Voltage Sag Classification Based on PDT-SVM

Lightning strike failure is one of the causes of voltage sags. This paper analyzes the characteristics of the type of lightning strike sag and summarizes the characteristics of 4 sag types' voltage RMS waveforms including short-circuit fault, lightning strike, transformer switching, induction motor starting, and introduces five kinds of voltage sag feature indicators to establish the voltage sag identification feature matrix. Then, the optimal decision tree support vector machine (DT-SVM) classifier based on particle swarm clustering is used to identify the 4 types of sag. Training data and test data are from the grid voltage sag monitoring terminal. The analysis results verify the effectiveness and superiority of the algorithm.