Adaptive Ultra-Short-Term Rolling Forecast Model of Photovoltaic Power

Photovoltaic (PV) power forecast can not only provide the reference for grid dispatching, but also improve the safety and stability of the distribution network. It is of great significance to use reasonable methods to predict photovoltaics. Based on this, this paper proposes an adaptive ultra-short-term rolling prediction method to improve the accuracy of the prediction model according to historical photovoltaic power. First, wavelet decomposition is used to extract photovoltaic energy, which divides the photovoltaic power curves into stable and fluctuating types. Then, prediction models are established according to the types of photovoltaic power curves respectively. Finally, a dynamic time warping (DTW) algorithm is used to measure similarity between the predicted photovoltaic power and the historical photovoltaic power during every rolling forecast, and an appropriate model is selected to predict the photovoltaic power at the next moment. This article will take the photovoltaic data of a 2.8MW photovoltaic power plant in a city in China as an example to verify the feasibility of the proposed model.