Fault diagnosis methodology based on nonlinear system modelling and frequency analysis

摘要

In this paper, a new fault diagnosis methodology based on nonlinear system modelling and frequency analysis is developed. The method is derived via resolving several fundamental issues associated with conducting system fault diagnosis using the NARMAX (Nonlinear Auto-Regressive Moving Average with eXougenous inputs) modelling and NOFRFs (Nonlinear Output Frequency Response Functions) based frequency analysis, which was recently proposed and extends the well-known linear system modelling and FRF (Frequency Response Function) based frequency analysis to the nonlinear case. Simulation studies verify the effectiveness of the new method and demonstrate the performance of the method when applied to address a mechanical structural system fault diagnosis problem. The new methodology has the potential to resolve a wide range of input output data based engineering system fault diagnosis problems.