Complex time series analysis for rotor dynamics identification and fault diagnosis

Abstract

On account of the ever increasing demand for high performance and efficiency ,the design of rotating machinery is pressed toward light weight and high operating speed so that in-depth study and accurate knowledge on vibration characteristics of the machines have become increasingly important to meet with the requirements such as the durability, reliability, performance, and environmental acceptability. Among others, modal testing of such machines has provided a major contribution to understand and to control many vibration phenomena encountered in practice. In recent years, time series methods have been widely and successfully used for system and/or modal parameter identification of stationary structures in many fields as an alternative to the frequency-domain approaches to overcome their inherent drawbacks. Unlike other stationary structures, rotors show peculiar modal characteristics known as the backward and forward modes with directivity information. Conventional modal testing methods often ignore the directivity of modes when they are applied to rotating machinery. It often results in heavy overlapping of the backward and forward modes in the frequency domain, leading to ineffective modal parameter identification. In the past, ARMAX (AutoRegressive Moving Average with eXogenous inputs) model-based approaches utilized the conventional modal testing theory developed for stationary structures. Thus it is not adequate to apply the conventional approach to rotating machinery, because both the conventional modal testing theory and ARMAX model adopt the real notation (real-valued signal processing), yielding complex conjugate pairs of eigensolutions, that is, losing the important directivity information of modes. Extensive investigations on the complex modal testing theory and modal parameter identification for rotating machinery have recently been reported, whose key idea is to use the directional frequency response functions (dFRFs) between complex inputs and ...