Analysis of Nonstationary Signals Generated by Nonlinear Dynamic Systems

摘要

Signals generated by nonlinear dynamic systems are nonstationary. To analyze such signals using linear methods, for example spectral or time-frequency methods, one has to use long signals, free of artifacts, denoised etc. In this paper I concisely present nonlinear methods of signal analysis we use in our Lab - Higuchi fractal dimension, Empirical Mode Decomposition, Symbolic Analysis. These methods are much better suited for analysis of nonstationary signals to extract important features of the nonlinear dynamic systems that generated those signals. In particular, living organisms in general, and human organism is particular, are inherently nonlinear dynamic systems. That is why biosignals like EEG or HRV are inherently nonstationary, so nonlinear methods find application in medical diagnosis and in assessment of therapy results.