ANN based controller to mitigate soiling loss on solar panels

Abstract

The hasty augmentation in PV cells technology made the installation and utilization of solar power very trouble-free. The solar cells fabricated by semiconductors, has reduced the efficiency of PV systems to 15–20%. Further, the performance of solar panel is still reduced by the accumulation of dust known as soiling. This paper introduces a technique of cleaning a solar panel using control model of automatic windshield wiper based on Levenberg — Marquardt (LM) neural network. A network model of pattern recognition based on LM algorithm is trained and tested with specialist's information data. The result showed that this model based on LM neural network is effectual to grip doubts and nonlinearities of the automatic windshield wiper system, without using mathematical model. The accuracy of developed ANN model to predict the accumulated dust over the panel was found to be 99%.