Surface discharge detection method of contaminated insulators based on ultraviolet images' parameters

Insulator flashover is a great threat of power system's safe and stable operation and surface discharge detection of contaminated insulators is an effective approach of flashover warning. Surface discharge detection method of contaminated insulators based on ultraviolet images' parameters is proposed in this paper. Experiment system of contaminated insulators surface discharge is designed to acquire discharge magnitude and ultraviolet(UV) images synchronously. After frame image process of UV videos, we define average facula area in 100 successive UV image frames as UV images' feature parameter. As shooting parameters vary in practical application, effects of imager gain and shooting distance on facula area under the same discharge magnitude are also studied. Then a model of support vector regression is built to predict discharge magnitude using the UV image parameter. And genetic algorithm is adopted to optimize kernel parameters and punishment parameters. The regression result indicates that discharge magnitude calculated by UV image only have an error of less than 9% compared with measured discharge magnitude. It meets the accuracy requirement of engineering application and thus provide a new surface discharge detection method of contaminated insulators based on ultraviolet images' parameters.