Superiority of decision-tree classifier on complicated cases for power system protection

Diagnosis of power system faults requires the identification and classification of voltage disturbances in power systems and smart grids. The objective of this approach is to develop state-of-the-art signal classification algorithms for classifying different types of power quality disturbances (faults) based on latest improvements in signal processing and pattern recognition techniques. This presentation proposes a new solution for power system monitoring against all possible power quality issues. The S-transform is used for analysing distorted power signals. As a classifier, a decision-tree algorithm is used, and its performance is compared to other classifiers. The proposed hybrid power-system monitoring system is able to detect common power system disturbances, such as voltage sag/swell/flicker, the DC component, electromagnetic interference, harmonics, transients and blackouts.