Method for monitoring galloping of transmission line without spacer without interval by joint ellipse detection

Abstract

The application discloses a kind of combined ellipse detection's no spacer rod transmission line galloping lightweight monitoring method, comprising the following steps: S1: collection transmission line video data, image enhancement pretreatment is carried out to video data;S2: using improved lightweight DeepLabV3+ model realizes transmission line segmentation, and fast locks transmission line in initial frame of video;S3: using SiamMask algorithm tracks target line, adopts refined path to obtain line mask information and carries out real-time positioning when continuously segmenting;S4: using ellipse detection algorithm based on arc segment extraction judges whether no spacer rod line running track presents oval, infers the probability of forming oval in succession;S5: using line ellipse detection fitting parameter analysis line galloping intensity;using oval formation probability analysis line galloping probability.The application uses lightweight model to realize the segmentation and tracking of line, and combined ellipse detection realizes the galloping lightweight monitoring of no spacer rod line.

Description

Technical Field

[0001] This invention belongs to the field of transmission line galloping monitoring, specifically a lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection. Background Technology

[0002] Line inspection is a crucial link in ensuring the safe operation of local power transmission and distribution networks and is therefore highly valued. For transmission lines without spacers, the lack of mechanical separation makes them more susceptible to line galloping compared to lines with spacers. Furthermore, in recent years, freezing rain and snow in many regions have caused varying degrees of icing on transmission lines, leading to uneven conductor stress and triggering more line galloping events, resulting in numerous power outages. Current methods for monitoring line galloping primarily rely on data from tension sensors, inertial sensors, and BeiDou satellite navigation, which are costly and ignore the external characteristics of line galloping, namely, the elliptical trajectory it often forms. This invention, based on transmission line data, employs a lightweight model for line segmentation and tracking, combined with ellipse detection to achieve galloping monitoring, providing a fast, reliable, lightweight, and low-cost solution. Summary of the Invention

[0003] In view of this, the purpose of this invention is to provide a lightweight monitoring method for galloping of barless transmission lines using combined elliptic detection. The method employs a lightweight model to achieve line segmentation and tracking, and combines elliptic detection to achieve lightweight monitoring of galloping of barless transmission lines.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection includes the following steps:

[0006] S1: Collect video data of transmission lines and perform image enhancement preprocessing on the video data;

[0007] S2: Transmission line segmentation is achieved using an improved lightweight DeepLabV3+ model, which quickly locks the transmission line in the initial frame of the video;

[0008] S3: Use the SiamMask algorithm to track the target line. During continuous segmentation, refine the path to obtain the line mask information and perform real-time positioning.

[0009] S4: Use an ellipse detection algorithm based on arc segment extraction to determine whether the running trajectory of the line without spacers is elliptical, and infer the probability of forming an ellipse in the future;

[0010] S5: Analyze the line galloping intensity using the line ellipse detection fitting parameters; analyze the line galloping probability using the ellipse formation probability.

[0011] Furthermore, in step S1, color space transformation is used to augment the time-frequency data, and the Retinex algorithm is used to process the low-light image.

[0012] Furthermore, the steps for processing low-light images using the Retinex algorithm are as follows:

[0013] S1.1: Calculate the logarithmic form of the reflection components from the scene to the camera:

[0014] Log(R(x,y))=Log(I(x,y))-Log(L(x,y))=Log(I(x,y))-Log(I(x,y)*G(x,y))

[0015] In the formula: R(x,y) is the reflection component of the scene reflected to the camera; L(x,y) is the incident light component; I(x,y) is the original image captured; G(x,y) represents the Gaussian convolution kernel; and * represents the convolution operation.

[0016] S1.2: Perform linear stretching to improve image contrast; limit pixel values ​​to within 0-255 to obtain the final image enhancement result.

[0017] R(x,y)=255×(Value-min) / (max-min)

[0018] In the formula: Value represents the pixel value to be restricted; max is the maximum result of the iteration; min is the minimum result of the iteration.

[0019] Furthermore, the improved lightweight DeepLabV3+ model includes an encoding part and a decoding part. The encoding part includes a cascaded backbone network, an ASPP module, and a 1×1 convolutional layer. The backbone network is composed of the first 8 layers of the MobileNetV2 model. The ASPP module is a sequentially cascaded ASPP module. Coordinate attention modules are provided between the ASPP module and the 1×1 convolutional layer, and between the backbone network and the decoding part.

[0020] The MobileNetV2 model is derived from the MobileNetv1 model by introducing inverse residuals and linear bottlenecks, in order to improve the representation ability while reducing the number of model parameters.

[0021] The sequentially concatenated ASPP module parallelizes dilated convolutions with selected dilation rates, shares parameters between convolutions with different dilation rates, and uses 1×1 convolutions to reduce the dimensionality of the input features.

[0022] The coordinate attention module is used to incorporate position information into channel attention in order to capture pixel feature information at different locations.

[0023] Furthermore, in step S3, the method for refining the path is as follows: deconvolution operation is performed on the candidate window response RoW of the model to obtain a higher resolution feature map, the feature map of the search image is added to the feature map of RoW, and the segmentation result of the line target is generated through convolution operation.

[0024] Furthermore, in step S4, the method steps of the ellipse detection algorithm based on arc segment extraction are as follows:

[0025] S4.1: Line arc detection: The Canny edge detection algorithm is used to identify the image edges. Edge pixels are connected to form edge arcs. A preset length threshold is used to determine whether the edge arc is an elliptical arc. If it is, a valid arc that conforms to the shape of an ellipse is obtained. If not, invalid arcs that cannot be fitted into an ellipse are removed. Finally, a set of valid arcs that conform to the shape of an ellipse is obtained.

[0026] S4.2: Candidate Ellipse Selection and Ellipse Parameter Estimation: Select arc segments on the same ellipse from the set of valid arc segments to form a triple; estimate the ellipse parameters of the triple, including the ellipse center (x...). c ,y c ), major semi-axis a, minor semi-axis b, and ellipse tilt angle ρ;

[0027] S4.3: Remove false ellipses with poor fit and duplicate ellipses that have been calculated multiple times;

[0028] S4.4: Use the interaction ratio to calculate the probability that the line will form an ellipse.

[0029] Furthermore, in step S4.1, edge pixels are represented by coordinates and gradient direction, and are divided into two categories based on the gradient direction:

[0030]

[0031] Among them, g(T) i ) represents edge pixel T i Classification; T i =(x i ,y i ,ω i ) represents edge pixels; x i With y i Representing edge pixel T i The x and y coordinates; ω i Represents edge pixel T i The gradient direction is obtained by the Canny edge detection algorithm;

[0032] Connecting pixels of the same category within 8 neighborhoods forms an edge arc segment, the k-th edge arc segment Ω k Represented as:

[0033]

[0034] Where: M k This represents the number of pixels in the k-th edge arc segment.

[0035] Furthermore, in step S4.4, the probability that the line forms an ellipse is:

[0036]

[0037] In the formula: IoU represents the probability that the line forms an ellipse; ε j Indicates the detected ellipse; ε z Represents a real ellipse; area(ε) j ) and area(ε z ) represent the ellipse ε j and ε z The area.

[0038] Furthermore, in step S5, the method for analyzing the line galloping intensity using the line ellipse detection fitting parameters is as follows: the coordinates of the ellipse center (x... c ,y c The ratio of the major semi-axis a to the minor semi-axis b reflects the area and direction of the galloping line. The larger the ratio between the major semi-axis a and the minor semi-axis b, the smaller the galloping intensity of the line.

[0039] Furthermore, in step S5, the method for analyzing the probability of line galloping using the probability of ellipse formation is as follows:

[0040] When the probability of forming an ellipse is less than the first probability threshold, it is determined that the trajectory of the line does not form an ellipse and no dancing occurs.

[0041] When the probability of ellipse formation is between the first probability threshold and the second probability threshold, it is determined that the line dance is small and the possibility of forming an ellipse is low, so no special measures are needed.

[0042] When the probability of ellipse formation is between the second probability threshold and the third probability threshold, it is determined that there is a possibility of ellipse formation, and it is recommended to strengthen inspection and monitoring.

[0043] When the probability of ellipse formation is greater than the third probability threshold, it is determined that the possibility of forming an ellipse is high, and immediate measures should be taken for inspection and maintenance.

[0044] The first probability threshold is less than the second probability threshold, and the second probability threshold is less than the third probability threshold.

[0045] The beneficial effects of this invention are as follows:

[0046] This invention presents a lightweight monitoring method for galloping of barless transmission lines, combining ellipse detection. Based on transmission line video data, it uses an improved lightweight DeepLabV3+ model to quickly lock the line in the initial frame of the video, balancing speed and accuracy. The SiamMask adaptive algorithm is used for line tracking and continuous segmentation. Then, an ellipse detection algorithm based on arc segment extraction determines whether the trajectory of the barless transmission line is elliptical. The galloping intensity and probability are analyzed using the line ellipse detection fitting parameters and ellipse formation probability, respectively, to assess the risk of transmission line galloping. This method has certain guiding and practical significance. Attached Figure Description

[0047] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the following figures are provided for illustration:

[0048] Figure 1 This is a flowchart of an embodiment of the lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to the present invention.

[0049] Figure 2 A schematic diagram of image generation using the Retinex algorithm;

[0050] Figure 3 Comparison of low-light image data enhancement results;

[0051] Figure 4 This is a diagram of the structure of the lightweight DeepLabv3+ network;

[0052] Figure 5 A visualization of the multi-scenario segmentation results for power transmission lines;

[0053] Figure 6 This is a structural diagram of the SiamMask network;

[0054] Figure 7 To refine the path mask generation method diagram;

[0055] Figure 8 The results of the transmission line tracking experiment;

[0056] Figure 9 A visualization of the dance trajectory in a typical scene;

[0057] Figure 10 This is a visualization of the ellipse detection results for the railway trajectory. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0059] like Figure 1 As shown in the figure, the lightweight monitoring method for galloping of spacerless transmission lines combined with elliptic detection in this embodiment includes the following steps.

[0060] S1: Collect video data of the transmission line and perform image enhancement preprocessing on the video data.

[0061] In this embodiment, color space transformation is used to augment the time-frequency data, and the Retinex algorithm is used to process low-light images. Specifically, in this embodiment, color space transformation is used to simulate different weather and lighting conditions in the external environment to augment the transmission line video data, including brightness transformation, contrast transformation, saturation transformation, grayscale transformation, and combined transformation.

[0062] In this embodiment, the Retinex algorithm is used to process low-light images, including video images captured in scenes such as nighttime, rain, fog, and sandstorms. Figure 2 As shown, the steps for processing low-light images using the Retinex algorithm are as follows:

[0063] S1.1: Calculate the logarithmic form of the reflection components from the scene to the camera:

[0064] Log(R(x,y))=Log(I(x,y))-Log(L(x,y))=Log(I(x,y))-Log(I(x,y)*G(x,y))

[0065] In the formula: R(x,y) is the reflection component of the scene reflected to the camera; L(x,y) is the incident light component; I(x,y) is the original image captured; G(x,y) represents the Gaussian convolution kernel; and * represents the convolution operation.

[0066] S1.2: Perform linear stretching to improve image contrast; limit pixel values ​​to within 0-255 to obtain the final image enhancement result.

[0067] R(x,y)=255×(Value-min) / (max-min)

[0068] In the formula: Value represents the pixel value to be restricted; max is the maximum result of the iteration; min is the minimum result of the iteration.

[0069] like Figure 3As shown, from left to right and from top to bottom, the original image of the transmission line and the effects of Gaussian filtering, histogram equalization, and Retinex algorithm processing are presented.

[0070] S2: Transmission line segmentation is achieved using an improved lightweight DeepLabV3+ model, which quickly locks the transmission line in the initial frame of the video.

[0071] Specifically, in this embodiment, the improved lightweight DeepLabV3+ model includes an encoding part and a decoding part. The encoding part includes a cascaded backbone network, an ASPP module, and a 1×1 convolutional layer. The backbone network is composed of the first 8 layers of the MobileNetV2 model. The ASPP module is a sequentially cascaded ASPP module. Coordinate attention modules are provided between the ASPP module and the 1×1 convolutional layer, and between the backbone network and the decoding part.

[0072] Specifically, the MobileNetV2 model is derived from the MobileNetv1 model by introducing inverse residuals and a linear bottleneck to improve representation capabilities while reducing the number of model parameters. Only the first 8 layers of MobileNetV2 are used as the DeepLabV3+ backbone network for feature extraction, and a 512×512 RGB image is input to the network for power transmission line feature extraction, ensuring model accuracy while accelerating training speed. The detailed optimized structure of MobileNetV2 is shown in Table 1, with a channel expansion factor of T, input and output channels of Input-c and Output-c respectively, repetitions of N, and a step size of S.

[0073] Table 1 MobileNetV2 Structure

[0074]

[0075] In this embodiment, the sequentially concatenated ASPP module parallelizes dilated convolutions with selected dilation rates, shares parameters between convolutions with different dilation rates, and uses 1×1 convolutions to reduce the dimensionality of the input features.

[0076] In this embodiment, the coordinate attention module is used to incorporate position information into channel attention in order to capture pixel feature information at different locations.

[0077] Using the improved lightweight DeepLabV3+ model in this embodiment for line segmentation, the transmission lines are quickly located in the initial frame of the video, such as... Figure 5 The image shows a comparison of the transmission line segmentation effects of different algorithms under different backgrounds.

[0078] S3: Use the SiamMask algorithm to track the target line. During continuous segmentation, refine the path to obtain the line mask information and perform real-time positioning.

[0079] Specifically, in this embodiment, the SiamMask algorithm is used for tracking based on the locked line, and its network structure is as follows: Figure 6 As shown, during continuous segmentation, a path refinement method is used to generate a mask for the target line. The path refinement method is as follows: deconvolution operation is performed on the candidate window response RoW of the model to obtain a higher resolution feature map. The feature map of the search image is added to the feature map of RoW, and the segmentation result of the line target is generated through convolution operation.

[0080] A detailed diagram of path mask generation is shown below. Figure 7 As shown in the figure. The final tracking results for different lines are shown in the figure below. Figure 8 As shown. The line galloping trajectory is obtained by integrating the tracking results using the information contained in the line mask. The specific visualization results are shown below. Figure 9 As shown.

[0081] S4: Use an ellipse detection algorithm based on arc segment extraction to determine whether the running trajectory of the line without spacers is elliptical, and infer the probability of forming an ellipse in the future.

[0082] Specifically, in this embodiment, the steps of the ellipse detection algorithm based on arc segment extraction are as follows:

[0083] S4.1: Line arc detection: The Canny edge detection algorithm is used to identify the image edges. Edge pixels are connected to form edge arcs. A preset length threshold is used to determine whether the edge arc is an elliptical arc. If it is, a valid arc that conforms to the shape of an ellipse is obtained. If not, invalid arcs that cannot be fitted into an ellipse are removed. Finally, a set of valid arcs that conform to the shape of an ellipse is obtained.

[0084] In this embodiment, edge pixels are represented by coordinates and gradient direction, and are divided into two categories based on the gradient direction:

[0085]

[0086] Among them, g(T) i ) represents edge pixel T i Classification; T i =(x i ,y i ,ω i ) represents edge pixels; x i With y i Representing edge pixel T i The x and y coordinates; ω i Represents edge pixel T iThe gradient direction is obtained by the Canny edge detection algorithm;

[0087] Connecting pixels of the same category within 8 neighborhoods forms an edge arc segment, the k-th edge arc segment Ω k Represented as:

[0088]

[0089] Where: M k This represents the number of pixels in the k-th edge arc segment.

[0090] S4.2: Candidate Ellipse Selection and Ellipse Parameter Estimation: Select arc segments on the same ellipse from the set of valid arc segments to form a triple; estimate the ellipse parameters of the triple, including the ellipse center (x...). c ,y c ), major semi-axis a, minor semi-axis b, and ellipse tilt angle ρ.

[0091] S4.3: Remove false ellipses with poor fit and duplicate ellipses that have been calculated multiple times.

[0092] S4.4: Use the interaction ratio to calculate the probability that the line will form an ellipse.

[0093] Specifically, the probability of the line forming an ellipse is:

[0094]

[0095] In the formula: IoU represents the probability that the line forms an ellipse; ε j Indicates the detected ellipse; ε z Represents a real ellipse; area(ε) j ) and area(ε z ) represent the ellipse ε j and ε z The area.

[0096] S5: Analyze the line galloping intensity using the line ellipse detection fitting parameters; analyze the line galloping probability using the ellipse formation probability.

[0097] Specifically, in this embodiment, the method for analyzing the line galloping intensity using the line ellipse detection fitting parameters is as follows: the coordinates of the ellipse center (x... c ,y c The ratio of the major semi-axis a to the minor semi-axis b reflects the area and direction of the galloping line. The larger the ratio between the major semi-axis a and the minor semi-axis b, the smaller the galloping intensity of the line.

[0098] In this embodiment, the method for analyzing the probability of line galloping using the probability of ellipse formation is as follows:

[0099] When the probability of forming an ellipse is less than the first probability threshold, it is determined that the trajectory of the line does not form an ellipse and no dancing occurs.

[0100] When the probability of ellipse formation is between the first probability threshold and the second probability threshold, it is determined that the line dance is small and the possibility of forming an ellipse is low, so no special measures are needed.

[0101] When the probability of ellipse formation is between the second probability threshold and the third probability threshold, it is determined that there is a possibility of ellipse formation, and it is recommended to strengthen inspection and monitoring.

[0102] When the probability of ellipse formation is greater than the third probability threshold, it is determined that the possibility of forming an ellipse is high, and immediate measures should be taken for inspection and maintenance.

[0103] The first probability threshold is less than the second probability threshold, and the second probability threshold is less than the third probability threshold.

[0104] In this embodiment, the first probability threshold is 20%, the second probability threshold is 50%, and the third probability threshold is 70%. Of course, the specific data of the first, second, and third probability thresholds can be set according to the actual application scenario.

[0105] In this embodiment, ellipse detection is performed on the galloping trajectory of a line in a typical scenario. The detection results for each line in this scenario are as follows: Figure 10 As shown in Table 2:

[0106] Table 2 Results of Line Galloping Detection

[0107]

[0108] The galloping patterns of each track in the field were analyzed using the ellipse detection results. Track 1 was not detected as an ellipse, indicating a low probability of its trajectory forming an ellipse and a relatively small galloping amplitude; therefore, immediate repair and maintenance are not required. However, tracks 2 and 3 were detected as ellipses, with probabilities of 0.728 and 0.661, respectively. Track 2 did not exceed the 70% detection threshold, indicating moderate galloping; increased inspection and monitoring are recommended. Track 3 exceeded the 70% detection threshold, indicating a high probability of forming an ellipse, requiring immediate repair and maintenance.

[0109] Further analysis of the parameters of the ellipse fitted to the trajectories of routes 2 and 3 reveals that the coordinates of the ellipse center and the tilt angle reflect the region and direction of the galloping lines, while the ratio of the major to minor axis reflects the intensity of the galloping to some extent. Specific parameters are shown in Table 3.

[0110] Table 3 Parameters of Line Dancing Ellipse

[0111]

[0112] As can be seen from the ellipse parameters, the ratio of the major axis to the minor axis of line 2 is smaller than that of line 3. This is reflected in the degree of dancing, indicating that line 2 dances more violently and requires close attention.

[0113] Once the relevant steps are executed, lightweight monitoring of galloping in spacerless transmission lines can be achieved using combined elliptic detection.

[0114] This embodiment presents a lightweight monitoring method for galloping of spacerless transmission lines using elliptic detection. First, low-quality images are preprocessed with image enhancement to improve data quality and facilitate subsequent segmentation and tracking. Second, the improved lightweight line segmentation model used can accelerate the running speed while ensuring model accuracy. Then, based on the discovery that the trajectory of the spacerless transmission line is elliptical, elliptic detection is used to identify the risk of galloping of the transmission line, providing a new approach to line galloping monitoring.

[0115] The above-described embodiments are merely preferred embodiments provided to fully illustrate the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are all within the scope of protection of the present invention. The scope of protection of the present invention is defined by the claims.

Claims

1. A lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection, characterized in that: Includes the following steps: S1: Collect video data of transmission lines and perform image enhancement preprocessing on the video data; S2: Transmission line segmentation is achieved using an improved lightweight DeepLabV3+ model, which quickly locks the transmission line in the initial frame of the video. The improved lightweight DeepLabV3+ model includes an encoding part and a decoding part. The encoding part includes a backbone network, an ASPP module, and a 1×1 convolutional layer connected in series. The backbone network is composed of the first 8 layers of the MobileNetV2 model. The ASPP module is a sequentially cascaded ASPP module. Coordinate attention modules are provided between the ASPP module and the 1×1 convolutional layer, and between the backbone network and the 1×1 convolutional layer of the decoding part. The MobileNetV2 model is derived from the MobileNetv1 model by introducing inverse residuals and linear bottlenecks, in order to improve the representation ability while reducing the number of model parameters. The sequentially concatenated ASPP module reduces the dimensionality of input features by parallelizing dilated convolutions with selected dilation rates, sharing parameters and using 1×1 convolutions among convolutions with different dilation rates. The coordinate attention module is used to integrate position information into channel attention in order to capture pixel feature information at different positions; S3: Use the SiamMask algorithm to track the target line. During continuous segmentation, refine the path to obtain the line mask information and perform real-time positioning. S4: Use an ellipse detection algorithm based on arc segment extraction to determine whether the running trajectory of the line without spacers is elliptical, and infer the probability of forming an ellipse in the future; S5: Analyze the line galloping intensity using the line ellipse detection fitting parameters; Analyze the probability of line galloping using the probability of ellipse formation; The method for analyzing the galloping intensity of railway lines using the fitting parameters of the ellipse detection is as follows: ellipse center coordinates... and tilt angle It reflects the area and direction of the dancing line, and the long half-axis With short half shaft The larger the ratio between them, the smaller the intensity of the line's galloping.

2. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 1, characterized in that: In step S1, color space transformation is used to augment the time-frequency data, and the Retinex algorithm is used to process the low-light image.

3. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 2, characterized in that: The steps for processing low-light images using the Retinex algorithm are as follows: S1.1: Calculate the logarithmic form of the reflection components from the scene to the camera: In the formula: The reflected components of the scene are transmitted to the camera; For the incident component of light, The original image obtained from the capture. Represents a Gaussian convolution kernel. Represents convolution operation; S1.2: Perform linear stretching to improve image contrast; limit pixel values ​​to within 0~255 to obtain the final image enhancement result. In the formula: Indicates the pixel value to be restricted; The maximum result of the iteration; This is the minimum result of the iteration.

4. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 1, characterized in that: In step S3, the method for refining the path is as follows: deconvolution operation is performed on the candidate window response RoW of the model to obtain a higher resolution feature map, the feature map of the search image is added to the feature map of RoW, and the segmentation result of the line target is generated through convolution operation.

5. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 1, characterized in that: In step S4, the ellipse detection algorithm based on arc segment extraction has the following steps: S4.1: Line arc detection: The Canny edge detection algorithm is used to identify the image edges. Edge pixels are connected to form edge arcs. A preset length threshold is used to determine whether the edge arc is an elliptical arc. If it is, a valid arc that conforms to the shape of an ellipse is obtained. If not, invalid arcs that cannot be fitted into an ellipse are removed. Finally, a set of valid arcs that conform to the shape of an ellipse is obtained. S4.2: Candidate Ellipse Selection and Ellipse Parameter Estimation: Select arc segments on the same ellipse from the set of valid arc segments to form a triple; estimate the ellipse parameters of the triple, including the ellipse center. Long half shaft short half shaft and the ellipse tilt angle ; S4.3: Remove false ellipses with poor fit and duplicate ellipses that have been calculated multiple times; S4.4: Use the interaction ratio to calculate the probability that the line will form an ellipse.

6. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 5, characterized in that: In step S4.1, edge pixels are represented by coordinates and gradient direction, and are divided into two categories based on the gradient direction: in, Represents edge pixels Classification; For edge pixels; and Representing edge pixels respectively The x and y coordinates; Represents edge pixels The gradient direction is obtained by the Canny edge detection algorithm; Connecting pixels of the same category within 8 neighborhoods forms an edge arc segment. Edge arc segment Represented as: In the formula: Indicates the first The number of pixels on the edge arc segment.

7. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 5, characterized in that: In step S4.4, the probability that the line forms an ellipse is: In the formula: This indicates the probability that the line will form an ellipse. Indicates the detected ellipse; Represents a real ellipse; and They represent ellipses respectively and The area.

8. The lightweight monitoring method for galloping of spacerless transmission lines using combined elliptic detection according to claim 1, characterized in that: In step S5, the method for analyzing the probability of line galloping using the probability of ellipse formation is as follows: When the probability of forming an ellipse is less than the first probability threshold, it is determined that the trajectory of the line does not form an ellipse and no dancing occurs. When the probability of ellipse formation is between the first probability threshold and the second probability threshold, it is determined that the line dance is small and the possibility of forming an ellipse is low, so no special measures are needed. When the probability of ellipse formation is between the second probability threshold and the third probability threshold, it is determined that there is a possibility of ellipse formation, and it is recommended to strengthen inspection and monitoring. When the probability of ellipse formation is greater than the third probability threshold, it is determined that the possibility of forming an ellipse is high, and immediate measures should be taken for inspection and maintenance. The first probability threshold is less than the second probability threshold, and the second probability threshold is less than the third probability threshold.

Images