Ultra-short-term Photovoltaic Generation Forecasting Model Based on Weather Clustering and Markov Chain

Random fluctuations of solar radiation result in low precision of ultra-short-term photovoltaic(PV) power prediction. To solve this problem, the modified model based on Adaboost, KNN clustering and Markov Chain is proposed in this paper. First, an improved classifier combining k-Nearest Neighbor (KNN) with Adaboost is adopted to classify the collected historical data. Furthermore, the concept of attenuation coefficient of solar radiation is presented to modify the Hottel model. Then, a weighted Markov chain model is built to predict the solar radiation. Active power output is calculated by PV engineering model with respect to solar radiation. Experimental results indicate that the proposed model can significantly improve the precision of power prediction in cloudy and rainy conditions.