Conductor icing weight evaluation method based on multi-source data fusion three-dimensional reconstruction technology

Abstract

This invention provides a method for assessing the weight of ice accretion on power transmission lines based on multi-source data fusion and 3D reconstruction technology. By deploying image acquisition devices at four shooting positions on the target section of the transmission line and acquiring environmental parameters, the current ice accretion type can be accurately identified, and the corresponding ice density can be determined. Furthermore, based on the image data from the four shooting positions, the overall structure of the transmission line and ice accretion is reconstructed in 3D, thereby accurately determining the overall volume of the ice accretion on the target section. This reduces the error in volume assessment caused by the irregular shape of the ice accretion, thus accurately determining the ice volume of the target section. Finally, the accurate overall ice accretion mass and ice accretion mass per unit length of the target section are calculated, providing accurate data support for subsequent ice condition warnings, ice melting decisions, and line condition maintenance.

Description

Technical Field

[0001] This invention relates to a method for assessing the weight of ice accretion on power transmission lines, and more particularly to a method for assessing the weight of ice accretion on power transmission lines based on multi-source data fusion and three-dimensional reconstruction technology. Background Technology

[0002] Icing on transmission lines is one of the major natural disasters threatening the safe operation of the power grid. Severe icing can lead to conductor galloping, line breakage, and even tower collapse. Accurately predicting the weight of ice on conductors is of great engineering significance for ice condition early warning, ice melting decisions, and line condition maintenance.

[0003] In existing technologies, the main methods for obtaining the weight of ice accretion on transmission lines include the weighing method and the mechanical model method. The weighing method uses tension sensors installed on insulator strings or tower measuring points to measure the total load on the conductor, and then combines parameters such as temperature and wind speed to indirectly estimate the ice load. However, this method cannot directly give the weight of ice accretion per unit length of conductor. Usually, the total load needs to be converted into an equivalent uniform ice thickness, and the ice density needs to be assumed in the conversion process. However, the actual ice types are complex and diverse, such as rime, hoarfrost, and mixed rime, and the ice density varies significantly (approximately 0.2 g / cm³ to 0.9 g / cm³). Using a fixed density value uniformly will introduce a large weight prediction error.

[0004] The mechanical model method relies on the conductor's state equation and meteorological parameters, and is also limited by assumptions about the shape and density of icing, making it difficult to guarantee prediction accuracy.

[0005] Therefore, in order to solve the above-mentioned technical problems, it is urgent to propose a new technical approach. Summary of the Invention

[0006] In view of this, the purpose of this invention is to provide a method for assessing the weight of ice accretion on transmission lines based on multi-source data fusion and three-dimensional reconstruction technology. By deploying image acquisition devices at four shooting positions on the target section of the transmission line and acquiring environmental parameters, the current ice accretion type can be accurately identified, and the corresponding ice density can be determined. Furthermore, based on the image data from the four shooting positions, a three-dimensional data reconstruction of the overall structure of the transmission line and the ice accretion can be performed, thereby accurately determining the overall volume of the ice accretion on the target section, reducing the error in volume assessment caused by the irregular shape of the ice accretion, and thus accurately determining the ice volume of the target section. Finally, the accurate overall ice accretion mass and ice accretion mass per unit length of the target section can be calculated, providing accurate data support for subsequent ice condition early warning, ice melting decision-making, and line condition maintenance.

[0007] This invention provides a method for assessing the weight of ice accretion on conductors based on multi-source data fusion and three-dimensional reconstruction technology, comprising the following steps:

[0008] S1. Four image acquisition devices are arranged on the target power transmission line. The optical axis of the image acquisition devices is perpendicular to the axis of the power transmission line and the optical axes of the four image acquisition devices are located in the same plane. The optical axes of two adjacent image acquisition devices are perpendicular.

[0009] S2. Four image acquisition devices simultaneously acquire icing images of the target section of the target transmission line, and input the icing images into a trained icing type evaluation neural network to determine the current icing type of the transmission line.

[0010] S3. Perform three-dimensional reconstruction based on the icing images output by four image acquisition devices, and determine the icing volume of the target section of the transmission line based on the reconstructed three-dimensional point cloud data;

[0011] S4. Based on the icing type, query the density-type database to determine the icing density corresponding to the current icing type, and then determine the icing weight of the target section of the current transmission line based on the icing density and icing volume.

[0012] Furthermore, in step S1, the four image acquisition devices are the same image acquisition devices, and the vertical distance from the geometric center of the four image acquisition devices to the power transmission line is the same.

[0013] Furthermore, when training the network to evaluate icing type, it is also necessary to input the environmental parameters of the sample when the sample images are acquired, including: temperature, humidity, wind speed, and diameter of supercooled water droplets.

[0014] While acquiring real-time monitoring images from four image acquisition devices, real-time environmental parameters are also acquired. The real-time image parameters are of the same category as the sample environmental parameters. The real-time environmental parameters form an auxiliary feature vector, which is simultaneously input into the trained icing type evaluation network along with the real-time icing image.

[0015] Furthermore, the 3D reconstruction based on the icing images output by the four image acquisition devices specifically includes:

[0016] Obtain preliminary point cloud data;

[0017] Denoising the point cloud data;

[0018] The point cloud data is segmented to separate the target, consisting of the conductor and the ice cover, from the background;

[0019] The Poisson surface reconstruction method is used to reconstruct the target point cloud data consisting of the conductor and ice cover to form a three-dimensional mesh model.

[0020] Furthermore, based on the reconstructed 3D point cloud data, the specific ice accretion volume of the target segment of the transmission line is determined, including:

[0021] The overall volume of the 3D mesh model is determined using the voxel integration method.

[0022] The volume of the target segment of the transmission line is determined by the cross-sectional area of ​​the transmission line and the length of the target segment.

[0023] The ice-covered volume of the target section of the transmission line is determined by subtracting the volume of the overall volume from the volume of the target section of the transmission line.

[0024] The beneficial effects of this invention are as follows: By deploying image acquisition devices at four shooting positions on the target section of the transmission line and acquiring environmental parameters, the current icing type can be accurately identified, and the corresponding icing density can be determined. Furthermore, based on the image data from the four shooting positions, a three-dimensional data reconstruction of the overall structure of the transmission line and the icing can be performed, thereby accurately determining the overall volume of the icing conductor and the icing in the target section. This reduces the error in volume assessment caused by the irregular shape of the icing, thus accurately determining the icing volume of the target section. Ultimately, the accurate overall icing mass and icing mass per unit length of the target section can be calculated, providing accurate data support for subsequent ice condition warnings, ice melting decisions, and line condition maintenance. Attached Figure Description

[0025] The present invention will be further described below with reference to the accompanying drawings and embodiments:

[0026] Figure 1 This is a schematic diagram of the process of the present invention. Detailed Implementation

[0027] The present invention will be further described in detail below:

[0028] This invention provides a method for assessing the weight of ice accretion on conductors based on multi-source data fusion and three-dimensional reconstruction technology, comprising the following steps:

[0029] S1. Four image acquisition devices are arranged on the target transmission line. The optical axis of the image acquisition devices is perpendicular to the axis of the transmission line and the optical axes of the four image acquisition devices are located in the same plane. The optical axes of two adjacent image acquisition devices are perpendicular. That is to say, the four image acquisition devices are orthogonal to each other, which can ensure that images of the target section of the transmission line and the complete shape of the ice accretion are obtained from completely different angles, and can also avoid the number of images and data burden.

[0030] S2. Four image acquisition devices simultaneously acquire icing images of the target section of the transmission line, and input the icing images into a trained icing type evaluation neural network to determine the icing type of the current transmission line. The icing type evaluation model uses existing neural networks, such as YOLOv5 and YOLOv8, or Faster R-CNN, SSD (Single Shot MultiBox Detector), etc. The training process of these neural networks is existing technology and will not be described in detail here.

[0031] S3. Perform three-dimensional reconstruction based on the icing images output by four image acquisition devices, and determine the icing volume of the target section of the transmission line based on the reconstructed three-dimensional point cloud data;

[0032] S4. Based on the icing type, query the density-type database to determine the icing density corresponding to the current icing type. Then, based on the icing density and icing volume, determine the icing weight of the target section of the transmission line. This invention, by deploying image acquisition devices at four shooting positions on the target section of the transmission line and acquiring environmental parameters, can accurately identify the current icing type and determine the corresponding icing density. Furthermore, based on the image data from the four shooting positions, a three-dimensional data reconstruction of the overall structure of the transmission line and icing is performed, thereby accurately determining the overall volume of the iced conductor and icing in the target section. This reduces the error in volume assessment caused by the irregular shape of the icing, thus accurately determining the icing volume of the target section. Finally, the accurate total icing mass and icing mass per unit length of the target section are calculated, providing accurate data support for subsequent ice condition warnings, ice melting decisions, and line condition maintenance.

[0033] In this embodiment, in step S1, the four image acquisition devices are identical, and the vertical distance from the geometric center of each of the four image acquisition devices to the power transmission line is the same. Of course, the specific vertical distance needs to be determined based on the actual image acquisition device, ensuring that the acquired images have sufficient clarity and resolution to effectively guarantee the determination of the icing type and the accuracy of the 3D reconstruction.

[0034] In this embodiment, when training the network for evaluating icing type, it is also necessary to input the environmental parameters of the sample when acquiring the sample images, including: temperature, humidity, wind speed, and diameter of supercooled water droplets.

[0035] Real-time environmental parameters are acquired simultaneously with real-time monitoring images obtained by four image acquisition devices. These real-time image parameters are of the same category as the sample environmental parameters. The real-time environmental parameters form an auxiliary feature vector, which, along with the real-time icing images, is simultaneously input into a trained icing type evaluation network. Icing types generally include rime, rain rime, mixed rime, snow rime, non-icing conditions, and other attachments. However, rime (hoarse frost) forms at lower temperatures, where supercooled water droplets rapidly condense on the surface of conductors. It has a loose structure and contains numerous air gaps. Its density is typically low, generally between 0.1 and 0.4 g / cm³, with a typical value of 0.25 g / cm³. For rime under different meteorological conditions, the table can be further subdivided into soft rime and hard rime, each corresponding to different recommended densities.

[0036] Glaze: Formed from freezing rain or supercooled large water droplets, it spreads on the surface of a conductor and then freezes. It has a dense, transparent or translucent structure. Its density is relatively high, generally between 0.8 and 0.9 g / cm³, with a typical value of 0.85 g / cm³. The density of glaze varies relatively little, so a fixed representative value can be used.

[0037] Mixed rime / Glaze: It is formed by alternating or coexisting rime and glaze, with a complex structure and uneven internal density distribution. Its density range is relatively large, typically between 0.4 and 0.8 g / cm³. As can be seen, the density varies significantly between different icing types, and even within the same type, the density can differ. Therefore, by considering environmental parameters during neural network training and subsequent type recognition, the specific icing type can be accurately identified. For example, rime ice can be divided into three categories: Rime Ice I, Rime Ice II, and Rime Ice III. The density ranges for these three categories are 0.1–0.2 g / cm³ (recommended density value for this category is 0.15 g / cm³), 0.2–0.3 g / cm³ (recommended density value for this category is 0.25 g / cm³), and 0.3–0.4 g / cm³ (recommended density value for this category is 0.35 g / cm³), respectively. These three categories correspond to different environmental parameter conditions. Therefore, considering environmental parameters allows for a more detailed classification of icing types.

[0038] Of course, there is still a problem here. Since four image acquisition devices acquire images simultaneously and input them into the icing type assessment network, different results may be output. For example, the types of images output by the four image acquisition devices and input into the icing type assessment network may be different. In this case, it is necessary to re-acquire images and re-identify and infer environmental parameters. If the categories corresponding to three images are the same, then the average confidence score of the three images is used as the final judgment result. Based on this result, the specific sub-category (such as rime I, rime II, or rime III) within the category (e.g., the major category is rime) is searched. If two images are the same and the other two are different, then the category output by the two images is taken as the standard, and the sub-category is determined by the average confidence score of the two images.

[0039] In this embodiment, the three-dimensional reconstruction based on the icing images output by four image acquisition devices specifically includes:

[0040] In fact, during 3D reconstruction, the four image acquisition devices are not fixed, but move simultaneously along the power transmission line at a set operating speed to acquire several images. Then, the motion reconstruction structure combined with multi-view stereo vision technology (which is an existing technology) is used to acquire preliminary point cloud data.

[0041] The point cloud data is denoised to remove the effects of noise.

[0042] A Poisson surface reconstruction method (a prior art, not detailed here) is used to reconstruct the target point cloud data consisting of the guide wire and ice floes, forming a 3D mesh model. This 3D mesh model incorporates irregular shapes such as protrusions, edges, and icicles of the ice floes, facilitating accurate determination of the ice volume. After generating the 3D mesh, post-mesh processing is performed. Small, isolated faces caused by incomplete segmentation boundaries are removed, local holes caused by occlusion or matching failures are filled (using a hole-filling algorithm based on radial basis functions), and elongated triangular faces are remeshed to ensure uniform face size.

[0043] The ice accretion volume of the target segment of the transmission line determined based on the reconstructed 3D point cloud data specifically includes:

[0044] The overall volume of the 3D mesh model is determined by the voxel integration method; the voxel integration method is an existing technology and will not be elaborated here.

[0045] The volume of the target segment of the transmission line is determined by the cross-sectional area of ​​the transmission line (which can be directly obtained from the product specifications of the transmission line) and the length of the target segment of the transmission line.

[0046] The icing volume of the target segment of the transmission line is determined by subtracting the volume of the overall volume from the volume of the target segment. After the icing of the target segment is determined, the volume is multiplied by the corresponding type of purpose to obtain the icing mass of the target segment of the current transmission line. Then, the icing mass is divided by the length of the target segment to determine the icing mass per unit degree, thus forming the final icing index, which is beneficial for the formulation of subsequent measures.

[0047] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the ice weight of a conductor based on multi-source data fusion three-dimensional reconstruction technology, characterized in that: Includes the following steps: S1. Four image acquisition devices are arranged on the target power transmission line. The optical axis of the image acquisition devices is perpendicular to the axis of the power transmission line and the optical axes of the four image acquisition devices are located in the same plane. The optical axes of two adjacent image acquisition devices are perpendicular. S2. Four image acquisition devices simultaneously acquire icing images of the target section of the target transmission line, and input the icing images into a trained icing type evaluation neural network to determine the current icing type of the transmission line. S3. Perform three-dimensional reconstruction based on the icing images output by four image acquisition devices, and determine the icing volume of the target section of the transmission line based on the reconstructed three-dimensional point cloud data; S4. Based on the icing type, query the density-type database to determine the icing density corresponding to the current icing type, and then determine the icing weight of the target section of the current transmission line based on the icing density and icing volume.

2. The method of claim 1, wherein the method comprises: In step S1, the four image acquisition devices are the same image acquisition devices, and the vertical distance from the geometric center of the four image acquisition devices to the power transmission line is the same.

3. The method for assessing the weight of icing on conductors based on multi-source data fusion and three-dimensional reconstruction technology according to claim 1, characterized in that: When training the network to evaluate icing type, it is also necessary to input the environmental parameters of the sample when the sample images are acquired, including: temperature, humidity, wind speed, and diameter of supercooled water droplets. While acquiring real-time monitoring images from four image acquisition devices, real-time environmental parameters are also acquired. The real-time image parameters are of the same category as the sample environmental parameters. The real-time environmental parameters form an auxiliary feature vector, which is simultaneously input into the trained icing type evaluation network along with the real-time icing image.

4. The method for assessing the weight of icing on conductors based on multi-source data fusion and three-dimensional reconstruction technology according to claim 1, characterized in that: The 3D reconstruction based on the icy images output by four image acquisition devices specifically includes: Obtain preliminary point cloud data; Denoising the point cloud data; The point cloud data is segmented to separate the target, consisting of the conductor and the ice cover, from the background; The Poisson surface reconstruction method is used to reconstruct the target point cloud data consisting of the conductor and ice cover to form a three-dimensional mesh model.

5. The method for assessing the weight of icing on conductors based on multi-source data fusion and three-dimensional reconstruction technology according to claim 4, characterized in that: The ice accretion volume of the target segment of the transmission line determined based on the reconstructed 3D point cloud data specifically includes: The overall volume of the 3D mesh model is determined using the voxel integration method. The volume of the target segment of the transmission line is determined by the cross-sectional area of ​​the transmission line and the length of the target segment. The ice-covered volume of the target section of the transmission line is determined by subtracting the volume of the overall volume from the volume of the target section of the transmission line.

Images