Time-frequency analysis of non-stationary fusion plasma signals using an improved Hilbert-Huang transform.

Abstract

An improved Hilbert-Huang transform method is developed to the time-frequency analysis of non-stationary signals in tokamak plasmas. Maximal overlap discrete wavelet packet transform rather than wavelet packet transform is proposed as a preprocessor to decompose a signal into various narrow-band components. Then, a correlation coefficient based selection method is utilized to eliminate the irrelevant intrinsic mode functions obtained from empirical mode decomposition of those narrow-band components. Subsequently, a time varying vector autoregressive moving average model instead of Hilbert spectral analysis is performed to compute the Hilbert spectrum, i.e., a three-dimensional time-frequency distribution of the signal. The feasibility and effectiveness of the improved Hilbert-Huang transform method is demonstrated by analyzing a non-stationary simulated signal and actual experimental signals in fusion plasmas.

DOI: 10.1063/1.4887415

6 Figures and Tables

Cite this paper

@article{Liu2014TimefrequencyAO, title={Time-frequency analysis of non-stationary fusion plasma signals using an improved Hilbert-Huang transform.}, author={Yangqing Liu and Yi Tan and Huiqiao Xie and Wenhao Wang and Zhe Gao}, journal={The Review of scientific instruments}, year={2014}, volume={85 7}, pages={073502} }