A future possibility of vibration based condition monitoring of rotating machines

Adopted conventional practice uses a number of vibration sensors at a bearing pedestal of a rotating machine for the vibration based condition monitoring. The number of bearings in a machine, say a Turbo-Generator (TG) set, is likely to be very high hence increasing sensors to a large number. Therefore results in huge data sets to analyse to track any fault(s) which often depends on the experience and the engineering judgments in fault detection process. The effort of the present study is to reduce the number of sensors per bearing pedestals by enhancing the computational effort in signal processing. The concept used was fusion of the data from all sensors in the frequency domain to get a composite spectrum for a machine and then the computation of the higher order spectra (HOS) so that the vibration data is managed efficiently and able to detect fault uniquely. The results of the suggested approach are discussed here. © 2012 Elsevier Ltd.