CN-122024175-A - Power channel tree obstacle early warning method and device, terminal equipment and storage medium

CN122024175ACN 122024175 ACN122024175 ACN 122024175ACN-122024175-A

Abstract

The application discloses a power channel tree obstacle early warning method, a device, terminal equipment and a storage medium, belonging to the field of power equipment monitoring, wherein the method comprises the following steps: acquiring point cloud data of a power channel at a current time point, screening current tree obstacle early warning points from the point cloud data, and extracting first space appearance features from each current tree obstacle early warning point; the method comprises the steps of obtaining historical tree obstacle early warning points of a power channel at historical time points, extracting second space appearance features from the historical tree obstacle early warning points, performing similarity matching on first space appearance features and second space appearance features, generating a plurality of early warning point sequences according to matching results, predicting the remaining time of each corresponding risk tree growing until the first space distance between each corresponding risk tree and a wire is smaller than a first preset safety distance according to the early warning point sequences, and sending an early warning report when the remaining time corresponding to any risk tree is smaller than a preset time threshold. The application can improve the safety of the operation process of the power channel.

Inventors

DING CHENXING

Assignees

广东电网有限责任公司广州供电局

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The utility model provides a power channel tree obstacle early warning method which is characterized in that the method comprises the following steps: Acquiring point cloud data of a power channel at a current time point, screening current tree obstacle early warning points from the point cloud data, and extracting first space appearance features from each current tree obstacle early warning point; acquiring historical tree obstacle early warning points of the power channel at historical time points, and extracting second space appearance features from each historical tree obstacle early warning point; Performing similarity matching on the first spatial appearance characteristic and the second spatial appearance characteristic, and generating a plurality of early warning point sequences according to a matching result, wherein each early warning point sequence corresponds to a risk tree, and each early warning point sequence comprises one current tree obstacle early warning point and a plurality of historical tree obstacle early warning points; And predicting the residual time of each corresponding risk tree growing until the first space distance between the corresponding risk tree and the lead is smaller than a first preset safety distance according to the early warning point sequence, and sending an early warning report when the residual time of any corresponding risk tree is smaller than a preset time threshold.
2. The method for early warning tree obstacle in power channel according to claim 1, wherein the step of obtaining the point cloud data of the power channel at the current time point and screening the current tree obstacle early warning point from the point cloud data comprises the steps of: extracting a wire point cloud and a tree point cloud from the point cloud data through semantic segmentation; performing space curve fitting on the wire point cloud to obtain a wire curve model; and calculating a second space distance between each cloud point in the tree point cloud and the wire curve model, and taking the point cloud with the second space distance smaller than the first preset safety distance as the current tree obstacle early warning point.
3. The power channel tree obstacle early warning method according to claim 1, wherein the first spatial appearance feature comprises a first spatial coordinate and a first color feature, the second spatial appearance feature comprises a second spatial coordinate and a second color feature, the similarity matching is performed on the first spatial appearance feature and the second spatial appearance feature, and a plurality of early warning point sequences are generated according to a matching result, and the method comprises the following steps: calculating a third spatial distance between each first spatial coordinate and each second spatial coordinate, and screening a plurality of candidate matching points corresponding to the current tree obstacle early-warning points from the historical tree obstacle early-warning points according to the third spatial distance aiming at each current tree obstacle early-warning point; Calculating cosine similarity between each first color feature and a second color feature corresponding to each candidate matching point; based on the third spatial distance and the cosine similarity, performing optimal assignment on each current tree-obstacle early-warning point and each corresponding candidate matching point by using a Hungary algorithm to determine a historical tree-obstacle early-warning point matched with each current tree-obstacle early-warning point; And generating an early warning point sequence corresponding to each risk tree according to each current tree obstacle early warning point and the corresponding matched historical tree obstacle early warning point.
4. The power channel tree obstacle warning method according to claim 3, wherein the screening the plurality of candidate matching points corresponding to the current tree obstacle warning point from the historical tree obstacle warning points according to the third spatial distance includes: Taking each second space coordinate as an original data set; Taking the original data set as a current data set, iteratively dividing the current data to generate a binary search tree, selecting the dimension with the largest variance in the current data set as a dividing axis in each iteration process, dividing the current data set into a left subspace and a right subspace based on the median of all coordinate values on the corresponding dimension of the dividing axis as a dividing point, and taking the left subspace and the right subspace as new data sets to enter the next iteration until the current data set only has single first space coordinates; and taking the first space coordinates as query points, executing radius search by using the binary search tree, and determining a plurality of candidate matching points by traversing nodes with the distance from the query points smaller than a second preset distance.
5. The power channel tree obstacle warning method according to claim 3, wherein the determining the historical tree obstacle warning points matched with each current tree obstacle warning point by optimally assigning each current tree obstacle warning point and each corresponding candidate matching point by using a hungarian algorithm based on the similarity between the third spatial distance and the cosine comprises: Obtaining the matching cost between each current tree obstacle early warning point and each corresponding candidate matching point by weighting and fusing the third space distance and the cosine similarity; Constructing an assignment benefit matrix between a current time point and a historical time point based on the matching cost, and performing iterative operation on the assignment benefit matrix by using the Hungary algorithm to obtain an optimal assignment scheme for minimizing the total global matching cost; And according to the optimal assignment scheme, determining candidate matching points with a one-to-one matching relationship corresponding to each current tree obstacle early warning point.
6. The method for pre-warning the power channel tree obstacle according to claim 1, wherein predicting the remaining time for each corresponding risk tree to grow to a first spatial distance from the wire less than a first preset safety distance according to the pre-warning point sequence comprises: smoothing each early warning point sequence by adopting a Kalman filter to remove observation noise in the early warning point sequence and output a basic growth speed which corresponds to the optimal estimation of the risk tree; Acquiring environmental factor data of the power channel in a preset period, and acquiring a growth prediction model for representing the mapping relation between the actual growth speed of the tree and the environmental factor data; Inputting the basic growth speed and the environmental factor data into the growth prediction model, so that the growth prediction model corrects the basic growth speed according to the environmental factor data to obtain the actual growth speed; and calculating a difference value between the first preset safety distance and the first space distance, and calculating the residual time according to the difference value and the actual growth speed.
7. The power channel tree obstacle early warning device is characterized by comprising an early warning point screening module, a characteristic extraction module, an early warning point matching module and an early warning report generation module; The early warning point screening module is used for acquiring point cloud data of the power channel at a current time point, screening current tree obstacle early warning points from the point cloud data, and extracting first space appearance features from each current tree obstacle early warning point; the feature extraction module is used for obtaining historical tree obstacle early warning points of the power channel at historical time points and extracting second space appearance features from the historical tree obstacle early warning points; The early warning point matching module is used for carrying out similarity matching on the first spatial appearance characteristic and the second spatial appearance characteristic and generating a plurality of early warning point sequences according to a matching result, wherein each early warning point sequence corresponds to a risk tree, and each early warning point sequence comprises one current tree obstacle early warning point and a plurality of historical tree obstacle early warning points; And the early warning report generation module is used for predicting the residual time of each corresponding risk tree growing until the first space distance between the corresponding risk tree and the lead is smaller than a first preset safety distance according to the early warning point sequence, and sending an early warning report when the residual time of any corresponding risk tree is smaller than a preset time threshold.
8. The power channel tree obstacle early warning device according to claim 7, wherein the early warning point screening module comprises a semantic segmentation unit, a curve fitting unit and a distance comparison unit, and the early warning point screening module is used for acquiring point cloud data of a power channel at a current time point, screening current tree obstacle early warning points from the point cloud data, and comprises the following steps: the semantic segmentation unit is used for extracting wire point clouds and tree point clouds from the point cloud data through semantic segmentation; The curve fitting unit is used for performing space curve fitting on the wire point cloud to obtain a wire curve model; The distance comparison unit is used for calculating a second space distance between each cloud point in the tree point cloud and the wire curve model, and taking the point cloud with the second space distance smaller than the first preset safety distance as the current tree obstacle early warning point.
9. A terminal device comprising a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, the processor implementing a power channel tree obstacle warning method according to any one of claims 1 to 6 when the computer program is executed.
10. A computer readable storage medium, characterized in that the computer readable storage medium comprises a stored computer program, wherein the computer program when run controls a device in which the computer readable storage medium is located to perform a power channel tree obstacle warning method according to any one of claims 1 to 6.

Description

Power channel tree obstacle early warning method and device, terminal equipment and storage medium Technical Field The application relates to the field of power equipment monitoring, in particular to a power channel tree obstacle early warning method, a device, terminal equipment and a storage medium. Background As a key infrastructure, the power transmission network needs to ensure safe and stable operation. In the operation and maintenance of overhead transmission lines, overgrowth of trees in a channel is one of the main natural factors that induce line tripping, initiate forest fires, and even cause large-scale power failure accidents of the power grid. In order to ensure that a sufficient safety distance is maintained between the lead and vegetation, the power transmission channel needs to be inspected, and potential tree obstacle risks are timely identified and treated, so that the power supply reliability of a power grid is ensured. The existing detection technology based on the point cloud data still belongs to a static instantaneous observation mode in nature. Although the tree obstacle hidden trouble during single inspection can be accurately identified, due to the lack of a cross-period association mechanism for observing targets in different periods, the inspection data in each period form isolated information islands. Such breaks between data make it difficult for the prior art to predict the probability of a risk point evolving into a true fault in the future, thereby affecting the safety during operation of the power channel. Disclosure of Invention The application provides a power channel tree obstacle early warning method, a device, terminal equipment and a storage medium, which can solve the problem that the safety of a power channel operation process is affected by the probability that a risk point is difficult to predict and evolve into a real fault in the future in the prior art. Some embodiments of the present application provide a power channel tree obstacle early warning method, including: Acquiring point cloud data of a power channel at a current time point, screening current tree obstacle early warning points from the point cloud data, and extracting first space appearance features from each current tree obstacle early warning point; acquiring historical tree obstacle early warning points of the power channel at historical time points, and extracting second space appearance features from each historical tree obstacle early warning point; Performing similarity matching on the first spatial appearance characteristic and the second spatial appearance characteristic, and generating a plurality of early warning point sequences according to a matching result, wherein each early warning point sequence corresponds to a risk tree, and each early warning point sequence comprises one current tree obstacle early warning point and a plurality of historical tree obstacle early warning points; And predicting the residual time of each corresponding risk tree growing until the first space distance between the corresponding risk tree and the lead is smaller than a first preset safety distance according to the early warning point sequence, and sending an early warning report when the residual time of any corresponding risk tree is smaller than a preset time threshold. Compared with the prior art, the method has the advantages that the current tree obstacle early warning points with real threat can be accurately screened out in the inspection stage by acquiring the wire curve model and combining the tree point cloud data at the current moment and taking the space distance between the tree cloud points and the wires as a constraint condition, so that the interference of irrelevant trees on subsequent analysis is avoided. On the basis, tree obstacle early warning points of historical time points are further introduced, and spatial appearance features are uniformly extracted aiming at the current and historical early warning points, so that point cloud observation results in different periods have comparability, and a cross-period association relationship is established. By matching the current early warning point with the historical early warning point to form an early warning point sequence aiming at a single-plant risk tree, the application can describe the continuous change trend of the tree relative to the position of the wire in the growth process, and is not limited to the instantaneous state at a certain moment. Further, the residual time of tree growth to the dangerous distance threshold value is predicted based on the early warning point sequence, so that tree obstacle risk is converted from 'whether existing' to 'when occurring', and early warning reports are output in advance when the residual time is lower than the preset threshold value, and therefore the prospective and operation safety of power channel tree obstacle risk assessment are remarkably improved. Further, the obtaining the point