Embedding dimension determination in phase space reconstruction is difficult. The purpose of this paper is to present a new approach for embedding dimension determination based on empirical mode, showing that embedding dimensions for phase space reconstruction could be easily determined according to the number of intrinsic mode functions decomposed by empirical mode decomposition.
Through the relation analysis of intrinsic mode functions and embedding dimensions, the approach for embedding dimension determination by the number of intrinsic mode functions is presented. First, a time series is decomposed into several intrinsic mode functions. Second, correlation analysis between intrinsic mode functions and original signals is investigated, and then false intrinsic mode functions could be eliminated by the analysis of correlation coefficient thresholds, which makes the embedding dimension precise. Finally, the method presented is applied to the Lorenz system, Chen's system, and the Duffing equation. Simulation results prove this method is feasible.
A new approach for embedding dimension determination based on empirical mode decomposition is presented. Compared with G‐P algorithms, this new method is effective and decreases computational complexity.
This method provides an effective qualitative criterion to the selection of embedding dimensions in phase space reconstruction.
This method could be used to determine embedding dimensions of phase space reconstruction and degree‐of‐freedom of nonlinear dynamical systems.
The paper proposes a new method of embedding dimension determination in phase space reconstruction.
