基于Hurst指数的风速时间序列研究

Abstract

In order to investigate self-similarity and long-rang dependence of wind speed time series, the rescaled range analysis and the detrended fluctuation analysis (DFA) were carried out to calculate the Hurst exponent of the wind speed time series. The power spectral density was analyzed and the spectral indexes were calculated. The results show that, the Hurst exponents calculated with the both methods are close to 1, which means that the wind speed time series has strong self-similarity and long-range positive dependence. However, the Hurst indexes from the 2 methods are different, which means that the DFA method could reflect the power-law feature of the non-stationary wind speed time series. Furthermore, the results of the Hurst exponents and the spectral indexes show that the wind speed fluctuation belongs to the “1/f fluctuation”. The conclusions offer some references for further study on fractal chaos and short-time wind speed prediction.