A Data Driven Condition Monitoring of Chain Type Structures by Using Nonlinear Frequency Analyses

Abstract

A novel data driven condition monitoring approach is proposed in this study to detect the location of nonlinear faults in chain type structures. In this approach, the chain type system is characterized by representing the relationships between any two adjacent measurement points by NARX (Nonlinear Auto-Regressive with Exogenous Inputs) model. A new indicator for condition monitoring, known as the NET (Nonlinear Energy Transmission) indicator, is introduced based on the evaluation of the NRSFs (Nonlinear Response Spectrum Functions) of the chain type system. A 3 DOF system is employed to demonstrate the application of the data driven condition monitoring by using the NET indicator. A case study on the condition monitoring of a rotor-bearing system is then discussed to validate the advantage of the proposed approach in engineering practice.