Discrimination of transmission line insulator contamination grades using visible light images

Flashover occurs more easily on contaminated transmission line insulators, which causes great economic losses and has bad effects on power system stability. An accurate and safe detection of insulator contamination grades is required. In this paper, a new method is proposed to discriminate insulator contamination grades using visible light images. Firstly, ZSW-10/4 insulators are smeared in different contamination grades, and their visible light images are shot under illumination ranges from 10,000lux to 100,000lux. Secondly, both software methods and hardware methods are adopted and compared to eliminate the effects of illumination. After image processing, insulator surface color features in several color spaces are calculated. Results of Fisher criterion shows that hardware methods work better in eliminating illumination effects, and mean value of V component in YUV color space is selected for discriminating contamination grades. Finally, BP (Back Propagation) neural networks are established, whose testing accuracy rates are over 90% in discriminating contamination grades. Further, a general formula between mean value of V component and ESDD (Equivalent Salt Deposit Density) is obtained.