Power transmission channel external implicit space distance measuring device fusing laser point cloud and image and starting method of power transmission channel external implicit space distance measuring device

Abstract

The invention discloses a laser point cloud and image fused power transmission channel external damage hidden space distance measuring device and a starting method thereof.The power transmission line external damage hidden danger monitoring device is installed on a power transmission tower and mainly comprises a laser point cloud acquisition module, a common light image acquisition module and a night vision image acquisition module; according to the external damage hidden danger space distance measuring device, a laser point cloud acquisition module and an image acquisition module are integrated into the device, and a direction adjusting mechanism is arranged at the installation position of the laser point cloud acquisition module and the image acquisition module; the method comprises the steps of starting point cloud data acquisition to reconstruct point cloud data and starting a distance measurement method. According to the method, the visualization function of an existing power transmission line can be considered, the distance from the external damage hidden danger to the power transmission body can be calculated, quantitative analysis of the external damage hidden danger is achieved, and invalid warning of the external damage hidden danger is greatly reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent operation and maintenance of power transmission lines, and in particular to a device for measuring distance outside a power transmission channel by integrating laser point cloud and image, and a method for starting the device. Background Art

[0002] Transmission lines are the main arteries of the power grid. As the core component of the power grid, their safe, reliable and stable operation is related to the safe operation of the entire power system itself. However, most transmission lines are distributed in the wild, and the lines cover a wide area. They are inevitably affected by over-height vehicles, construction machinery and tall trees, which damage the safe and stable operation of the power system.

[0003] In order to prevent damage to the power system caused by external forces such as over-high vehicles, construction machinery, and tall trees, the visual online monitoring device for transmission lines with hidden danger identification has been vigorously promoted and applied, realizing the identification and early warning of hidden dangers of external damage in the transmission channel environment and the main body. However, although the hidden dangers of external damage far away from the transmission line channel and the main body appear in the picture taken by the visualization device, they do not cause damage to the transmission channel and the main body, thereby increasing the frequency of pushing the alarm information of external damage hidden dangers of the transmission line visualization device, causing the operation and maintenance personnel to constantly confirm the alarm information of external damage hidden dangers, which greatly increases the workload of the operation and maintenance personnel.

[0004] In order to achieve effective distance measurement of hidden dangers inside and outside the transmission channel, the existing tower monitoring devices can be used to collect images and integrate them with the point cloud data scanned by drones for distance measurement. However, due to tree growth and human activities in some areas, the topography has changed, and the original point cloud data scanned by drones is relatively old and cannot meet the current distance measurement needs. If the point cloud data in the transmission channel is collected again, it will not only increase the workload of the operation and maintenance personnel, but also require the re-collection of point cloud data as the scene in the transmission channel changes over time. In addition, at this stage, the distance measurement of hidden dangers in the internal transmission line is mainly completed on the platform side. When measuring the distance of external hidden dangers in a large number of transmission channels, the equipment resources on the platform side cannot meet the distance measurement algorithm analysis, and need to be continuously upgraded.

[0005] In summary, the existing visual monitoring device for power transmission lines only realizes the qualitative analysis of the hidden dangers of external force damage. Even if the spatial ranging algorithm is embedded in the platform, it is impossible to perform real-time analysis of a large amount of tower data. Therefore, it is urgent to solve the distance measurement problem between the hidden dangers of external damage and the power transmission body in the front-end equipment, and conduct quantitative analysis of the hidden dangers of external damage. Summary of the invention

[0006] Purpose of the invention: The purpose of the present invention is to provide a device for breaking the invisible space distance measurement outside the power transmission channel by integrating laser point cloud and image and a method for starting the device.

[0007] Technical solution: The power transmission channel external hidden danger space ranging device integrating laser point cloud and image described in the present invention includes a laser point cloud acquisition module and an image acquisition module. The image acquisition module includes a general light image acquisition module and a night vision image acquisition module. The external hidden danger space ranging device integrates the laser point cloud acquisition module and the image acquisition module into the device, and sets a direction adjustment mechanism at the installation position of the laser point cloud acquisition module and the image acquisition module to ensure that the data acquisition directions of the two modules remain consistent after installation.

[0008] Furthermore, the external damage potential space distance measuring device includes a main control module, a power supply module, a remote transmission module and auxiliary fixing structures.

[0009] Furthermore, the laser point cloud acquisition module and the image acquisition module are kept in the same position, and are periodically or actively awakened to acquire point cloud data in the power transmission channel, thereby realizing real-time acquisition of three-dimensional data of the power transmission channel and construction of a three-dimensional space model.

[0010] Furthermore, the image acquisition module for collecting image data of the transmission channel and the main body is integrated with a starlight-level night vision lens to collect color pictures of the transmission channel and the main body at night, thereby realizing 24-hour uninterrupted monitoring of hidden dangers in the transmission channel.

[0011] The method for starting distance measurement outside a power transmission channel by fusing laser point cloud and image of the present invention comprises starting point cloud data collection, reconstructing point cloud data and starting a distance measurement method.

[0012] Furthermore, the starting point cloud data collection and reconstructing point cloud data comprises the following steps:

[0013] (1) After the device is installed, the scene in the power transmission channel captured by the image module of the current device is set as the initial image through local connection or remote connection;

[0014] (2) The device uses a transmission line hidden danger identification model to identify hidden dangers in the transmission channel;

[0015] (3) Pre-embedding the optimized twin neural network, and using the optimized twin neural network to identify the difference between the current image and the initial image;

[0016] (4) The device uses the optimized twin neural network to detect the type of different parts and determine whether there are changes in tree growth, topography, and buildings in the transmission channel;

[0017] (5) Calling the laser point cloud acquisition module to re-acquire the number of point clouds in the transmission line channel;

[0018] (6) The laser point cloud data module in the device also supports scheduled start and remote start, and the operation and maintenance personnel configure it according to the operation and maintenance requirements.

[0019] Furthermore, the starting ranging method includes:

[0020] (7) The device uses a transmission line hidden danger identification model to identify hidden dangers in the transmission channel;

[0021] (8) When hidden dangers of tower cranes, cranes, and construction machinery channels are detected in the transmission channel, the spatial ranging algorithm is activated.

[0022] Furthermore, the spatial ranging algorithm includes:

[0023] The point cloud data and image data are integrated to construct a three-dimensional spatial data model; and the nearest distance from the hidden dangers inside and outside the transmission line channel to the power line or transmission tower is measured with the help of spatial Euclidean distance measurement and nearest neighbor search method.

[0024] Furthermore, the construction of the three-dimensional space data model includes the following steps:

[0025] S1: extract feature values ​​from two-dimensional images;

[0026] S2: construct the relationship between the 2D pixel coordinates of the feature point and its 3D spatial coordinates;

[0027] S3: Calculate the rotation matrix and translation vector of the image coordinate system and the three-dimensional space coordinate system;

[0028] S4: For image sequences, a dense point cloud is obtained by using a dense matching method after relative pose calculation between images and encryption of sparse points. After removing outliers and noise points, a preprocessed scene dense point cloud is obtained.

[0029] s5: transform the scene dense point cloud into the LiDAR coordinate system according to the initial registration parameters. At this time, the image dense point cloud and the laser point cloud are aligned.

[0030] S6: Search the nearest corresponding points between the two point clouds with distance as the constraint, and calculate the initial mean square distance d1 between the two point clouds;

[0031] S7: Calculate the rotation and translation parameters of the set of neighboring point pairs according to the least squares rule, and transform and adjust the pose of the dense point cloud according to the calculation results;

[0032] S8: Calculate the mean square distance d2 between the nearest corresponding point sets of the two point clouds after transformation. If the difference between d2 and d1 is less than a given threshold, the two point clouds achieve optimal matching. Otherwise, continue to repeat steps S6 and S7 to perform the next iteration.

[0033] Furthermore, the measurement of the nearest distance from the hidden danger inside and outside the transmission line channel to the power line or transmission tower includes:

[0034] S9: using the embedded transmission channel external damage hidden danger identification algorithm to obtain the hidden danger target pixel area, and then obtain the coordinate point of the target in the transmission three-dimensional space;

[0035] S10: traverse and calculate the nearest distance from the hidden danger target to the power line or transmission tower, quickly calculate the nearest distance from the external hidden danger target point area to the power line or transmission tower, adopt the idea of ​​space segmentation, and continuously search from the bottom to the top for the distance from the hidden danger target area to the power line and power equipment.

[0036] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages:

[0037] (1) Compared with the existing power transmission line visualization device, the device of the present invention can not only take into account the existing power transmission line visualization function, but also calculate the distance from the external failure hidden danger to the power transmission body, thereby realizing the quantitative analysis of the external failure hidden danger and greatly reducing the invalid alarm of the external failure hidden danger;

[0038] (2) The device integrates a laser point cloud acquisition module and an image acquisition module and is installed on a transmission tower. When the transmission channel scene changes, such as tree growth or terrain changes, the transmission channel point cloud data can be collected in time to re-model the three-dimensional model;

[0039] (3) The installation position and monitoring angle of the laser point cloud acquisition module and the image acquisition module in the device are consistent to achieve the alignment of the point cloud and the image. Compared with the point cloud and image alignment achieved through the visual positioning algorithm, it does not require a large number of accurate pose marker samples;

[0040] (4) The spatial ranging startup method based on the twin neural network can be used to identify changes in fixed scenes in the transmission channel, and the laser point cloud module can be awakened in time to collect point cloud data in the transmission channel;

[0041] (5) The spatial ranging startup method based on the twin neural network can adjust the working and sleeping modes of the laser point cloud module according to the business, avoiding the energy consumption caused by the operation of the laser point cloud module during the period when there is no point cloud collection demand, reducing the average power consumption of the device, and thus improving the stability and effectiveness of the installation work;

[0042] (6) The spatial distance measurement startup method based on the twin neural network can identify the movement trend of potential danger targets such as cranes and construction machinery, and timely adjust the device working module to calculate the distance from the moving potential danger target to the transmission channel in real time;

[0043] (7) In the process of fusion of graphics and point cloud, the nearest corresponding points between two point clouds are searched with distance as constraint to fuse the point cloud data with the image data, and the process is iterated continuously until the optimal match is achieved;

[0044] (8) The three-dimensional distance measurement method that utilizes the idea of ​​space segmentation can effectively improve the speed of space distance calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a flow chart of the method for starting the distance measurement in the outer space of the transmission channel;

[0046] Figure 2 2(a) is a schematic diagram of the structure of the space distance measuring device, 2(b) is a front view of the distance measuring device with hidden danger of external damage, and 2(a) is a rear view of the distance measuring device with hidden danger of external damage;

[0047] Figure 3 This is the image difference comparison network architecture diagram based on the twin neural network;

[0048] Figure 4 This is the startup flow chart of the laser point cloud module based on the twin neural network;

[0049] Figure 5 It is a flow chart of the external damage hidden danger distance measurement algorithm that integrates laser point cloud and image;

[0050] Figure 6 It is the flow chart of laser point cloud and image data fusion;

[0051] Figure 7 This is a flow chart for calculating the closest distance from external damage hazards to power lines and towers;

[0052] Figure 8 This is the overall structure diagram of the main control module of the space ranging device. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0054] The present invention proposes a transmission line external damage hidden danger ranging start mechanism, method and device for real-time measurement of the distance between the external damage hidden danger and the transmission line body, wherein the starting mechanism can automatically or controllably start the laser power acquisition module to collect the monitoring scene point cloud data in the transmission channel in combination with the daily operation and maintenance requirements of the transmission line and the on-site conditions; the ranging method can quickly measure the distance between the external damage hidden danger and the adjacent transmission line and the tower; the device is installed on the transmission tower to quantitatively analyze the external damage hidden danger in the transmission channel, laying a foundation for the refined management of transmission operation and maintenance, and specifically includes the following parts:

[0055] First, the overall design of the device described in this invention is as follows:

[0056] like Figure 2As shown, the external damage hidden danger distance measuring device of the present invention adopts an integrated design scheme, which is convenient for the installation of the equipment. The device integrates a laser point cloud acquisition module 3, an image acquisition module including a general light image acquisition module 1, a night vision image acquisition module 2, a main control module (in which a remote transmission module is embedded) 4, and a power supply module 5. The solar panel used for power supply is installed independently, which is convenient for selecting the appropriate power according to the installation area, and can also be installed facing the sun. Figure 2 (a) is the front view of the external damage potential distance measuring device. Figure 2 (b) is the rear view of the external damage potential distance measuring device.

[0057] The specific design scheme of each module is as follows:

[0058] The main control module is the core component of the device. In order to reduce the data processing pressure of the operation and maintenance platform, the main control module needs to have image processing, three-dimensional space ranging and other analysis capabilities to achieve on-site analysis of hidden dangers of power transmission and internal and external damage. The overall design framework of the main control module of the space ranging device proposed in this patent is as follows: Figure 8 As shown in the figure, the main control module is embedded with Linux system, and includes edge computing frameworks such as lightweight model for identifying hidden dangers of external damage to transmission line channels and three-dimensional space ranging algorithm. In addition, the main control module is embedded with remote module to upload analysis results in real time. The remote transmission module includes 4G communication module, security verification and power safety chip, which encrypts data while uploading analysis results to ensure the security of power grid Internet.

[0059] The image acquisition module and laser point cloud acquisition module collect the transmission channel image and power data. The image acquisition module is divided into general light image acquisition and night vision image acquisition. At the same time, the device is equipped with a photosensitive element. The device automatically switches between the general light image acquisition module and the night vision image acquisition module according to the light data collected by the photosensitive element. The laser point cloud module collects point cloud data in the transmission channel environment. It can not only collect data based on remotely issued instructions, but also automatically collect data based on changes in the monitoring scene.

[0060] The power supply module includes a lithium iron phosphate battery and a solar panel to power the device. At the same time, the power supply module includes a power management module, which can automatically adjust the battery charging and discharging mode according to the output power of the solar panel and the current power consumption of the device.

[0061] like Figure 1 As shown in the figure, the main idea of ​​the spatial ranging startup method based on the twin neural network described in the present invention is to use the twin neural network to judge the hidden dangers inside and outside the device channel and whether the scene has changed. The network architecture is as follows Figure 3As shown, the initially set image and the image collected by the device in a timed or real-time manner are taken as two inputs, the input image feature values ​​are extracted using the ResNet network structure, and the feature values ​​are input into the loss function to calculate the image differences. The loss function described in the present invention includes two parts, namely the Softmax target loss function and the maximum weighted perfect matching algorithm based on metric learning.

[0062] Among them, the calculation formula of Softmax target loss function is as follows:

[0063]

[0064] In addition, the calculation formula of the maximum weight perfect matching algorithm based on metric learning is as follows:

[0065]

[0066] During the actual operation of the device, when there is no significant change in the scene, there is no need to collect point cloud data. In order to reduce the average power consumption of the device, the working mode of the laser point cloud acquisition module is switched to sleep mode during routine monitoring. Then the device uses the twin neural network to periodically judge the scene in the channel. The device decides whether to wake up the laser point cloud module based on the output result. The process of the spatial ranging startup method based on the twin neural network described in this article is as follows: Figure 4 As shown, the specific steps are as follows:

[0067] Step 1: Set the initial image of the device remotely. The initial image should be the image taken by the device when there is no hidden danger of external damage in the channel;

[0068] Step 2: The device collects images regularly or in real time and inputs them into the twin neural network before identifying hidden dangers to determine whether the current image has changed from the initial image;

[0069] Step 3: If there is no change, the device does not alarm and continues to enter the normal monitoring mode;

[0070] Step 4: If any changes occur, the twin neural network described in the present invention is used to determine any changes in the transmission channel scene. At the same time, the environmental device internal and external hidden danger algorithm analyzes the images collected by the device. If there are changes in the topography caused by tree growth and human activities, the device wakes up the laser point cloud module to re-collect point cloud data, and at the same time wakes up the spatial ranging algorithm to calculate the proximity of hidden dangers such as trees and temporary buildings to power lines and towers in the new scene; if there are external hidden dangers such as cranes, excavators and other construction machinery in the transmission channel, the device automatically adjusts the image acquisition interval (the default is 30s, which can also be set remotely), and then uses the twin neural network to analyze the target position of the external hidden danger. If the construction machinery stays in the transmission channel, the spatial ranging algorithm is woken up to calculate the distance from the construction machinery to the nearest power line and transmission tower. If it leaves the channel, no action is taken.

[0071] Next, the transmission channel external damage hidden danger distance measurement algorithm of the present invention which fuses the image and point cloud data is introduced. The algorithm flow is as follows: Figure 5 After acquiring the image and point cloud data, the device extracts the feature value of the image and matches it with the data in the point cloud, then constructs a mapping relationship between the two-dimensional image and the three-dimensional point cloud data, and then uses the spatial Euclidean distance algorithm to calculate the distance. Finally, the calculated distance value is screened to find out the distance between the external damage hidden danger target and the nearest power line and transmission tower.

[0072] The key point of the distance measurement algorithm described in this paper is to construct the mapping relationship between the two-dimensional image and the three-dimensional point cloud data, that is, to calibrate the image acquisition module and the point cloud data, and to evaluate the calibrated data using the test set. The specific process is as follows: Figure 6 As shown, the detailed fusion steps are as follows:

[0073] Step 1: Extract features of two-dimensional images and point cloud data. The method used is consistent with the image feature extraction algorithm in the spatial ranging algorithm based on the twin neural network. The ResNet network structure is used to extract the input image feature values.

[0074] Step 2: Mapping 2D (image) coordinates to 3D (point cloud) space, mapping 2D images and 3D point clouds into a shared latent representation space, and then simultaneously describing and detecting key points;

[0075] Step 3: Calculate the mapping coordinate values, where the coordinate system conversion is the basis for the fusion of the two data. When fusing point cloud and image data, the coordinate conversion parameters of the two data need to be solved. The fusion image and point cloud ranging method described in the present invention uses the coordinate system of the point cloud as the reference coordinate system, and then performs the coordinate relationship of the corresponding point-pixel pairs in the point cloud and image based on the position and posture of the panoramic image in the coordinate system of the laser point cloud acquisition module. In the coordinate system of the laser point cloud acquisition module, let the coordinates of any laser point C be (X, Y, Z), the coordinates of the pixel point P on the image corresponding to point C be (Row, Col), and the height and width of the affected point cloud and pixel coordinates be r and c, respectively. The corresponding relationship between the point cloud coordinates (X, Y, X) and the pixel coordinates (Row, Col) can be converted using the following three equations:

[0076]

[0077] Among them, (x, y, z) is the point cloud coordinate after the origin of the coordinate system is normalized to the center of the sphere, X s , Y s , Z s , R Ψ , R φ , R θ It is determined by the six pose parameters of the panoramic image, where (X s , Y s , Z s ) is the coordinate of the sphere center, that is, the coordinate of the imaging center of the panoramic camera during shooting, R Ψ , R φ , R θ They are the rotation matrices corresponding to the three attitude angles, and Ψ, φ, θ are the three attitude angle parameters of the panoramic camera during shooting. The three rotation matrices are as follows:

[0078]

[0079]

[0080] Step 4: preprocess the point cloud data to convert the sparse point cloud data into denser point cloud data. The specific steps are as follows:

[0081] Step 4.1: For the image sequence, use the dense matching method to calculate the relative pose between images and encrypt the sparse points to obtain a dense point cloud, and remove outliers and noise points to obtain the pre-processed scene dense point cloud;

[0082] Step 4.2: Transform the scene dense point cloud into the LiDAR coordinate system according to the initial registration parameters. At this time, the image dense point cloud and the laser point cloud should be roughly aligned.

[0083] Step 4.3: Search for the nearest corresponding points between the two point clouds with distance as the constraint, and calculate the initial mean square distance d1 between the two point clouds;

[0084] Step 4.4: Calculate the rotation and translation parameters of the set of neighboring point pairs according to the least squares rule, and transform and adjust the pose of the dense point cloud according to the calculation results;

[0085] Step 4.5: Calculate the mean square distance d2 between the nearest corresponding point sets of the two point clouds after transformation. If the difference between d2 and d1 is less than the given threshold, the two point clouds achieve optimal matching. Otherwise, continue to repeat the above steps 4.4 and 4.5 for the next iteration.

[0086] Step 5: List the coordinate values ​​and make corrections. This is mainly for training the model after the image and point cloud data are calibrated. The label distribution predicted by the model training can be as close to the label distribution of the training set as possible. The loss function of the present invention is shown as follows:

[0087]

[0088] Among them, x, y, z, p, q, r are the components of the posture, P l With D l is a certain pose component and the predicted distribution and pose distribution.

[0089] After the above-mentioned image and point cloud data processing is completed, the structured data after pixel point and spatial coordinate mapping is generated. Subsequently, the nearest distance from the hidden danger inside and outside the transmission line channel to the power line or transmission tower is calculated by using the spatial Euclidean distance measurement and the nearest neighbor search method. The algorithm flow is as follows: Figure 7 First, the fused point cloud and image data are used to distinguish the point cloud data of ground buildings, construction machinery, trees, poles, wires, etc. Then the edge points of various point cloud data are extracted and saved. The edge points include sharp points, boundary points and some flat points. Finally, based on the idea of ​​space segmentation, the distance from the external damage hidden danger to the power line and pole is calculated, and the nearest distance is obtained. The detailed calculation steps are as follows:

[0090] Step 1: dividing the point cloud data, dividing the fused point cloud data of the present invention, and distinguishing the point cloud data of ground buildings, construction machinery, trees, pole towers and conductors, etc.;

[0091] Step 2: Extract edge points. Extract edge data of ground buildings, construction machinery, trees, poles and wires, etc., including sharp points, boundary points and some flat points. The edge extraction method is divided into three steps: the first step is to partition the external damage potential area, with an interval length of 1 meter; the second step is to search upward from the interval length of the bottom three-dimensional point of the external damage potential. After each search, the searched three-dimensional points are projected onto the image to check whether the vertical coordinates of the projected pixels are greater than the upper limit of the external damage potential projection. If so, stop the search. The third step is to use the binary search method to search for the edge points of the external damage potential based on the interval obtained in the second step, repeatedly divide the interval into two halves, and determine the sub-interval containing the top three-dimensional point of the external damage potential to approximate the top three-dimensional point. The vertical coordinates of the three-dimensional points searched by the binary search method after projection are equal to the highest pixel point of the external damage potential target, stop the search, and then obtain the three-dimensional edge points of the external damage potential.

[0092] Step 3: traverse and calculate the nearest neighbor distance from the external damage hidden danger to the power line and the transmission tower. The method of the present invention finally uses the Euclidean formula to calculate the distance from the target area of ​​the external damage hidden danger in the three-dimensional point cloud to the power line and the transmission tower, and compares them to finally obtain the nearest neighbor distance, where the distance calculation formula is as follows:

[0093]

Claims

1. A device for measuring distance outside a power transmission channel by fusing laser point cloud and image, characterized in that: The device comprises a laser point cloud acquisition module (3) and an image acquisition module, wherein the image acquisition module comprises a general light image acquisition module (1) and a night vision image acquisition module (2). The device for measuring distance between external damage potential hazards integrates the laser point cloud acquisition module (3) and the image acquisition module into the device, and provides a direction adjustment mechanism at the installation position of the laser point cloud acquisition module (3) and the image acquisition module to ensure that the data acquisition directions of the two modules remain consistent after installation.

2. The device for measuring distance outside the power transmission channel by fusing laser point cloud and image according to claim 1 is characterized in that: The external damage potential space distance measuring device comprises a main control module (4), a power supply module (5), a remote transmission module and auxiliary fixing structural parts.

3. The device for measuring distance outside the transmission channel by fusing laser point cloud and image according to claim 1 is characterized in that: The laser point cloud acquisition module (3) is kept at the same position as the image acquisition module, and is periodically or actively awakened to acquire point cloud data in the power transmission channel, thereby achieving real-time acquisition of three-dimensional data of the power transmission channel and construction of a three-dimensional space model.

4. The device for measuring distance outside the power transmission channel by fusing laser point cloud and image according to claim 1 is characterized in that: The image acquisition module and the laser point cloud acquisition module (3) are kept in the same position and are used to acquire image data of the power transmission channel and the main body. The night vision image acquisition module (2) is integrated with a starlight-level night vision lens to acquire color images of the power transmission channel and the main body at night, thereby achieving 24-hour uninterrupted monitoring.

5. A method for starting distance measurement outside a power transmission channel by integrating laser point cloud and image, characterized in that: It includes starting point cloud data collection, reconstructing point cloud data, and starting distance measurement methods.

6. The method for starting distance measurement outside the power transmission channel by fusing laser point cloud and image according to claim 5 is characterized in that: The starting point cloud data collection and reconstructing point cloud data comprises the following steps: (1) After the device is installed, the scene in the power transmission channel captured by the image module of the current device can be set as the initial image through local connection or remote connection; (2) The device uses a transmission line hidden danger identification model to identify hidden dangers in the transmission channel; (3) An optimized twin neural network is embedded in the device, and the optimized twin neural network is used to identify the difference between the current image and the initial image; (4) The device uses the transmission line hidden danger identification model to detect the type of different parts and determine whether there are changes in tree growth, topography and buildings in the transmission channel; (5) Calling the laser point cloud acquisition module to re-acquire point cloud data within the transmission line channel; (6) The laser point cloud data module in the device also supports scheduled start and remote start, and the operation and maintenance personnel configure it according to the operation and maintenance requirements.

7. The method for starting distance measurement outside the power transmission channel by fusing laser point cloud and image according to claim 5 is characterized in that: The starting ranging method comprises: (7) The device uses a transmission line hidden danger identification model to identify hidden dangers in the transmission channel; (8) When hidden dangers of tower cranes, cranes, and construction machinery channels are detected in the transmission channel, the spatial ranging algorithm is activated.

8. The method for starting distance measurement outside the power transmission channel by fusing laser point cloud and image according to claim 7 is characterized in that: The spatial ranging algorithm includes: The point cloud data and image data are integrated to construct a three-dimensional spatial data model; and the nearest distance from the hidden dangers inside and outside the transmission line channel to the power line or transmission tower is measured with the help of spatial Euclidean distance measurement and nearest neighbor search method.

9. The method for starting distance measurement outside the power transmission channel by fusing laser point cloud and image according to claim 8 is characterized in that: The construction of the three-dimensional space data model comprises the following steps: S1: extract feature values ​​from two-dimensional images; S2: construct the relationship between the 2D pixel coordinates of the feature point and its 3D spatial coordinates; S3: Calculate the rotation matrix and translation vector of the image coordinate system and the three-dimensional space coordinate system; S4: For image sequences, a dense point cloud is obtained by using a dense matching method after relative pose calculation between images and encryption of sparse points. The preprocessed scene dense point cloud is obtained after removing outliers and noise points. S5: transform the scene dense point cloud into the laser radar coordinate system according to the initial registration parameters, and the image dense point cloud and the laser point cloud are aligned; S6: Search the nearest corresponding points between the two point clouds with distance as the constraint, and calculate the initial mean square distance d1 between the two point clouds; S7: Calculate the rotation and translation parameters of the set of neighboring point pairs according to the least squares rule, and transform and adjust the pose of the dense point cloud according to the calculation results; S8: Calculate the mean square distance d2 between the nearest corresponding point sets of the two point clouds after transformation. If the difference between d2 and d1 is less than a given threshold, the two point clouds achieve optimal matching. Otherwise, continue to repeat steps S6 and S7 to perform the next iteration.

10. The method for starting distance measurement outside the power transmission channel by fusing laser point cloud and image according to claim 8, characterized in that: The measurement of the closest distance from the hidden danger inside and outside the transmission line channel to the power line or transmission tower includes: S9: using the embedded transmission channel external damage hidden danger identification algorithm to obtain the hidden danger target pixel area, and then obtain the coordinate point of the target in the transmission three-dimensional space; S10: traverse and calculate the nearest distance from the hidden danger target to the power line or transmission tower, quickly calculate the nearest distance from the external hidden danger target point area to the power line or transmission tower, adopt the idea of ​​space segmentation, and continuously search from the bottom to the top for the distance from the hidden danger target area to the power line.

Images