Artificial neural network model of photovoltaic generator for power flow analysis in PSSSINCAL

Output of a three-phase photovoltaic generator (PVG) is a function of sunlight irradiance, temperature, and three-phase terminal voltage phasors. Three-phase PVGs are largely connected to rural distribution systems feeders that are predominantly unbalanced. Models of PVGs that are only a function of sunlight irradiance and temperature disregarding unbalanced three-phase terminal voltages phasors are simple to use with three-phase power flow analysis but yield inaccurate solutions. Detailed three-phase PVG models are complex and non-linear, hence unsuitable for power flow analysis applications. This study proposes an artificial neural network (ANN) model to represent a PVG comprising photovoltaic panels, a boost chopper and a three-phase inverter. Main advantages of the ANN model are that it can be readily used to model a PVG of any size and type, mathematical simplicity, high accuracy with unbalanced systems and computational speed. The model was tested with the unbalanced distribution system feeder from a Canadian utility. The results show that the ANN model of a PVG is computationally fast and more accurate than simple model that ignores unbalanced three-phase terminal voltage phasors. In addition, simplicity of the proposed ANN model of PVG allows easy integration into commercial software packages such as PSSSINCAL as reported in this study.