CN-121982622-A - Safety risk determination method and device for objects near cable and robot

CN121982622ACN 121982622 ACN121982622 ACN 121982622ACN-121982622-A

Abstract

The application provides a safety risk determining method, a device and a robot for objects near a cable, which comprise the steps of obtaining point cloud data of a power transmission line corridor to obtain initial point clouds, conducting semantic segmentation on the initial point clouds by means of a pre-trained semantic segmentation model to obtain first point clouds corresponding to the power transmission line and second point clouds corresponding to underground objects, determining clearance distances between the power transmission line and the underground objects according to the first point clouds and the second point clouds by means of a plane orthogonal projection fitting method, and determining whether the underground objects have safety risks according to the clearance distances and the maximum safety distances. The method solves the problem that dangerous points are not diagnosed accurately because the method based on clustering and image recognition in the prior art cannot accurately recognize the power transmission line because the point cloud acquired by the power transmission line corridor is irregular.

Inventors

LI SHANGGUO
ZHANG LANTAO
JIAO YUYANG
Men Haochen
ZHOU XIAONAN

Assignees

北京卓越电力建设有限公司
国网北京市电力公司

Dates

Publication Date: 20260505
Application Date: 20251111

Claims (10)

1. A method of determining a security risk of an object in proximity to a cable, comprising: acquiring point cloud data of a power transmission line corridor to obtain an initial point cloud; performing semantic segmentation on the initial point cloud by using a pre-trained semantic segmentation model to obtain a first point cloud corresponding to the power transmission line and a second point cloud corresponding to the underground object; Determining the clearance distance between the power transmission line and the underground object according to the first point cloud and the second point cloud by adopting a plane orthogonal projection fitting method; And determining whether the underground object has safety risk according to the clearance distance and the maximum safety distance.
2. The method of claim 1, wherein semantically segmenting the initial point cloud data using a pre-trained semantic segmentation model to obtain a first point cloud corresponding to the transmission line and a second point cloud corresponding to the underground object, comprising: performing network sampling, local feature coding, global feature extraction and up-sampling on the initial point cloud through RandLA-Net models in the semantic segmentation model to obtain a first target feature, wherein the semantic segmentation model is obtained based on SSLNet model training; Extracting geometric features and semantic features from the first target features through a local aggregation layer of the semantic segmentation model to obtain second target features; and processing the second target features through the shared multi-layer perceptron of the semantic segmentation model to obtain the first point cloud and the second point cloud.
3. The method of claim 1, wherein determining a headroom distance between the transmission line and the subsurface object from the first point cloud and the second point cloud using a planar orthogonal projection fitting method comprises: projecting the first point cloud to an xoy plane to obtain a first line point, and fitting according to the first line point to obtain a first projection line; Projecting the second point cloud to an xoy plane to obtain a first object point; Projecting the first point cloud to a xoz plane to obtain a second line point, and fitting according to the second line point to obtain a second projection line; projecting the second point cloud to a xoz plane to obtain a second object point; And calculating the clearance distance according to the first projection line, the first object point, the second projection line and the second object point.
4. A method according to claim 3, wherein fitting based on the first line points results in a first projection line, comprising: Clustering the first line points by adopting a DBSCAN algorithm to obtain a plurality of cluster clusters, wherein the cluster clusters are in one-to-one correspondence with the power transmission lines; Fitting is carried out according to the first line points in each cluster, and the corresponding first projection line is obtained.
5. The method of claim 4, wherein fitting each of the first line points in each cluster to obtain the corresponding first projection line comprises: constructing an equation according to the coordinates of the first line point in the cluster by adopting a least square method, wherein the equation is that Wherein n is the number of the first line points in the cluster, k is the highest degree of the polynomial, And The x-axis coordinate and the y-axis coordinate of the ith first line point on the xoy plane are respectively, ... Is a polynomial coefficient; And solving an equation to obtain a corresponding target polynomial coefficient, and constructing a curve equation according to the target polynomial coefficient to obtain the first projection line.
6. A method according to claim 3, wherein calculating the headroom distance from the first projection line, the first object point, the second projection line, and the second object point comprises: Calculating the vertical distance between the first projection line and the first object point to obtain a first target distance Y'; calculating the vertical distance between the second projection line and the second object point to obtain a second target distance Z'; According to the first target distance Y 'and the second target distance Z' through a formula Calculating the clearance distance 。
7. The method of claim 1, wherein obtaining point cloud data for the transmission line corridor to obtain an initial point cloud comprises: Controlling a robot to walk along the power transmission line corridor, and controlling a laser space scanner arranged on the robot to scan to obtain an alternative point cloud; And downsampling the alternative point cloud to obtain the initial point cloud.
8. A security risk determination device for an object in the vicinity of a cable, the device comprising: The acquisition unit is used for acquiring point cloud data of the transmission line corridor to obtain an initial point cloud; the processing unit is used for carrying out semantic segmentation on the initial point cloud by utilizing a pre-trained semantic segmentation model to obtain a first point cloud corresponding to the power transmission line and a second point cloud corresponding to the underground object; The computing unit is used for determining the clearance distance between the power transmission line and the underground object according to the first point cloud and the second point cloud by adopting a plane orthogonal projection fitting method; And the diagnosis unit is used for determining whether the underground object has safety risk according to the clearance distance and the maximum safety distance.
9. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored program, wherein the program, when run, controls a device in which the computer readable storage medium is located to perform the method of any one of claims 1 to 7.
10. A robot comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs comprising instructions for performing the method of any of claims 1-7.

Description

Safety risk determination method and device for objects near cable and robot Technical Field The invention relates to the technical field of power transmission line inspection, in particular to a method and a device for determining safety risk of objects nearby a cable, a computer readable storage medium and a robot. Background The underground cable network is used as urban energy artery and communication nerve, and the safe operation of the underground cable network is directly related to the stability of an urban life line system. Traditional manual inspection mode faces the bottlenecks such as inefficiency, monitoring blind area are many, data accuracy is not enough, especially in complicated underground environment, and manual inspection is difficult to realize high-frequency, full-coverage accurate monitoring. In order to improve efficiency and safety, new methods are needed to replace manual inspection. To solve the above problems, the use of a robot dog in combination with a ranging technique (cable corridor laser space scanner) has become common in cable inspection and transmission line planning. And how to extract useful information from the detected data is important based on cable detection by the robot dog cable corridor laser space scanner. In the prior art, the original point cloud in the original grid is preprocessed, residual fitting is applied to find cable points, and then a double-k-means algorithm is used for reconstructing and fitting a cable line, but because the clustering method is applied to uneven point cloud density, irregular clustering is easy to cause, the performance of the clustering algorithm depends on an initial clustering center, and the point cloud data of a power transmission line corridor is difficult to determine a good initial clustering center, so that a better clustering cluster cannot be formed, and the accuracy of automatic inspection is difficult to guarantee. Disclosure of Invention The application mainly aims to provide a safety risk determining method, a safety risk determining device, a safety risk determining computer-readable storage medium and a safety risk determining robot for objects near a cable, so as to at least solve the problem that dangerous points are inaccurately diagnosed due to the fact that point clouds collected by a corridor of a power transmission line are irregular, and the method based on clustering and image recognition in the prior art cannot accurately identify the power transmission line. In order to achieve the above object, according to one aspect of the present application, there is provided a security risk determining method for an object near a cable, including obtaining point cloud data of a corridor of a power transmission line to obtain an initial point cloud, performing semantic segmentation on the initial point cloud by using a pre-trained semantic segmentation model to obtain a first point cloud corresponding to the power transmission line and a second point cloud corresponding to an underground object, determining a clearance distance between the power transmission line and the underground object by using a planar orthogonal projection fitting method according to the first point cloud and the second point cloud, and determining whether the underground object has a security risk according to the clearance distance and a maximum security distance. Optionally, semantic segmentation is carried out on initial point cloud data by utilizing a pre-trained semantic segmentation model to obtain a first point cloud corresponding to a power transmission line and a second point cloud corresponding to an underground object, wherein the method comprises the steps of carrying out network sampling, local feature coding, global feature extraction and up-sampling on the initial point cloud through a RandLA-Net model in the semantic segmentation model to obtain a first target feature, carrying out geometric feature and semantic feature extraction on the first target feature through a local aggregation layer of the semantic segmentation model to obtain a second target feature, and processing the second target feature through a shared multi-layer perceptron of the semantic segmentation model to obtain the first point cloud and the second point cloud. The method comprises the steps of projecting a first point cloud to an xoy plane to obtain a first line point, fitting according to the first line point to obtain a first projection line, projecting a second point cloud to the xoy plane to obtain a first object point, projecting the first point cloud to a xoz plane to obtain a second line point, fitting according to the second line point to obtain a second projection line, projecting the second point cloud to a xoz plane to obtain a second object point, and calculating the clearance distance according to the first projection line, the first object point, the second projection line and the second object point. The method comprises the steps of clu