Survey of Nonlinear Detection and Identification Techniques for Experimental Vibrations

Nonlinear structural dynamic system identification is often more a subjective art than it is a direct application of some particular method in systems theory. The nonlinear problem is subjective because although there are many analytical methods from which to choose, there is no general approach to detect, characterise, or model input-output relationships in nonlinear systems. This paper is a survey of the numerous experimental nonlinear structural dynamic system identification techniques that have been implemented by vibration engineers in recent years for SDOF and MDOF systems in both the time and frequency domain. The survey discusses the following detection and identification methods from an experimental perspective: Spectral analysis and the reverse-path formulation, Volterra and Wiener series, nonlinear auto-regressive moving average models, the restoring force method for SDOF and MDOF systems, the describing function methods, direct parameter estimation, Hilbert transforms and wavelet transforms, neural networks and general nonlinear experimental testing techniques.