Data-Based Testing for Nonlinearity in Dynamical Systems: The Use of Surrogate Data

Abstract

The use of Fourier-based surrogates for data-based testing for nonlinearity within system identification is introduced. It is shown that the presented method handles multi-input multi-output systems as easily as single-input single-output systems. In addition, the approach can be readily applied as a tool for validating nonlinear models by residual analysis. Furthermore, an extensive empirical analysis shows that the proposed methodology appears to be the most reliable alternative for detecting nonlinearity in general dynamical systems.