Transforming and separating rotating disk vibrations using a sensor array

This paper presents a method to separate the complex vibration pattern of a rotating disk into simpler entities. The decomposition transforms data measured by an array of sensors into time domain signals representing the contribution of individual modes of vibration. Having performed the decomposition with respect to wavelength, speed and direction of travel, the obtained measurements can be projected onto a rotating body experiencing variable rotational speed relative to the sensors. Unlike previous works, the vibrations here are decomposed into time domain signals that provide better insight into stress levels and fatigue than frequency domain based decompositions. Furthermore, the proposed method works under non-stationary conditions, e.g. under rapid angular acceleration and during transient motions. By exploiting the spatial deployment of sensors, the proposed transformation can produce information about the deformations in the body-fixed or material coordinates which is essential for stress analysis. The main feature of the method is the ability to separate modes of vibration that normally overlap in the frequency domain, to enable better insight into the sources of vibration. The method is demonstrated by analytical, numerical and experimental examples.