Fault detection in photovoltaic arrays via sparse representation classifier

In recent years, there has been an increasing interest in the integration of photovoltaic (PV) systems in the power grids. Although PV systems provide the grid with clean and renewable energy, their unsafe and inefficient operation can affect the grid reliability. Early stage fault detection plays a crucial role in reducing the operation and maintenance costs and provides a long lifespan for PV arrays. PV Fault detection, however, is challenging especially when DC short circuit occurs under the low irradiance conditions while the arrays are equipped with an active maximum power point tracking (MPPT) mechanism. In this case, the efficiency and power output of a PV array decrease significantly under hard-to-detect faults such as active MPPT and low irradiance. If the hard-to-detect faults are not detected effectively, they will lead to PV array damage and potential fire hazard. To address this issue, in this paper we propose a new sparse representation classifier (SRC) based on feature extraction to effectively detect DC short circuit faults of PV array. To verify the effectiveness of the proposed SRC fault detection method, we use numerical simulation and compare its performance with the artificial neural network (ANN) based fault detection.