CN-121998898-A - Unmanned aerial vehicle hyperspectral-based insulator aging intelligent detection method and system

CN121998898ACN 121998898 ACN121998898 ACN 121998898ACN-121998898-A

Abstract

The invention discloses an insulator aging intelligent detection method and system based on hyperspectrum of an unmanned aerial vehicle, wherein the method comprises the steps of completing radiation, spectrum, camera internal reference and time synchronous calibration, flying according to a preset air route, synchronously collecting data, unifying timestamp marks, carrying out radiation correction on hyperspectral data, carrying out distortion correction on visible light images, establishing a one-to-one correspondence of the data through timestamp matching, constructing insulator string sparse three-dimensional point cloud based on visible light image characteristic points, realizing coarse registration of the hyperspectral data and the three-dimensional point cloud, carrying out iterative optimization by taking the insulator characteristic points as references, correcting terrain deviation according to a digital elevation model, ensuring that registration errors meet threshold pixels, segmenting and extracting a single insulator, establishing a local coordinate system, carrying out geometric correction according to actual space gestures, eliminating oblique deflection distortion, extracting corrected insulator spectrum characteristics and three-dimensional coordinates, judging aging levels, and outputting a detection report containing accurate positioning.

Inventors

Xu Dongzi
ZHANG XUEKUN
SONG JIANMING
LIU WENJU
WEI WEI
CEN ZHONGYANG
LI SUHUA
SUN BIN
ZHANG DIANKE
LI ZHIXIAO
ZHANG DELU
ZHANG DONG
WANG HONGLI
ZHOU YANJUN
LI ZHAOGUO
LI JIE
ZHAO SHUHUI
ZHENG YI
QU XINQI
AN XIAOJING
LI TAO
TANG MINGWEI
TANG YAZHOU

Assignees

国网河南省电力公司柘城县供电公司

Dates

Publication Date: 20260508
Application Date: 20251209

Claims (10)

1. The intelligent insulator aging detection method based on the hyperspectrum of the unmanned aerial vehicle is characterized by comprising the following steps of: The unmanned aerial vehicle is provided with a hyperspectral imager, a visible light camera, an IMU and a GPS module to complete synchronous calibration of radiation, spectrum, camera internal parameters and time; Performing radiation correction on the hyperspectral data, and performing distortion correction on the visible light image, wherein a one-to-one correspondence of the data is established through time stamp matching; based on visible light image characteristic points and IMU-GPS data, constructing an insulator string sparse three-dimensional point cloud, and realizing coarse registration of hyperspectral data and the three-dimensional point cloud; taking the insulator characteristic points as reference iteration optimization, correcting terrain deviation according to the digital elevation model, and ensuring that the registration error meets a threshold pixel; A single insulator is extracted in a segmentation way, a local coordinate system is established, geometric correction is carried out according to the actual space posture, and inclination deflection distortion is eliminated; And extracting the corrected spectral characteristics and three-dimensional coordinates of the insulator, judging the ageing grade, and outputting a detection report containing accurate positioning.
2. The intelligent detection method for insulator aging based on hyperspectral of unmanned aerial vehicle is characterized in that the unmanned aerial vehicle is provided with a hyperspectral imager, a visible light camera, an IMU and a GPS module to complete radiation, spectrum, camera internal parameters and time synchronization calibration, specifically, the unmanned aerial vehicle is provided with the hyperspectral imager, the visible light camera, the IMU and the GPS module to form an integrated detection system, the hyperspectral imager is subjected to radiation and spectrum calibration to determine the corresponding relation between pixels and spectrum wavelengths, the visible light camera is subjected to internal parameter calibration to obtain relevant parameters, the IMU and the GPS module are subjected to time synchronization calibration, the installation angle parameters of a cloud platform of the unmanned aerial vehicle are recorded, and a spatial position correlation model among all devices is established.
3. The intelligent detection method for the aging of the insulator based on the hyperspectrum of the unmanned aerial vehicle is characterized in that the unmanned aerial vehicle flies according to a preset route, data are synchronously collected, and a uniform time stamp is marked, specifically, the unmanned aerial vehicle is controlled to fly along the preset route of a power transmission line, an integrated detection system is started to synchronously collect hyperspectral cube data, visible light sequence images, flight attitude data and space position data of the insulator, and all collected data are marked and stored in a uniform time stamp.
4. The intelligent detection method for insulator aging based on the hyperspectral of the unmanned aerial vehicle is characterized in that radiation correction is carried out on hyperspectral data, specifically, based on a radiation calibration result of a hyperspectral imager, an original digital quantized value of the hyperspectral data is converted into a radiation brightness value with unified dimension, atmospheric related scattering interference is eliminated through atmospheric scattering correction, dark current correction deduction instrument electronic noise is carried out, and flat field correction is adopted to correct brightness deviation caused by uneven pixel response, so that corrected data truly reflect the spectral radiation characteristics of the surface of the insulator.
5. The unmanned aerial vehicle hyperspectral-based insulator aging intelligent detection method is characterized in that distortion correction is carried out on the visible light image, specifically, based on calibration of an internal reference of a visible light camera, a correction model coupling radial distortion and tangential distortion is adopted to process a visible light sequence image, an original pixel coordinate is extracted, the original pixel coordinate is converted into a normalized imaging coordinate, radial offset and tangential offset are corrected, the original pixel coordinate is restored into a pixel coordinate to generate a geometric distortion-free image, key feature point distortion correction precision is verified in the correction process, and a precise spatial feature basis is provided for subsequent three-dimensional point cloud construction and insulator string segmentation.
6. The unmanned aerial vehicle hyperspectral-based insulator aging intelligent detection method is characterized in that the one-to-one correspondence of data is established through timestamp matching, specifically, the timestamps of hyperspectral data after radiation correction, visible light images after distortion correction and IMU-GPS combined data are extracted, unified timestamp references of three types of data are used as cores, a timestamp sequence alignment algorithm is adopted for comparison, tiny time differences generated by response delay of positive equipment are eliminated, abnormal timestamp data are removed, timestamp association logs of effective matched data are recorded, and a structured dataset corresponding to the three types of data one by one is constructed.
7. The method for intelligently detecting the aging of the insulator based on the hyperspectrum of the unmanned aerial vehicle is characterized by constructing an insulator string sparse three-dimensional point cloud based on visible light image characteristic points and IMU-GPS data and realizing the rough registration of hyperspectral data and the three-dimensional point cloud, specifically comprises the steps of extracting matched stable characteristic points based on a distortion corrected visible light image, solving camera external parameters according to IMU-GPS data and holder installation parameters, generating a sparse three-dimensional point cloud containing the insulator string and hardware fitting by means of SfM fusion of the internal and external parameters, filtering noise, matching corresponding camera external parameters according to an inter-equipment space correlation model and relative installation parameters by means of a timestamp synchronization mechanism, converting hyperspectral data pixel coordinates to a global GPS coordinate system, and establishing initial space mapping.
8. The unmanned aerial vehicle hyperspectral-based insulator aging intelligent detection method is characterized by taking insulator characteristic points as reference iterative optimization, specifically comprising the steps of screening stable characteristic points from insulator string sparse three-dimensional point cloud as a target point set, extracting hyperspectral projection points after coarse registration as a source point set, calculating Euclidean distance based on ICP, screening matching point pairs, removing abnormal points, iteratively adjusting hyperspectral space parameters by taking average distance errors as a target function, updating the matching point pairs, setting termination conditions, and ensuring that registration errors reach standards.
9. The intelligent detection method for insulator aging based on the hyperspectral of the unmanned aerial vehicle, which is disclosed by claim 1, is characterized in that terrain deviation is corrected according to a digital elevation model, registration errors are guaranteed to meet threshold pixels, specifically, insulator string areas are divided and interference is eliminated based on visible light images after fine registration, a single insulator is divided according to three-dimensional point clouds, a dedicated local coordinate system is constructed by utilizing three-dimensional point cloud data of the insulator, corresponding hyperspectral data are geometrically corrected according to actual spatial postures of the insulator, distortion caused by inclination and deflection is eliminated, and corrected hyperspectral data pixels are enabled to accurately correspond to the actual surface positions of the insulators.
10. The intelligent detection method for the aging of the insulator based on the hyperspectral of the unmanned aerial vehicle is characterized by comprising the steps of extracting spectral features and three-dimensional coordinates of the corrected insulator, judging an aging grade, outputting a detection report containing accurate positioning, specifically, extracting the multidimensional hyperspectral features of the single insulator after the three-dimensional geometric correction, synchronously acquiring the three-dimensional coordinates of the single insulator, converting the three-dimensional coordinates into global coordinates, inputting the spectral features and the three-dimensional coordinates into a trained model, judging the aging grade and the spatial position of an aging area of the insulator, summarizing the detection results, registering and correcting evaluation data of the insulator, and forming a structured detection report through three-dimensional visualization to provide support for the operation and maintenance of a power transmission line.

Description

Unmanned aerial vehicle hyperspectral-based insulator aging intelligent detection method and system Technical Field The invention belongs to the field of intelligent detection, and particularly relates to an intelligent detection method and system for insulator aging based on hyperspectrum of an unmanned aerial vehicle. Background With the development of society, especially the progress of scientific technology, the rapid development of social productivity is greatly promoted, and especially for some high-risk and high-precision detection, more advanced equipment is adopted for auxiliary data acquisition and detection, and in the electric power field, the ageing detection of insulators is a more common and frequent work, wherein in the electric power field, unmanned aerial vehicle equipment is most representative. Patent document with publication number CN120847115A discloses a method and a system for identifying and positioning an unmanned aerial vehicle target based on multispectral fusion. The method comprises the steps of firstly obtaining multispectral image data, then analyzing the multispectral image data in multiple phases into pixel-level data, identifying early-stage infection areas of diseases and insect pests, then constructing a geometric distortion correction model, determining the relative positions of the early-stage infection areas of the diseases and insect pests through multi-view spatial intersection calculation, and finally fusing real-time differential global navigation satellite system positioning data of an unmanned aerial vehicle and the relative positions of the early-stage infection areas of the diseases and insect pests to output absolute geographic coordinates of the early-stage infection areas of the diseases and insect pests. The mapping method and system based on the unmanned aerial vehicle remote sensing technology comprise the following steps of collecting multispectral image data of an unmanned aerial vehicle, carrying a multispectral camera on the bottom of the unmanned aerial vehicle, setting flight parameters of the unmanned aerial vehicle, collecting image data by the multispectral camera in the flight process of the unmanned aerial vehicle, carrying out radiation model correction and geometric correction on the collected image by image processing, eliminating illumination change and geometric distortion, ensuring accurate physical meaning of the image data, matching the characteristics of the multiview images, establishing corresponding relations among images of different view angles, providing matching point pairs for three-dimensional reconstruction, reconstructing three-dimensional point cloud, recovering a scene three-dimensional structure from a two-dimensional image, and generating a dense point cloud model. In the insulator aging detection technology based on unmanned aerial vehicle hyperspectrum, hyperspectral data and unmanned aerial vehicle visible light images need to be synchronously acquired, and spatial registration needs to be carried out on the hyperspectral data and the unmanned aerial vehicle visible light images so as to realize fusion analysis of spectral characteristics and spatial positions. However, geometric distortion can occur in the registration process due to the change of the flying attitude (pitching, rolling and yawing) and the topography fluctuation of the unmanned aerial vehicle, if the registration precision is insufficient, the spectrum data of the same insulator are not matched with the space position, and the positioning of an aging area is affected. The insulator string is generally in a suspended type structure with multiple insulators connected in series, the angles and the distances of the different insulators are different, and the traditional geometric correction method based on plane assumption is not applicable any more. Therefore, the intelligent insulator aging detection method based on the hyperspectrum of the unmanned aerial vehicle is a problem worthy of research. Disclosure of Invention In order to solve the defects in the prior art, the invention provides an insulator aging intelligent detection method and system based on unmanned aerial vehicle hyperspectrum. The purpose of the invention is realized in the following way: an intelligent detection method for insulator aging based on hyperspectrum of unmanned aerial vehicle comprises the following steps: The unmanned aerial vehicle is provided with a hyperspectral imager, a visible light camera, an IMU and a GPS module to complete synchronous calibration of radiation, spectrum, camera internal parameters and time; Performing radiation correction on the hyperspectral data, and performing distortion correction on the visible light image, wherein a one-to-one correspondence of the data is established through time stamp matching; based on visible light image characteristic points and IMU-GPS data, constructing an insulator string sparse three-dimensional point cloud, and r