A classification based on random forest for partial discharge sources

The identification of Partial Discharge Sources (PD) is an important task in the monitoring and diagnosis of high voltage components, and the classification of their discharge sources is extremely important. In this paper, three major features of Partial Discharge Sources have been extracted and various machine learning algorithms are applied to classify them. The final experiments in implementing the classification of partial discharge sources show that Random Forest is more robust to noise compared to decision trees and AdaBoost, and runs at a speed comparable to AdaBoost.