Wind speed prediction by a swarm intelligence based deep learning model via signal decomposition and parameter optimization using improved chimp optimization algorithm

Accurate prediction of wind speed plays a very important role in the stable operation of wind power plants. In this study, the goal is to establish a hybrid wind speed prediction model based on Time Varying Filtering based Empirical Mode Decomposition (TVFEMD), Fuzzy Entropy (FE), Partial Autocorrelation Function (PACF), improved Chimp Optimization Algorithm (IChOA) and Bi-directional Gated Recurrent Unit (BiGRU). Firstly, the original wind speed data was decomposed by TVFEMD to obtain modal components, and FE aggregation is used to decrease the computational complexity. Secondly, the components are processed by PACF to extract important input features. Thirdly, the BiGRU parameters are optimized using IChOA which is an improved version of ChOA. Finally, the optimized BiGRU is used to predict the decomposed components, and the predicted components are summed to obtain the final prediction result. In this experiment, the proposed model is used to predict the data of four months of a year from Station 46,060 of National Data Buoy Center, and the performance of eight benchmark models is analyzed. Experimental results show that TVFEMD and PACF can improve the prediction accuracy of the model. IChOA is feasible to optimize the parameters of BiGRU and can improve the prediction performance.