Optimal Feature and Decision Tree-Based Classification of Power Quality Disturbances in Distributed Generation Systems

Penetration of distributed generation systems in conventional power systems leads to power quality (PQ) disturbances. This paper provides an improved PQ disturbances classification, which is associated with load changes and environmental factors. Various forms of PQ disturbances, including sag, swell, notch, and harmonics, are taken into account. Several features are obtained through hyperbolic S-transform, out of which the optimal features are selected using a genetic algorithm. These optimal features are used for PQ disturbances classification by employing support vector machines (SVMs) and decision tree (DT) classifiers. The study is supported by three different case studies, considering the experimental setup prototypes for wind energy and photovoltaic systems, as well as the modified Nordic 32-bus test system. The robustness and precision of DT and SWM are performed with noise and harmonics in the disturbance signals, thus providing comprehensive results.