All‐factor short‐term photovoltaic output power forecast
Na Zhang, Shouxiang Wang, Guangchen Liu, Jianwei Zhang
Abstract
Abstract Accurately predicting photovoltaic output power is one of the most important basic tasks for the rapid development of the smart grid. The factors that influence photovoltaic output power are not fully considered in the current forecast model. To this end, an all‐factor photovoltaic short‐term output power prediction model based on improved ensemble empirical mode decomposition (EEMD) and the seeker optimization algorithm (SOA) ‐BP neural network model is proposed here. First, different similarity algorithms are used in the selection of similar days of different weather types. Second, the factors that affect the PV output power are divided into deterministic parameters and stochastic parameters. The stochastic parameters of similar daily samples are used to constitute a time series, which is decomposed into several stationary signals by EEMD, and then, these signals are trained separately by the SOA‐BPNN model. Finally, the final forecast result can be synthesized with the deterministic parameters and the accumulation of the predicted data of individual signals. Compared with the results of different prediction models, the prediction accuracy of the proposed model is much higher.