Non-Stationary Wind Pressure Prediction Based on A Hybrid Decomposition Algorithm of Wavelet Packet Decomposition and Variational Mode Decomposition

2018 International Conference on Civil and Hydraulic Engineering

土木建筑工程;水利工程

Jin, Mengya^1 ; Li, Chunxiang^1

Department of Civil Engineering, Shanghai University, Shanghai

200444, China^1

关键词: Decomposition algorithm;    Decomposition methods;    Extreme learning machine;    Forecasting accuracy;    Mode decomposition;    Variational mode decomposition (VMD);    Wavelet Packet Decomposition;    Wavelet packet decompositions;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/189/5/052038/pdf
DOI  :  10.1088/1755-1315/189/5/052038

学科分类：土木及结构工程学

来源: IOP

【 摘 要 】

A hybrid decomposition algorithm is proposed for wind pressure prediction on structures based on wavelet packet decomposition (WPD), unit root test, and variational mode decomposition (VMD). In the proposed hybrid decomposition, the wavelet packet decomposition is employed to drop intermittent of the wind pressure. Then, the unit root test is adopted to ensure the stationary of the decomposed subseries. For the non-stationary subseries, the variational mode decomposition is employed to further decrease intermittent of the subseries. For the rest of the stationary subseries, they are reconstructed by frequency, which can be divided into high frequency, middle frequency and low frequency. Finally, the prediction process is established by the extreme learning machine (ELM). The three cases forecasting results indicated that: the proposed decomposition algorithm has the best forecasting accuracy compared to the relative decomposition methods.