Disturbances Classification based on a Model Order Selection Method for Power Quality Monitoring

Abstract : In this paper, a new technique for power quality disturbances classification is proposed. It focuses on voltage sags and swells that are first pre-classified into four classes that depend on the number of non-zero symmetrical components and can contain different types of sag and swell. Using the estimated symmetrical component values, we can afterward classify the corresponding sag or swell signature. In this study, we show that the pre-classification can be reformulated as a pure model order selection problem. To solve this problem, we propose two pre-classifiers based on Information Theoretical Criteria. The former yields the highest statistical performances, while the latter has a lower computation complexity. The performances of the proposed classification algorithms are evaluated using Monte Carlo simulations on synthetic signals and using real power system data obtained from the DOE/EPRI National Database of Power System Events. The achieved simulations and experimental results clearly illustrate the effectiveness of the proposed algorithms for voltage sag and swell classification.