Abstract: A prediction method for time series based on Hilbert-Huang transform and ARMA model is proposed. The Hilbert-Huang transform is used to decompose the original time series into a number of intrinsic mode function components and the instantaneous frequencies and amplitudes of each intrinsic mode function component are obtained. Then the ARMA model of each instantaneous frequency and amplitude sequence is established. Finally，ARMA prediction model of the original sequence is obtained through compounding. Experimental examples demonstrate that the method based on Hilbert-Huang transform and ARMA model can be applied to predict non-stationary time series effectively.

Key words: Hilbert-Huang transform, ARMA model, instantaneous frequency, instantaneous amplitude, intrinsic mode function