Short-term Wind Speed Prediction Based on Improved CEEMD-FOA-LSSVM
Minjie Li, Guige Gao · 2020 IEEE International Conference on Power, Intelligent Computing and Systems (ICPICS) · 2020
In order to solve the problem caused by the nonstationarity of wind speed sequence in wind speed prediction, an improved CEEMD-FOA-LSSVM model is proposed: Firstly, the standard Complementary ensemble empirical mode decomposition (CEEMD) algorithm is improved to decompose the original wind speed sequence and reduce the end effect. Secondly, calculate the permutation entropy value of each modal component after decomposition, and the sub-sequences are recombined and merged according to the permutation entropy value to obtain multiple new sub-sequences, and the improved Fruit Fly Optimization Algorithm (FOA) algorithm can improve the optimization effect and optimize the Least Squares Support Vector Machine(LSSVM) parameters. Finally, the optimized model is used for predicti on for each new sequence, and the predicted values are combined to complete the model prediction. The simulation results prove that the proposed improved model not only effectively solves the above problems but also improves the prediction accuracy.