Multi-factor Hybrid Learning Method Based on Poisson Noise and Optimized Extreme Learning Machine and Its Application
Considering the characteristics of high fluctuation and intermittency of wind power data,this paper proposes a multi-factor hybrid learning approach based on complementary ensemble empirical mode decomposition with Poisson noise(CEEMDPN),modified snake optimizer(MSO)and extreme learning machine(ELM).Firstly,CEEMDPN is used to decompose the wind power sequence into subsequences.Then,the snake optimization algorithm is improved by introducing curve adaptive adjust-ment parameters.Finally,MSO-optimized ELM is utilized to predict and integrate each subsequence.In order to test the validity of the CEEMDPN-MSO-ELM model,the wind power data from Longyuan Power Group is used for ultra-short-term forecast.The empirical results are shown as below:The CEEMDPN algorithm can strengthen the main frequency part of wind power series and improve the decomposition accuracy,and MSO algorithm can well balance the optimization speed and convergence accuracy of the algorithm,so as to effectively improve the prediction performance of ELM model.The prediction accuracy and robustness of the proposed model are better than other models.
万方数据
维普
