Hilbert-Huang Transform Based Pseudo-Periodic Feature Extraction of Nonlinear Time Series

It is significant that analyze the periodic or pseudo-periodic disciplines of complex systems from the random component. Focused on the problems of difficult extraction and low accuracy of pseudo-periodic features of complex system, and taken the nonlinear time series generated by the complex system as the main research objects, a method of pseudo-periodic feature extraction for nonlinear time series is proposed based on the Hilbert-Huang transform. The empirical mode decomposition is used to decompose a signal into various intrinsic mode functions (IMFs) with the properties of complete and nearly orthogonal basis, the Hilbert spectrum analysis is applied to obtain the frequency-time distribution of IMFs, and the pseudo-periodic feature of the original time series is calculated finally. Three cases of classical nonlinear datasets are studied to describe the analysis and applying processes of the proposed method in detail. Through the contrastive analysis with the traditional methods of pseudo-periodic of extraction, the method presented in this paper can be used to extract the pseudo-periodic feature of nonlinear time series effectively and the extracted results are more believable than those obtained by traditional methods.