Simple and efficient approach to detect and diagnose electrical faults and partial shading in photovoltaic systems

Abstract

Photovoltaic (PV) systems are continuously exposed to many potential faults, causing significant power generation losses. Accurate fault detection procedures are crucial to ensure the reliable operating condition. This paper presents simple, efficient and easy to implement approach to detect the most common failures, due to short-circuit (SC) and open-circuit (OC) faults, inverter disconnection (ID) and partial shading (PS). The proposed procedure introduces three indicators namely, current indicator α, voltage indicator β and power indicator γ with the main function to distinguish healthy and faulty operating conditions. To this end, single-diode model is adopted to generate a trusted PV model in combination with the best-so-far artificial bees colony (ABC) optimization algorithm in order to extract the unknown model parameters. Subsequently, the maximum power point (MPP) coordinates are estimated to mimic the real operating PV system. Measured and predicted MPP coordinates allow the calculation of current, voltage and power indicators. For each indicator, an upper and lower thresholds has been established by trial and error. The computed value of each indicator will reveal the healthy or faulty operation of the PV system when it is within or outside the predefined threshold respectively. An experimental evaluation is presented by using the monitored data of the 3.2 kWp grid-connected PV system located at the roof of the Renewable Energy Development Centre (CDER), Algeria.