Estimating levels of surface contamination of silicone rubber insulators by employing wavelet decomposition and machine learning-based techniques

An evolutionary strategy of green evaluation in the power system as well as in the era of artificial intelligence (AI) is machine learning (ML). Evolutionary techniques are employed in various fields of research. One of the important applications of evolutionary techniques is in the field of condition monitoring study of overhead line insulators. Insulators of overhead lines play a significant role in power system networks for power transmission and distribution. The power may go out completely if the overhead line insulators fail to operate. It is therefore effective to monitor overhead line insulators through surface leakage current (SLC) studies. The study employs a laboratory configuration to capture leakage current signals from multiple contaminated surfaces of silicone rubber (SiR) insulators. As a result of analyzing the SLC data, some important features about contamination levels can be extracted. Additionally, based on these characteristics, machine learning-based eco-friendly evaluation techniques are used to identify contamination classes in SiR insulators. To extract and classify features, a signal analysis tool that is wavelet decomposition and a few popular machine learning classifiers (i.e., random forest, support vector machine, Gaussian naive Bayes, and k-nearest neighbor), are employed in this work. A comparative analysis was performed to predict the degradation levels of the SiR insulators using individual ML classifiers employing advanced evaluation models in this work. This chapter determined the optimal factors that can predict the extent of dirt accumulation on SiR insulators surface.