Wavelet optimized EWMA for fault detection and application to photovoltaic systems

Electrical power generation using photovoltaic (PV) became an active and continuous growing area for academic and industrial research. The complexity of PV systems and the increase in reliability requirement become a very important issue in automation. Grid-connected PV systems are among the top power technologies with the highest rate of development. Therefore, their proper operation and safe handling is a top priority. To respond for this exigency, we develop a novel technique for PV power systems monitoring. Various key variables can be monitored in PV systems, which include the voltage and frequency of the grid, the voltage and the current of the AC and DC converters, as well as climate data, such as the temperature and irradiance. Tight monitoring of these variables will provide more effective and less interrupted energy supplies. The developed monitoring method is applied and validated using simulated data of PV systems. The developed technique combines the advantages of Exponentially Weighted Moving Average (EWMA), multi-objective optimization (MOO) and Wavelet representation. The MOO is used here to solve the problem of choosing an optimal solution of the following two objective functions: (i) missed detection rate (MDR) and (ii) false alarm rate (FAR) where both of them are simultaneously minimized. Additionally, the use of wavelet representation improves the monitoring performances by reducing the MDR and FAR. The wavelet representation is applied to obtain precise deterministic characteristics besides decorrelation of autocorrelated measurements. The new proposed technique, called Wavelet Optimized EWMA (WOEWMA), is compared with the classical EWMA and Shewhart charts where they are used for detecting single and multiple faults (for example, Bypass, Mismatch, Mix and Shading faults). The performances of the monitoring scheme are evaluated using MDR and FAR indicators.