LAUSR

Strong ultraviolet (UV) radiation is one of the main factors that lead to the aging of composite insulators of transmission lines in high-altitude areas, which has brought security risks to the operation of power systems. The aging status information obtained by traditional insulation surface aging detection methods is indirect and limited, or time-consuming and labor-intensive. In this article, a non-contact non-destructive testing evaluation method based on hyperspectral technology is proposed to quickly detect the UV aging state of composite insulator surface. First, the hyperspectral imaging platform is used to image a number of continuous bands on the insulator surface. Combining the aging calibration data, a locally reserved projection-convolutional neural network (LPP-CNN) model is established to extract the characteristic bands of the aging surface and provide a way to accurately evaluate the surface UV aging degree of insulators. The model has a high resolution of aging degree, which provides a new idea and theoretical basis for realizing non-contact non-destructive online detection of surface aging under UV light.