A Novel Hybrid Model Based on Secondary Decomposition and Artificial Intelligence Approach for Abnormal Data Reconstruction

Abstract

The abnormal anemometer of wind turbines may be caused by environmental and weather effects, which can adversely affect the correctness of other system parameters and the efficiency of the wind farm. To reconstruct the abnormal data accurately and efficiently, this study proposes a newly hybrid model for reconstruction based on variational mode decomposition (VMD), improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN), improved grey wolf optimization (IGWO), and Long short term memory network (LSTM). In this model, the VMD is utilized to decompose the initial wind speed dates, the residual component is subjected to secondary decomposition using the ICEEMDAN, and the IGWO-LSTM model is built to reconstruct the wind speed data. To verify the validity of the developed approach 10-minute actual wind speed data from three stations in Hunan, China, are used. The experimental results of the developed technology are $\mathrm{RMSE}_{\text {1-step}}{=}0.1827$ , $\mathrm{RMSE}_{\text {2-step}}{=}0.2682$ , and $\mathrm{RMSE}_{\text {3-step}}{=}0.3649$ at Site 1; $\mathrm{RMSE}_{\text {1-step}}{=}0.2084$ , $\mathrm{RMSE}_{\text {2-step}}{=}0.3049$ , and $\mathrm{RMSE}_{\text {3-step}}{=}0.3785$ at Site 2; $\mathrm{RMSE}_{\text {1-step}}{=}0.1994$ , $\mathrm{RMSE}_{\text {2-step}}{=}0.2415$ , and $\mathrm{RMSE}_{\text {3-step}}{=}0.3625$ at Site 3. As a result, the reconstruction performance of this model is available to enhances the efficiency of wind farms.