Solar photovoltaic power prediction based on similar day approach

The output power of photovoltaic (PV) power generation system is related to solar irradiance, temperature, humidity and other meteorological conditions. The output powers in the similar days, which are much alike in meteorological conditions and social activities, will be more likely close. In this paper, a prediction algorithm combining the time series and similar day is proposed to predict the power output of a PV system. In this algorithm, the meteorological conditions of target days are predicted by the time series analysis, and the subjective weight and the entropy weight method are used to select the similar days. After that, the target day's power output will be forecasted by tuning the weights of similar days' power output. Based on the actual data collected from an experimental system, the accuracy of prediction method is verified.