|Institution:||University of New South Wales|
|Department:||Mechanical & Manufacturing Engineering|
|Keywords:||Condition monitoring; Order-tracking; Angular resampling; Gear diagnostics; Bearing diagnostics; Wind turbines; Operational modal analysis; Cepstrum|
|Full text PDF:||http://handle.unsw.edu.au/1959.4/54328|
Order-tracking is a method to remove speed fluctuations from a varying frequency vibration signal allowing constant frequency based Machine Condition Monitoring (MCM) analysis techniques to be employed. This thesis describes the generalised Phase Demodulation based Order-Tracking (PDOT) methodology whereby the “mapping” between rotation angle and time is obtained by phase demodulation of a reference signal. A variety of reference signal types can be used, including a tachometer, shaft encoder, or an extracted reference signal from the response data signal itself. Each has different properties, which have to be taken into account. The primary advantage of PDOT, as opposed to methods based in the time domain, is that in principle it gives samples of the true relationship between rotation angle and time. However, there has to be no aliasing, in the sense of overlap of sidebands of the order to be demodulated with those of higher orders in the frequency domain. This thesis defines the conditions for which this is valid. The PDOT method can be employed using a single stage, or the result improved by using progressive iterations in a multi-stage approach, even with large speed variations. For iterations to give an improvement there must be no aliasing at the first stage. A summary of the basic PDOT method is presented, highlighting the maximum speed variations of approximately ±30% which can be compensated for. Then the generalised PDOT method is presented, which can be employed in a modular fashion, using a variety of reference signal types and numbers of stages. One new development is using a segmented approach for very large speed variations, and another is the ability to reverse an order-tracked signal back to the time domain. Multiple experimental examples are presented for different applications of PDOT, highlighting the suitability of PDOT for many variable-frequency applications. Examples include bearing diagnostic and gear diagnostic applications, in the presence of small to large speed variations, the latter using a segmented approach to order-track a run-up signal. Another uses PDOT in a reversible fashion, combined with cepstrum editing techniques, to pre-process a signal for subsequent Operational Modal Analysis (OMA).