Solar PV parameter forecast using generalized neural networks

Abstract

Solar photovoltaic (PV) is widely used in the world due to the increased demand for renewable energy. The environmental concern and depletion of natural energy sources are also a factor that leads to the shift to renewable solutions. An efficient operation of solar PV depends on the accuracy of the modelling and control of the module prior to the installation. This research work proposes a new neural network algorithm, Generalised Linear Hopfield Neural Network (GLHNN) to determine the maximum power point (MPP) of a solar photovoltaic (PV) module accurately under dynamic environmental condition, while comparing the accuracy with the commonly used Newton Raphson (NR) method. A single diode model is used and whose equivalent circuit parameters are derived from the mathematical expressions developed. The five parameters of the model such as the series resistance (R_se), shunt resistance (R_sh), ideality factor of the diode (A), light generated current (I_LG) and diode reverse saturated current (I_sat) are used to analyse the operating performance of PV panel and to estimate the maximum power at MPP.