Analytical Review of Machine Learning Techniques for Transmission Line Insulator Fault Detection and Classification

To guarantee that transmission lines operate normally, insulations are crucial. Regular inspections must be performed since the integrity of the insulating system affects the transmission system’s dependability. There are currently three primary categories of defect picture identification and detection research for aerial insulators: methods based on image processing, methods based on pattern recognition, and approaches based on machine learning. Automated fault identification has the potential to greatly reduce inspection time and associated expenses. This paper provides important highlights on research work carried out in recent past to detect insulator fault. The various approaches used for detection of insulator faults has been studied and comparative analysis is given based on different hyper-parameters like size of dataset, types of faults identified, techniques used and its performances. The comparative analysis shows that performance of machine learning based system can be improve by using complex design techniques for CNN model with large size of dataset. Based on comprehensive analysis, the novel CNN model based on hybrid pooling technique has been proposed for detection of multiple insulators faults.