Open AccessResearch on short‐term output power forecast model of wind farm based on neural network combination algorithm
Author(s) -
Qu Boyang,
Xing Zuoxia,
Liu Yang,
Chen Lei
Publication year - 2022
Publication title -
wind energy
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.743
H-Index - 92
eISSN - 1099-1824
pISSN - 1095-4244
DOI - 10.1002/we.2763
Subject(s) - hilbert–huang transform , artificial neural network , algorithm , benchmark (surveying) , wind power , wind power forecasting , electric power system , noise (video) , computer science , signal (programming language) , power (physics) , term (time) , mode (computer interface) , control theory (sociology) , artificial intelligence , engineering , white noise , physics , quantum mechanics , electrical engineering , telecommunications , control (management) , geodesy , image (mathematics) , programming language , geography , operating system
Abstract Reliable wind farm short‐term output power prediction plays a crucial role in the safety and economy of power system operation. By embedding multiple neural network methods, such as long short‐term memory (LSTM) neural network, multiple optimized BP neural network and wavelet neural network (WNN), combined with multiple methods based on signal decomposition and reconstruction to evaluate the short‐term prediction effect. In this study, the variational modal decomposition (VMD) method and the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) are used to reduce the fluctuation of the original wind power signal. Besides, the intrinsic mode function (IMF) signals obtained by VMD and CEEMDAN methods are used. By predicting the IMF signals, the prediction error of each signal can be reduced, and the accuracy of wind power prediction can be indirectly improved. In order to test the validity of the proposed composite model, the validation prediction and rolling prediction of 72 h with a resolution of 10 min were carried out, respectively, using the data of Caijiagou wind farm in 2018. Through experiments, the relationship between the two predictions is analyzed, which provides an important reference for the selection of the prediction model. Corresponding results show that the proposed VMD‐LSTM and CEEMDAN‐LSTM methods significantly outperform other benchmark methods and provide very satisfactory results in both the accuracy and stability of wind farm power prediction.