, obtaining the iron tower inclination data, height data, transmission line-to-ground angle data and transmission line swing amplitude data in the iron tower transmission line area.

[0109] It can be understood by those skilled in the art that the target detection model in deep learning, such as the SSD model, is used to detect the target of the pylon transmission line area on the image block to obtain the coordinates of the pylon transmission line area in each image block. This embodiment uses Inception V2 as the base model of the SSD model. When selecting the base model, other models such as Mobile Net can also be selected. Of course, the deep detection of the target is not limited to the SSD detection model listed in this embodiment, and detection models such as Faster R-CNN, R-FCN, etc. can also be used;

[0110] First, prepare some license plate pictures for marking the pylon transmission line area, preferably in lines, and use the mouse to swipe the license plate picture from the upper left corner to the lower right corner, so that the rectangle generated with the current diagonal can completely contain the current pylon transmission line area, and record the coordinates of the upper left corner of the current rectangle (xmin, ymin) and the coordinates of the lower right corner (xmax, ymax) and the text contained therein as the collection information of the current area, and form a training set, and then train it through the SSD model until convergence, and then perform image preprocessing and subsequent area segmentation on the pictures to be detected, and then use the target detection model to detect the pylon transmission line area on these picture blocks, and derive the coordinates of the detected pylon transmission line area;

[0111] After obtaining the coordinates of the pylon transmission line area in each image block, the pylon transmission line area needs to be merged, including: processing the M image blocks from bottom to top and from right to left according to the relative position of each image block originally located in the license plate image, which is the reverse order of the original segmentation order, and connecting the coordinates of the detected pylon transmission line area based on the overlapping area of ​​N pixels in the two adjacent image blocks on the left and top of the current image block to obtain the coordinates of the complete pylon transmission line area;

[0112] The fusion process specifically includes the following:

[0113] (1) Based on the area of ​​50 pixels from the top to the bottom of the current picture block, i.e., b5,6[0:50,:,:], and the area of ​​50 pixels from the bottom to the top of the adjacent upper picture block, b4,6[450:500,:,:], if a certain iron tower transmission line area in the area of ​​the current picture block intersects with a certain iron tower transmission line area in the area of ​​the upper picture block, then the iron tower transmission line area in the current picture block is integrated into the iron tower transmission line area corresponding to the upper picture block and the iron tower transmission line area in the current picture block is deleted;

[0114] (2) Based on the area of ​​50 pixels from left to right of the current image block, i.e., b5,6[:,0:50,:], and the area of ​​50 pixels from right to left of the adjacent left image block, b5,5[:,450:500,:], if a certain pylon transmission line area in the area of ​​the current image block intersects with a certain pylon transmission line area in the area of ​​the left image block, then the pylon transmission line area in the current image block is integrated into the pylon transmission line area corresponding to the left image block and the pylon transmission line area in the current image block is deleted;

[0115] (3) Based on the relative position of the current image block, with the lower left corner of the current image block as the center, select the overlapping areas in the four image blocks, namely the current b5,6[450:500,:,:], the adjacent left b5,5[450:500,:,:], the adjacent bottom b6,6[0:50,:,:], and the adjacent lower left b6,5[0:50,:,:], and select all areas in these four image blocks where the tower transmission line area intersects with the corresponding area. Among all the areas, if one area is completely within another area, delete the smaller one;

[0116] Based on the area of ​​50 pixels from bottom to top of the current picture block, i.e., b5,6[450:500,:,:], and the area of ​​50 pixels from top to bottom of the adjacent lower picture block, b6,6[0:50,:,:], if a certain iron tower transmission line area in this area of ​​the current picture block intersects with a certain iron tower transmission line area in this area of ​​the lower picture block, then the iron tower transmission line area in the current picture block is integrated into the iron tower transmission line area corresponding to the lower picture block and the iron tower transmission line area in the current picture block is deleted.

[0117] Finally, after the above fusion and deletion operations, the iron tower transmission line areas in all image blocks are collected as the complete iron tower transmission line areas in the original bill image.

[0118] The pre-set transmission line risk prediction model is:

[0119]

[0120] In the formula, S i is the similarity between the real-time construction parameter set and the i-th feature of the historical construction parameter set before the risk occurs, w ij is the jth characteristic index value of the ith characteristic of the real-time construction parameter set, v ij is the jth characteristic index value of the ith characteristic of the historical construction parameter set before the risk occurs, and n is the total number of ith characteristic index values.

[0121] Based on the risk value of the transmission line installation, the following steps are used to determine whether to issue an alarm to the construction personnel:

[0122] S501, comparing the risk value of the transmission line installation with a threshold value pre-set in the system;

[0123] S502: If the risk value of the transmission line installation is greater than or equal to a preset threshold, it indicates that there is a high risk in the transmission line installation, and an alarm is issued to the construction personnel;

[0124] S503: If the risk value of the transmission line installation is less than a preset threshold, it means that there is a small risk in the transmission line installation and there is no need to issue an alarm.

[0125] In summary, by collecting images during the construction process and removing noise from the construction images, the accuracy of subsequent identification of iron tower transmission lines is improved. In order to enhance the contrast of the image and improve the image clarity to enhance the effect of the iron tower transmission lines, the denoised image is enhanced and the image is divided into different areas to better identify and locate the iron tower transmission line areas. When enhancing the image, the present invention also introduces correction of the image to make it in a fixed histogram state so that subsequent image analysis is in the same standard. Based on the neural network model and by establishing coordinates in the image, the inclination angle, height, transmission line angle to the ground and transmission line swing amplitude data of the iron tower are obtained to form a construction parameter set, which is input into the transmission line risk prediction model set in advance by the expert. The model is made by the expert through a large amount of parameter data when sending the transmission line risk, so that the risk value of the transmission line erection can be obtained, and the alarm can be automatically issued to quickly arrange the transmission line construction rectification.

[0126] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A visual monitoring method for tension wire construction of a power engineering vehicle, characterized in that: include: The video acquisition device on the vehicle-mounted tensioning equipment collects the high-voltage line tower stringing construction process in real time to obtain construction images; Preprocessing the construction image, wherein the preprocessing includes image denoising, image enhancement and image segmentation; Analyze the preprocessed image to obtain the tower inclination data, height data, transmission line-to-ground angle data, and transmission line swing amplitude data in the image, and form a real-time construction parameter set; Retrieve the transmission line risk prediction model pre-set in the database, input the real-time construction parameter set into the transmission line risk prediction model, and obtain the transmission line erection risk value; Based on the risk value of the transmission line installation, it is determined whether to issue an alarm to the construction workers.

2. The visual monitoring method for tension wire construction of a tension wire electric engineering vehicle according to claim 1 is characterized in that: The image denoising comprises: Decompose high-frequency and low-frequency information in construction images; The pixels of low-frequency information are processed by local pixel grouping, and the unbiased estimation of the error is used to approximate the similarity between the local pixel block and the target pixel block, and a sample set of similar local pixel blocks is obtained; Traverse each sample set obtained, use the principal component analysis algorithm to denoise it in turn, calculate the covariance matrix, obtain the orthogonal transformation matrix, and combine it with the eigenvalue matrix to remove the dimensions containing a small amount of information in the sample set to obtain the reconstructed low-frequency components; Decompose high-frequency information into overlapping blocks of the same size, calculate the Euclidean distance to construct similar blocks into groups, use singular value decomposition to learn the adaptive learning dictionary of each group, calculate sparse coding through the split Bregman iterative algorithm combined with convex optimization algorithm, and reconstruct the high-frequency components using sparse coding and adaptive learning dictionary; The inverse wavelet transform aggregates the high-frequency components and the low-frequency components to obtain the denoised image.

3. The visual monitoring method for tension wire construction of a tension-and-release electric power engineering vehicle according to claim 2 is characterized in that: The image enhancement comprises: Perform adaptive histogram equalization on the denoised image; Determine the cumulative distribution function of the local area by calculating the grayscale histogram of the denoised image; Based on the cumulative distribution function, the grayscale mapping relationship of the local area is adaptively adjusted to obtain an enhanced image.

4. The visual monitoring method for tension wire construction of a tension-release electric power engineering vehicle according to claim 3 is characterized in that: The image segmentation comprises: Segmenting the enhanced image into multiple superpixels using a superpixel segmentation algorithm; Determine the similarity matrix between superpixels as the edge weight of the undirected graph, and construct a superpixel graph cut undirected graph; Combined with the pre-set probability distribution of the iron tower transmission line area, the global optimal solution of the superpixel graph cut undirected graph is solved by the maximum flow minimum cut algorithm; The segmentation result of the iron tower transmission line area is obtained to obtain the pre-processed image.

5. The visual monitoring method for tension wire construction of a tension-and-payout electric power engineering vehicle according to claim 3 is characterized in that: The step of performing adaptive histogram equalization processing on the denoised image comprises: Perform edge detection on the image in the horizontal and vertical directions, and sum the detection results in the horizontal and vertical directions; The barcode part of the image is processed based on the first morphological closing operation to obtain the connected area, and then the single line interference on the image is removed by the second morphological closing operation; Filling holes in the image processed by the first morphological closing operation; Calculate the minimum outer rectangle according to each white area in the image after filling the holes, and obtain information of each outer rectangle, wherein the rectangle information includes the coordinates of the four vertices of the rectangle, the tilt angle, and the number of white pixels in the rectangle area; According to the calculated rectangular tilt angles, their distribution within the right angle range is counted, and the angles with more concentrated distribution are used to perform preliminary tilt angle correction.

6. The visual monitoring method for tension wire construction of a tension wire-laying electric power engineering vehicle according to claim 5 is characterized in that: In the image after the initial tilt angle correction, the transmission line area of ​​the pylon is located in the 0 degree direction or the 90 degree direction, and the error is within the range of plus or minus 10 degrees; Screen candidate rectangles of the iron tower transmission line area in the directions of 0 degrees and 90 degrees respectively to determine the final position and final angle of the iron tower transmission line area; Perform a final rotation to obtain the final rectified image of the pylon transmission line area.

7. The visual monitoring method for tension wire construction of a tension-and-payout electric power engineering vehicle according to claim 5 is characterized in that: The minimum enclosing rectangle is: S min =min(S1,S2,.....S n ) In the formula, S min It is the smallest outer rectangle; S1,S2,.....S n is the area of ​​a number of enclosing rectangles.

8. The visual monitoring method for tension wire construction of a tension-and-payout electric power engineering vehicle according to claim 7 is characterized in that: The pre-processed image is analyzed to obtain the tower inclination data, height data, transmission line-to-ground angle data and transmission line swing amplitude data in the image, including: Mark the area of ​​the iron tower transmission line in rows; The labeled tower transmission line areas are used as training sets, and the target detection model is trained until convergence. The segmented iron tower transmission line area blocks are put into the trained target detection model to perform target detection on the iron tower transmission line area, and the coordinates of the iron tower transmission line area in each image block are obtained and output; Based on the coordinates of the iron tower transmission line area, the iron tower inclination data, height data, transmission line-to-ground angle data and transmission line swing amplitude data in the iron tower transmission line area are obtained.

9. The visual monitoring method for tension wire construction of a tension-and-payout electric power engineering vehicle according to claim 8 is characterized in that: The pre-set transmission line risk prediction model is: In the formula, S i is the similarity between the real-time construction parameter set and the i-th feature of the historical construction parameter set before the risk occurs, w ij is the jth characteristic index value of the i-th characteristic of the real-time construction parameter set, v ij is the jth characteristic index value of the ith characteristic of the historical construction parameter set before the risk occurs, and n is the total number of ith characteristic index values.

10. The visual monitoring method for tension wire stringing construction of a tension wire-laying electric power engineering vehicle according to claim 9, characterized in that: The step of determining whether to send an alarm to construction personnel based on the risk value of the transmission line installation includes: Compare the risk value of transmission line installation with the pre-set threshold in the system; If the risk value of the transmission line installation is greater than or equal to the preset threshold, it means that there is a high risk in the transmission line installation, and an alarm is issued to the construction personnel; If the risk value of the transmission line installation is less than the preset threshold, it means that there is a small risk in the transmission line installation and there is no need to issue an alarm.

Images