An evolutionary deep learning model based on TVFEMD, improved sine cosine algorithm, CNN and BiLSTM for wind speed prediction

Abstract

Accurate prediction of wind speed is of great significance to the stable operation of wind power equipment. In this study, a hybrid deep learning model based on convolutional neural network (CNN), Bi-directional long short-term memory (BiLSTM), improved sine cosine algorithm (ISCA) and time-varying filter based empirical mode decomposition (TVFEMD) is proposed for wind speed prediction. Firstly, the original wind speed data is decomposed into intrinsic mode functions (IMFs) by TVFEMD to improve the data stability. Then, the importance of each decomposed subcomponent is analyzed using random forest (RF). Thirdly, CNN-BiLSTM is employed to predict the wind speed. And, an improved sine and cosine algorithm (ISCA) is utilized to optimize the model parameters BiLSTM. Finally, the forecasting results of each sub-model are combined to get the final prediction results. In this study, the proposed model is utilized to four monthly wind speed data sets, and different comparison models are established. The experimental results of this study show that TVFEMD and RF can process data more effectively and improve the prediction accuracy. ISCA can optimize the parameters of BiLSTM model and improve the prediction performance. The proposed model in this study can obtain good prediction results on all data sets.