Modeling of soiled photovoltaic modules with neural networks using particle size composition of soil

Abstract

The performance of PV systems is said to be affected due to soiling predominantly in dry and arid regions. It is therefore necessary to develop methods for estimating the losses that occur due to soiling. For development of this model the particle size composition of the soil is taken as the quantifying parameter. Particle size composition was determined from Sieve Analysis. A series of experiments were conducted on PV panel by artificially soiling with five different soils taken from Shekhawati region of Rajasthan in India. A neural network based modelling of a soiled PV module using particle size composition is proposed. The experimental data obtained is then used to train and develop a neural network which is the approximate model of a soiled solar PV panel using which the power losses of a soiled panel can be predicted.