A transmission line conductor ice thickness detection system and method

Abstract

The present invention relates to a transmission line conductor ice thickness detection system and method, the method collects and uploads images and video data of on-site transmission line conductors through an image acquisition terminal, and the ice detection and early warning system detects the ice thickness of the conductors according to the following steps: 1) preprocessing the image; 2) using the Canny edge detection algorithm to extract the edge in the image; 3) using the Hough line detection algorithm to obtain the conductor edge sub-segments; 4) using the improved K-means algorithm to determine the edge to which each sub-segment belongs; 5) fitting to obtain a complete and continuous conductor edge; 6) based on a monocular vision distance mapping model, mapping the conductor pixel width to the actual width; 7) comparing the current actual conductor thickness value with the un-iced conductor thickness value, calculating the conductor ice thickness value, and issuing an early warning if the ice thickness threshold is exceeded. The system and method are conducive to accurately and conveniently detecting the ice condition of the transmission line conductors.

Description

Technical Field

[0001] The present invention belongs to the technical field of power transmission line detection, and in particular relates to a system and method for detecting ice thickness of a transmission line conductor. Background Art

[0002] Transmission lines in mountainous or high-altitude areas often experience ice in winter. This phenomenon mainly depends on the humidity, temperature, wind speed, wind direction and other environmental conditions. In the cold winter environment, when the ambient temperature and humidity reach a certain level, the transmission lines will be covered with ice. If de-icing measures are not taken in time, the thickness of the ice layer on the conductor will gradually accumulate. When the thickness exceeds the designed anti-icing capacity, line tripping and line breaking accidents may occur. In severe cases, the tower may even collapse, causing large-scale power outages and huge economic losses. When the conductor is covered with ice, the insulation distance between the conductors or between the conductors and the ground will be reduced, and a short circuit will occur, endangering the safe and stable operation of the power system. Ice disasters often occur under harsh climatic conditions, and are mostly in high-altitude and mountainous areas. It is very difficult and risky to observe them manually on site. Setting up an ice observation station can reduce the difficulty and risk of observation to a certain extent, but this method has certain errors and requires observers to go deep into ice and snow covered areas, which poses a great safety hazard.

[0003] The traditional method of detecting the ice thickness of wires based on mechanical models requires the establishment of a complex model. The acquired data must strictly meet the timing requirements and is easily affected by environmental changes. The ice detection method based on a single machine vision algorithm is not accurate enough, the extracted wire edges are not smooth, and cannot effectively reflect the wire status. When classifying the attribution of wire edge segments, simple classification or clustering algorithms are not accurate enough, and complex algorithms have the defect of poor timeliness. There is no algorithm designed specifically for wire edge classification. In terms of calculating the true thickness of wires, the commonly used method is the linear mapping method, which converts the pixel width of the wire into the actual distance in direct proportion. Since the width of the near-end wire and the far-end wire in the image is the same, but visually due to the "near big and far small" effect, the width of the near-end and far-end wires is different, which is inconsistent with the actual situation and has a large error.

[0004] It can be seen that the problem of ice detection on transmission lines still needs to be solved urgently and has high practical value. Summary of the invention

[0005] The object of the present invention is to provide a system and method for detecting ice thickness of transmission line conductors, which are conducive to accurately and conveniently detecting the ice coating condition of transmission line conductors.

[0006] To achieve the above-mentioned purpose, the technical solution adopted by the present invention is: a transmission line conductor ice thickness detection system, including an image acquisition terminal and an ice detection and early warning system, the image acquisition terminal includes a camera unit, an image processing unit, a communication unit and a power supply unit, the camera unit is installed on the on-site tower and the shooting direction is facing the wires erected on the transmission line, so as to take real-time photos and video monitoring of the transmission line wires, the camera unit is connected to the image processing unit to transmit the collected image and video data to the image processing unit for processing, the image processing unit is connected to the communication unit to send the processed image and video data to the ice detection and early warning system through wireless communication; the power supply unit supplies power to each unit;

[0007] The ice detection and early warning system includes a communication and analysis unit, a database unit, a historical data query unit, a parameter setting unit, a thickness calculation unit, a display unit and an early warning unit. The communication and analysis unit communicates with the image acquisition terminal to receive the image and video data uploaded by the terminal. The database unit includes a real-time database and a historical database for storing the collected data. The historical data query unit provides the historical ice thickness information of the line as needed. The parameter setting unit sets the conductor parameters and environmental parameters to calculate the ice thickness of the conductor. The thickness calculation unit calculates the thickness value of the current conductor based on the collected image and video data, and calculates the ice thickness value of the conductor by comparing it with the thickness value of the un-iced conductor. The display unit is used to display the collected data and the calculation result of the ice thickness of the conductor. The early warning unit is used to compare the calculated ice thickness of the transmission line conductor with the line design bearing capacity threshold. If the threshold is exceeded, an alarm message is issued to the operation and maintenance personnel.

[0008] Furthermore, the power supply unit adopts solar energy storage to supply power, mainly including solar panels, charge and discharge controllers, battery packs and POE power supply modules. Solar energy is used for power supply during the day, and excess electric energy is stored in the battery pack. The battery pack is used for power supply at night.

[0009] The present invention also provides a method for detecting ice thickness of a transmission line conductor. The image and video data of the transmission line conductor on site are collected and uploaded by an image acquisition terminal. The ice detection and early warning system detects the ice thickness of the conductor according to the following steps based on the received image data:

[0010] 1) Preprocess the image;

[0011] 2) The Canny edge detection algorithm is used to extract the edges in the image. The extracted edges include the edges of the wires and a large number of interfering line segments;

[0012] 3) Using Hough line detection algorithm to detect straight line segments in the image, and obtain wire edge sub-segments;

[0013] 4) Using the improved K-means algorithm to cluster the wire edge sub-segments and determine the edge to which each sub-segment belongs;

[0014] 5) Using the least square method to fit the obtained wire edge, a complete and continuous wire edge is obtained;

[0015] 6) Based on the monocular vision distance mapping model, the pixel width of the wire is mapped to the actual width, that is, the actual thickness value of the current wire;

[0016] 7) Compare the calculated actual thickness value of the current conductor with the thickness value of the un-iced conductor to calculate the conductor ice thickness value. If the conductor ice thickness value exceeds the ice thickness threshold, an early warning is issued.

[0017] Furthermore, in step 1, first, Gaussian filtering is performed on the image according to formula (1):

[0018]

[0019] Then, the image is grayed according to formula (2) to reduce the three-channel color image to a single-channel grayscale image:

[0020] Gray=0.114B+0.587G+0.299R (2).

[0021] Furthermore, the step 2 specifically includes the following steps:

[0022] 201) Calculate the gradient magnitude and direction to obtain the gradient edge:

[0023] The gradient calculation formulas such as equation (3) and equation (4) are used to calculate the gradient amplitude and gradient angle:

[0024]

[0025] θ=tanh -1 (G x , G y ) (4)

[0026] The Sobel operator is used to calculate the gradient to obtain an edge with good effect;

[0027]

[0028] 202) The obtained gradient edge is multiple pixels wide, and the non-maximum suppression method is used to retain the local maximum gradient and suppress all other gradient values, so that the edge with multiple pixels wide is thinned to a single pixel wide;

[0029] 203) The refined edges are screened using high and low threshold screening methods to obtain all the real edges in the image.

[0030] Furthermore, in step 3, the Hough transform relies on a voting mechanism. A point in the parameter space corresponds to a straight line in the original space. The number of curves passing through a certain point in the parameter space is counted. When the number of curves passing through the point is higher than a threshold, it is the desired detection target. For a straight line, in the two-dimensional plane XY, the straight line equation is described as a point-slope form:

[0031] y=kx+b (6)

[0032] If the (k, b) of the straight line is known, the only straight line on the plane that satisfies the condition is determined; (k, b) is transferred to the parameter space (K, B) according to formula (7):

[0033] B=-Kx+y (7)

[0034] There are countless straight lines passing through a point (x0, y0) on the XY plane, that is, there are countless points satisfying (k, b) in the parameter space. To avoid straight lines with infinite slopes, equation (7) is rewritten as the trigonometric function form of equation (8):

[0035] ρ=xcosθ+ysinθ (8)

[0036] A threshold is set as the number of points required for a point set to form a straight line, and the frequency of each edge point in the image intersecting at a point in the θ-ρ coordinate system is counted. When the frequency is higher than the threshold, it is considered that this series of points are all on a straight line; in this way, the edges that are straight lines in the edge of the image are determined, and these straight line edges contain all the wire edge sub-segments.

[0037] Furthermore, in step 4, the improved K-means algorithm uses the slope of each sub-segment and its distance to the baseline as a feature value, instead of the x and y coordinates of each point for clustering, determines the direction by the slope, determines the position by the distance, and then characterizes each sub-segment with two parameters respectively. Finally, these two feature values ​​are substituted into the K-means algorithm for clustering, and the edge to which each sub-segment belongs can be determined, thereby completing the division of the sub-segments.

[0038] Furthermore, in step 6, based on the distance mapping model of monocular vision, the distance from the lens to the point A to be measured is calculated using equations (11), (12), and (13), and then the pixel width in the graph is mapped to the actual width using equation (14), where A and B are the actual points to be measured, and A' and B' are the projection points of A and B in the image:

[0039]

[0040]

[0041]

[0042]

[0043] Compared with the prior art, the present invention has the following beneficial effects: the system and method collect and upload the image and video data of the on-site power transmission line conductor through the image acquisition terminal, and then in the ice detection and early warning system, combine Canny edge detection, Hough line detection, improved K-means clustering algorithm, and least squares fitting algorithm to accurately identify the edge of the power transmission line conductor, and based on the distance mapping model of monocular vision, realize the mapping of pixel width and actual width, accurately obtain the actual thickness of the detected conductor, and finally realize automatic detection and early warning of ice through comparison, the detection is accurate and convenient, saving manpower and material resources, and has strong practicality and broad application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a system structure block diagram of an embodiment of the present invention.

[0045] Figure 2 It is a block diagram of the ice detection and warning system in an embodiment of the present invention.

[0046] Figure 3 It is a flow chart of the method implementation of the embodiment of the present invention.

[0047] Figure 4 Schematic diagram of parameter space in an embodiment of the present invention.

[0048] Figure 5 Schematic diagram of a distance mapping model based on monocular vision in an embodiment of the present invention.

[0049] Figure 6 It is a schematic diagram of a plane model in the Y direction in an embodiment of the present invention.

[0050] Figure 7 It is a schematic diagram of the X-direction plane model in an embodiment of the present invention.

[0051] Figure 8 Schematic diagram of a distance mapping model in an embodiment of the present invention.

[0052] Fig. 9 It is the original wire image in the embodiment of the present invention.

[0053] Fig.10 is the preprocessed image in the embodiment of the present invention.

[0054] Fig.11 It is an image after Canny edge detection in an embodiment of the present invention.

[0055] Fig.12 It is an image after Hough line detection in an embodiment of the present invention.

[0056] Fig.13 This is an image after the wire is fitted using the least squares method in an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0058] The present invention provides a transmission line conductor ice thickness detection system, such as Figure 1 As shown, it includes an image acquisition terminal and an ice detection and warning system, the image acquisition terminal includes a camera unit, an image processing unit, a communication unit and a power supply unit, the camera unit is installed on the on-site tower and the shooting direction is facing the wires installed on the transmission line, so as to take real-time photos and video monitoring of the transmission line wires, the camera unit is connected to the image processing unit to transmit the collected image and video data to the image processing unit for processing, the image processing unit is connected to the communication unit to send the processed image and video data to the ice detection and warning system through wireless communication; the power supply unit adopts solar energy storage to power each unit, mainly including solar panels, charge and discharge controllers, battery packs and POE power supply modules, uses solar energy for power supply during the day, and stores excess electricity in the battery pack, uses the battery pack for power supply at night, and the continuous power supply time is not less than 10 days.

[0059] like Figure 2 As shown, the ice detection and warning system includes a communication and analysis unit, a database unit, a historical data query unit, a parameter setting unit, a thickness calculation unit, a display unit and an early warning unit. The communication and analysis unit communicates with the image acquisition terminal to receive the image and video data uploaded by the terminal. The database unit includes a real-time database and a historical database for storing the collected data. The historical data query unit provides the historical ice thickness information of the line as needed. The parameter setting unit sets the conductor model parameters and environmental parameters according to different types of transmission line conductors to select appropriate parameters for conductor ice thickness calculation. The thickness calculation unit calculates the thickness value of the current conductor based on the collected image and video data, and calculates the conductor ice thickness value by comparing it with the thickness value of the un-iced conductor. The display unit is used to display the collected data and the calculation results of the conductor ice thickness in the form of numerical values, images, tables, etc. The early warning unit is used to compare the calculated transmission line conductor ice thickness with the line design bearing capacity threshold. If the threshold is exceeded, an alarm message is issued to the operation and maintenance personnel.

[0060] like Figure 3As shown, the present invention also provides a method for detecting ice thickness of transmission line conductors, wherein an image acquisition terminal is used to collect and upload images and video data of on-site transmission line conductors, and an ice detection and early warning system detects ice thickness of conductors according to the received image data in the following steps:

[0061] 1) In order to remove a large amount of noise and background interference in the image returned by the on-site camera unit, the image is preprocessed.

[0062] First, perform Gaussian filtering on the image according to formula (1):

[0063]

[0064] The image resolution returned by the 2-megapixel camera is 1920*1080. To improve the efficiency of the algorithm, the image is grayed according to formula (2) to reduce the three-channel color image to a single-channel gray image:

[0065] Gray=0.114B+0.587G+0.299R (2).

[0066] 2) Extracting the edge of the transmission line conductor: The first step is to extract the edge of the image. The present invention uses the Canny edge detection algorithm to extract the edge in the image, and the extracted edge includes the edge of the conductor and a large number of interfering line segments.

[0067] This algorithm is a multi-level detection algorithm with a low edge detection error rate. The detected edge points can be accurately located at the edge center. Any edge in the image is marked only once, and the image will eventually have a delicate edge without generating false edges. The algorithm specifically includes the following steps:

[0068] 201) Calculate the gradient magnitude and direction to obtain the gradient edge:

[0069] The edge of the image can point to any direction, so four gradient operators in the horizontal, vertical and diagonal directions are used for calculation. The gradient calculation formulas such as equations (3) and (4) are used to calculate the gradient amplitude and gradient angle:

[0070]

[0071] θ=tanh -1 (G x , G y ) (4)

[0072] The Sobel operator is used to calculate the gradient to obtain an edge with good effect;

[0073]

[0074] 202) The gradient edge usually obtained is multiple pixels wide, and the gradient map at this time cannot reflect the edge information well. The edge can only have an accurate point width. Therefore, the non-maximum suppression method is used to retain the local maximum gradient and suppress all other gradient values, so that the edge with multiple pixel widths is refined to a single pixel width.

[0075] 203) The refined edges are screened using high and low threshold screening methods to obtain all the real edges in the image.

[0076] Specifically, the gradient amplitude greater than the high threshold is considered a strong edge point, the gradient amplitude less than the low threshold is considered a non-edge point, and the gradient amplitude between the two thresholds is considered a weak edge point. By finding whether there is a strong edge point in the eight-neighborhood of the weak edge point, its nature is determined. If there is, it is an edge point, otherwise it is not.

[0077] 3) After obtaining the edge in the image, the background interference line segments should also be removed. The transmission line site environment is complex and mixed with a lot of background interference, so the wires need to be extracted from the background interference. The present invention uses the posture of the transmission line wires presenting straight lines in the image, and uses the Hough line detection algorithm to detect the straight line segments in the image to obtain the wire edge sub-segments.

[0078] Hough transform relies on a voting mechanism. A point in the parameter space corresponds to a straight line in the original space. The number of curves passing through a point in the parameter space is counted. When the number of curves passing through the point is higher than the threshold, it is the desired detection target. For a straight line, in the two-dimensional plane XY, the straight line equation is described as the point-slope form:

[0079] y=kx+b (6)

[0080] If the (k, b) of the line is known, the only line on the plane that satisfies the condition is determined. Transfer (k, b) to the parameter space (K, B) according to equation (7):

[0081] B=-Kx+y (7)

[0082] There are countless straight lines passing through a point (x0, y0) on the XY plane, that is, there are countless points satisfying (k, b) in the parameter space. To avoid straight lines with infinite slopes, equation (7) is rewritten as the trigonometric function form of equation (8):

[0083] ρ=xcosθ+ysinθ (8)

[0084] A threshold is set as the number of points required for a point set to form a straight line, and the frequency of each edge point in the image intersecting at a point in the θ-ρ coordinate system is counted. When the frequency is higher than the threshold, it is considered that this series of points are all on a straight line. In this way, the edges of the image edge that are straight lines are determined, and these straight line edges contain all the wire edge sub-segments.

[0085] 4) Use the improved K-means algorithm to cluster the wire edge sub-segments and determine the edge to which each sub-segment belongs.

[0086] The obtained sub-segments are distributed on each edge of the wire, and the ownership of each sub-segment needs to be determined. Each wire contains two left and right edges, that is, if the number of transmission lines in the image is 2, the number of edges is 4, then the wire sub-segments need to be divided into 4 categories. Since the classification categories are known, the present invention uses an improved K-means algorithm to classify the ownership of wire sub-segments.

[0087] K-means algorithm (K-means clustering algorithm) is a partitioning algorithm based on a known number of clustering categories. It uses distance as the criterion for similarity. The closer the distance between two objects, the stronger their similarity. The algorithm can handle large data sets and is efficient. Its input is the data set and the number of categories, and the clustering result is k data sets divided into k categories. The algorithm flow is:

[0088] (1) Given a data set of size n, let O = l and select k initial cluster centers Z j (O), j = 1, 2, 3, ..., k, O represents the clustering center of different iteration rounds.

[0089] (2) Calculate the distance D(x) between each sample data object and the aggregation center i , Z j (O)), i = 1, 2, 3, ..., and classify.

[0090] (3) Let O = O + 1, and calculate the new cluster center and the error square sum criterion f (objective function) value:

[0091]

[0092] (4) Judgment: |f(O+1)-f(O)|<theta(f converges) or the object has no category change, then the algorithm ends; otherwise, O=O+1, and return to step (2).

[0093] However, the traditional K-means algorithm is suitable for point sets distributed in groups or clusters, and has poor effect on point sets distributed linearly. To improve this defect, the present invention proposes an improved K-means clustering algorithm. The traditional K-means algorithm classifies each point by distance, calculates the distance from each point to the center point of each cluster or group, finds the center point with the smallest distance, and takes the group or cluster to which this center point belongs as the affiliation, and continuously corrects the position of the midpoint in this process, so as to gradually determine the affiliation of each point. In the same way, the dimension is raised from point to line, a center line is taken in the image, and the distance from each sub-line segment to the center line is calculated respectively, and the distance is used as the criterion to determine the distance between each sub-line segment and the center line. However, to calculate the distance between the two lines, it is necessary to ensure that the two are parallel. Each sub-line segment is often not parallel to the center line, and the distance between the two is difficult to solve, so the two must be parallel in space first. Take a center point on the center line of the image, make auxiliary lines through this point parallel to each sub-line segment, calculate the distance from the auxiliary line to the sub-line segment parallel to it, and the distance between the center line and each sub-line segment can be obtained by this processing method. This distance and the slope of the sub-segment can determine a unique sub-segment. Compared with the point-slope equation of a straight line, the two parameters of this method have clearer physical meanings and are more convenient to use (the edge straight line segment is a partial edge of a conductor, also called an edge sub-segment).

[0094] The improved K-means algorithm uses the slope of each sub-segment and its distance to the baseline as the feature value, instead of the x and y coordinates of each point for clustering, determines the direction by the slope, determines the position by the distance, and then characterizes each sub-segment with two parameters respectively. Finally, these two feature values ​​are substituted into the K-means algorithm for clustering, and the edge to which each sub-segment belongs can be determined, thereby completing the division of the sub-segments.

[0095] 5) The wire edge is fitted using the least square method to obtain a complete and continuous wire edge.

[0096] After obtaining the edge sub-segments corresponding to each wire and its left and right edges, they need to be fitted to obtain complete and continuous wire edges. Since the results obtained by the improved K-means clustering algorithm are highly accurate, only a simple fitting algorithm is needed to fit the sub-segments to obtain the edges of the wire.

[0097] 6) Based on the monocular vision distance mapping model, the pixel width of the wire is mapped to the actual width, that is, the actual thickness value of the current wire.

[0098] After fitting the wire edge, the wire thickness needs to be calculated. Since the near and far ends of the wire have different widths in the image, but the actual binary width values ​​are basically the same, this problem cannot be solved by linear mapping. Therefore, the present invention adopts a distance mapping model based on monocular vision to achieve this. The mapping model is as follows: Figure 5For the convenience of analysis and calculation, it is decomposed into two directions, Y and X, for analysis, respectively. Figure 6 , Figure 7 .

[0099] Based on the monocular vision distance mapping model and equations (11), (12), and (13), the distance from the lens to the measured point A is calculated. Figure 8 Based on the distance mapping model and geometric knowledge, the pixel width in the image is mapped to the actual width using formula (14), where A and B are the actual points to be measured, and A' and B' are the projection points of A and B in the image:

[0100]

[0101]

[0102]

[0103]

[0104] 7) Compare the calculated actual thickness value of the current conductor with the thickness value of the un-iced conductor to calculate the conductor ice thickness value. If the conductor ice thickness value exceeds the ice thickness threshold, an early warning is issued.

[0105] Assume that the thickness of the un-iced conductor is D1, and the current conductor thickness is D2. According to formula (15), the conductor ice thickness D is calculated as:

[0106] D=D1-D2 (15)

[0107] According to experimental and historical data, the ice thickness threshold T is set as the basis for judging the warning. When D exceeds the warning value T, a warning is issued, and the wire status is continuously observed. If necessary, de-icing is performed. If D is less than T, the system enters the next operation cycle.

[0108] The following is a specific example to verify the ability of the present invention to detect the ice thickness of the conductor.

[0109] Example 1

[0110] 1. Original wire image

[0111] There is a lot of noise in the background of the original wire image, such as Fig. 9 shown.

[0112] 2. Image preprocessing

[0113] Image preprocessing removes image noise and improves the contrast between the foreground and background areas of the image. Fig.10 for easy subsequent identification.

[0114] 3. Canny edge detection

[0115] Use the Canny edge detection algorithm to extract the edges in the image, such as Fig.11 As shown, it can be seen that there are a large number of interfering line segments in the image, and the edges of the wires can also be completely detected.

[0116] 4. Hough line detection

[0117] Based on Canny detection, the Hough line detection algorithm is used to detect straight line segments in the image. The detection effect is shown in the figure below. At this time, only the edges of the wires are in the image.

[0118] 5. Least squares method for fitting wires

[0119] The wire edge obtained by least squares fitting is made into a complete wire edge and is drawn on the original image, as shown in Fig.13 As shown, it can be seen that the fitted wire edge (yellow line) wraps around both sides of each wire very well, and the algorithm works well.

[0120] 6. Calculation of conductor thickness

[0121] The wires in the image have uniform widths. This method is used to calculate that the pixel widths in the middle of the wires are 17.1667 pixels and 18.0414 pixels, respectively. The left wire is smaller in width because the two wires in the image are split into two and the height of the left wire is 80 cm lower than that of the right wire. After adjusting the height difference, the pixel width of the left wire is calculated to be 17.9257 pixels. The two widths are very close, which is consistent with the actual situation.

[0122] 7. Distance Mapping

[0123] The pixel width is mapped to the actual width, and the model accuracy meets the requirements.

[0124] 8. Analysis of ice cover status

[0125] The calculated conductor thickness value is close to the conductor un-ice-covered thickness value, and the difference does not exceed the threshold value. Therefore, it is judged that the conductor is in an un-ice-covered state, which is consistent with the actual situation.

[0126] The above are preferred embodiments of the present invention. Any changes made according to the technical solution of the present invention, as long as the resulting functions do not exceed the scope of the technical solution of the present invention, belong to the protection scope of the present invention.

Claims

1. A method for detecting ice thickness of a transmission line conductor, characterized in that: The image and video data of the on-site power transmission line conductors are collected and uploaded by the image acquisition terminal. The ice detection and warning system detects the ice thickness of the conductors according to the received image data in the following steps: 1) Preprocess the image; 2) The Canny edge detection algorithm is used to extract the edges in the image. The extracted edges include the edges of the wires and a large number of interfering line segments; 3) Using Hough line detection algorithm to detect straight line segments in the image, and obtain wire edge sub-segments; 4) Using the improved K-means algorithm to cluster the wire edge sub-segments and determine the edge to which each sub-segment belongs; 5) Using the least square method to fit the obtained wire edge, a complete and continuous wire edge is obtained; 6) Based on the monocular vision distance mapping model, the pixel width of the wire is mapped to the actual width, that is, the actual thickness value of the current wire; 7) Compare the calculated actual thickness value of the current conductor with the thickness value of the un-iced conductor to calculate the conductor ice thickness value. If the conductor ice thickness value exceeds the ice thickness threshold, issue an early warning; In the step 1), first, the image is Gaussian filtered according to formula (1): Then, the image is grayed according to formula (2) to reduce the three-channel color image to a single-channel grayscale image: Gray=0.114B+0.587G+0.299R (2) The step 2) specifically comprises the following steps: 201) Calculate the gradient magnitude and direction to obtain the gradient edge: The gradient calculation formulas such as equation (3) and equation (4) are used to calculate the gradient amplitude and gradient angle: θ=tanh -1 (G x ,G y ) (4) The Sobel operator is used to calculate the gradient to obtain an edge with good effect; 202) The obtained gradient edge is multiple pixels wide, and the non-maximum suppression method is used to retain the local maximum gradient and suppress all other gradient values, so that the edge with multiple pixels wide is thinned to a single pixel wide; 203) Using high and low threshold screening methods to screen the refined edges to obtain all real edges in the image; In step 3), Hough transform relies on a voting mechanism. A point in the parameter space corresponds to a straight line in the original space. The number of curves passing through a point in the parameter space is counted. When the number of curves passing through the point is higher than a threshold, it is the desired detection target. For a straight line, in the two-dimensional plane XY, the straight line equation is described as a point-slope form: y=kx+b (6) If the (k, b) of the straight line is known, the only straight line on the plane that satisfies the condition is determined; (k, b) is transferred to the parameter space (K, B) according to formula (7): B=-Kx+y (7) There are countless straight lines passing through a point (x0, y0) on the XY plane, that is, there are countless points satisfying (k, b) in the parameter space. To avoid straight lines with infinite slopes, equation (7) is rewritten as the trigonometric function form of equation (8): ρ=xcosθ+ysinθ (8) Set a threshold as the number of points required for a point set to form a straight line, and count the frequency of each edge point in the image intersecting at a point in the θ-ρ coordinate system. When the frequency is higher than the threshold, it is considered that this series of points are all on a straight line. In this way, the edges of the image that are straight lines are determined, and these straight line edges contain all the wire edge sub-segments; In the step 4), the improved K-means algorithm uses the slope of each sub-line segment and its distance to the baseline as a feature value, instead of the x and y coordinates of each point for clustering, determines the direction by the slope, determines the position by the distance, and then characterizes each sub-line segment with two parameters respectively. Finally, these two feature values ​​are substituted into the K-means algorithm for clustering, and the edge to which each sub-line segment belongs can be determined, thereby completing the division of the sub-line segments; In step 6), based on the distance mapping model of monocular vision, the distance from the lens to the point A to be measured is calculated using equations (11), (12), and (13), and then the pixel width in the image is mapped to the actual width using equation (14), where A and B are the actual points to be measured, and A' and B' are the projection points of A and B in the image:

2. A transmission line conductor ice thickness detection system for implementing the transmission line conductor ice thickness detection method as claimed in claim 1, characterized in that: It includes an image acquisition terminal and an ice detection and early warning system. The image acquisition terminal includes a camera unit, an image processing unit, a communication unit and a power supply unit. The camera unit is installed on a pole tower on site and its shooting direction is facing the wires installed on the transmission line, so as to take real-time photos and video monitoring of the transmission line wires. The camera unit is connected to the image processing unit to transmit the collected image and video data to the image processing unit for processing. The image processing unit is connected to the communication unit to send the processed image and video data to the ice detection and early warning system through wireless communication. The power supply unit supplies power to each unit. The ice detection and early warning system includes a communication and analysis unit, a database unit, a historical data query unit, a parameter setting unit, a thickness calculation unit, a display unit and an early warning unit. The communication and analysis unit communicates with the image acquisition terminal to receive the image and video data uploaded by the terminal. The database unit includes a real-time database and a historical database for storing the collected data. The historical data query unit provides the historical ice thickness information of the line as needed. The parameter setting unit sets the conductor parameters and environmental parameters to calculate the ice thickness of the conductor. The thickness calculation unit calculates the thickness value of the current conductor based on the collected image and video data, and calculates the ice thickness value of the conductor by comparing it with the thickness value of the un-iced conductor. The display unit is used to display the collected data and the calculation result of the ice thickness of the conductor. The early warning unit is used to compare the calculated ice thickness of the transmission line conductor with the line design bearing capacity threshold. If the threshold is exceeded, an alarm message is issued to the operation and maintenance personnel.

3. A transmission line conductor ice thickness detection system according to claim 2, characterized in that: The power supply unit adopts solar energy storage mode for power supply, and mainly includes solar panels, charge and discharge controllers, battery packs and POE power supply modules. Solar energy is used for power supply during the day, and excess electric energy is stored in the battery pack. The battery pack is used for power supply at night.

Images