Evaluation method of composite insulator aging status based on hyperspectral imaging technology

Abstract

Silicone rubber insulators' long-term operation in complex environments can lead to accelerated aging, causing the risk of flashovers and grid outages. Timely detection of the insulators' aging status is crucial to ensure the stable operation of grid. Therefore, this paper proposes a pixel-level fast, non-contact detection method based on hyperspectral imaging (HSI) technology for insulators' aging state. Firstly, the hyperspectral spectral line of the samples' six aging levels was extracted, and 14 features were extracted using pre-processing methods like segmented principal component analysis (SPCA) to reduce irrelevant information interference. Secondly, the aging state classification model was established, and through comparison using overall accuracy, F1 score, Precision, and Recall, the random forest (RF) model demonstrated optimal performance with a classification accuracy of 96.81%. Finally, this study utilized HSI technology to achieve a pixel-level evaluation of composite insulators' aging status, and the technique's feasibility was verified through specific parameters such as roughness.