Generic vibration-based faults identification approach for identical rotating machines installed on different foundations

Vibration-based faults identification (VFI) in rotating machines is well-known and adopted in industries. However, it has proved challenging to directly apply historical measured vibration data from one rotating machine with a particular foundation to another identical rotating machine installed on a different foundation layout during VFI. This is generally due to the different dynamic behaviours exhibited by the identical rotating machines as a result of their different foundation layouts. Hence, VFI could be significantly simplified through the application of a generic approach that enables the sharing of measured vibration data and experience between several identical rotating machines with similar or different foundations. In order to realise this objective, two identical rigs with different flexible foundations were used. Vibration data were collected on the rigs with different foundations under several operating conditions. The measured vibration data were then analysed in the frequency domain to compute poly coherent composite higher order spectra (pCCHOS). pCCHOS is a data fusion method that generates a single composite bispectrum or trispectrum irrespective of the number of vibration measurement locations. The components of pCCHOS were jointly analysed using pattern classification and the results showed clear separation among all considered machine conditions, which is an indication of its ability to satisfy the generic VFI objectives. Sensitivity analysis was also conducted to assess the robustness of the method in case there are faulty measurements and/or sensors from certain locations of the rotating machine.