Power smoothing control of photovoltaic generation systems using a difference vector method

This paper proposes a novel method for the short-term prediction of photovoltaic generation (PV) power using difference vectors. The predicted data are used for the smoothing control of PV power fluctuation with the moving average method in order to reduce the capacity of the energy storage device. Three smoothing control methods are discussed in this paper: a persistence method, a local fuzzy reconstruction method, and the proposed method for prediction, respectively. The results of the smoothing control methods are evaluated in terms of the prediction error, reduction rate of fluctuation, power in load frequency control (LFC) band, and capacity of the energy storage device. The proposed method has the lowest prediction error among the three methods. Furthermore, the smoothing control scheme with the function of managing the stored energy is advantageous in that it can reduce the capacity of the energy storage device by approximately 44% compared to a smoothing control scheme without the abovementioned function.