Particle swarm optimized neuro-fuzzy system for photovoltaic power forecasting model

摘要

In this paper, a neuro-fuzzy system tuned by particle swarm optimization algorithm has been applied for representing the photovoltaic characteristics. The resulting model has optimum compactness and interpretability and can online estimate and predict the maximum power point of individual photovoltaic modules. Experimental data has confirmed its improved accuracy. The particle swarm tuned neuro-fuzzy model has been applied to a practical photovoltaic power generation system for maximum power point control. The simulated results showed an average error of 0.25% with respect to the maximum extractable power of the panel used under static conditions; this percentage remains universal for a range of dynamic weather conditions at sampling rate of 1 sample/12 min. The errors obtained, on average, are reduced to one fourth in comparison of the genetic algorithm based model proposed in a previous research.