Short-term photovoltaic power prediction based on similar days with meteorological factors

Short-term photovoltaic power prediction based on similar days is an effective prediction method. Selecting accurate similar days directly affects the accuracy of the forecast results. In this paper, the k-means clustering algorithm is proposed to identify the types of weather clusters, and the main meteorological factors are extracted by using the correlation analysis of photovoltaic power and meteorological factors. Based on the Gray Relational Analysis, the meteorological factors of different types are obtained. And the weighted similarity calculation formula is given to obtain a similar day sample training set. Then, based on the similar days selected, Random Forest Regression prediction model was established for prediction. By comparing with the traditional similar days' selection algorithm, this model has higher accuracy and applicability.