Contents I. Introduction
Smart electricity meters are one of the basic equipment of smart grid data acquisition [1]. In recent years, intelligent fault diagnosis of power metering devices based on image has gradually become a development trend to ensure stable operation of the power grid [2]. However, when obtaining inspection images, due to the fact that lens optical imaging is not ideal for small hole imaging, there is a certain deviation between the actual imaging point on the imaging plane and the ideal imaging point, and there is a complex nonlinear relationship between the two, namely lens distortion [3]. Image distortion can lead to the inability of intelligent fault diagnosis models to accurately capture the features of measuring devices, resulting in a decrease in diagnostic efficiency. Therefore, distortion correction of images before fault diagnosis is also an important guarantee for intelligent fault diagnosis of power metering devices.
