Insulator Contamination Diagnosis using Unsupervised Machine Learning

Overhead line insulators play a key role in high voltage power distribution and transmission systems. Due to unavoidable ambient conditions, these insulators are exposed to dust, moisture, rain, fog, etc. As a result, contaminated insulators lead to flashover and cause grid failures. For uninterrupted operations, leakage current monitoring of insulators is a very popular method. For the determination of insulator health, artificial pollution tests were carried out on 11 kV porcelain insulators by recording their leakage current dataset. K-means clustering algorithm was applied to those datasets to evaluate the severity of contamination.