Development of a generic rotating machinery fault diagnosis approach insensitive to machine speed and support type

• The transferability of three FD methods is tested and a new technique is proposed.
• Experiments are done on a laboratory rig with different flexible bearing supports.
• Previously introduced methods are transferable to dynamically different machines.
• In comparison, proposed approach gives improved isolation of conditions tested.
• The integration of dynamic features improves the efficacy of the diagnosis.

Despite numerous difficulties that can be encountered when using trend monitoring of harmonic components from the simple amplitude spectra to aid diagnosis of rotor related faults on large multi-stage multi-bearing rotating machines, the technique continues to be the mainstay in industry. This is due in part to factors including a lack of adequate experimental validation of newly proposed techniques aimed at improving or replacing this traditional practice. Nevertheless, in recent studies, simple but robust Individual Speed Individual Foundation (ISIF) and Multi-Speed Individual Foundation (MSIF) fault diagnosis (FD) methods that both used a single vibration sensor per bearing without the use of phase information was applied with good results to fixed and variable speed machines respectively. A similar Individual Speed Multi-Foundation (ISMF) technique later enabled FD by direct comparison of vibration data between similarly configured machines with different dynamic characteristics operating at the same steady-state speed. However, the efficacy of these techniques was questioned as they were all applied to experimental rigs with the same few rotor related faults. Thus, the objective of this study is to test the transferability of these said techniques on a wider range of rotor related faults on different machines. A new Multi-Speed Multi-Foundation (MSMF) method which facilitates FD by the direct comparison of vibration data from similarly configured machines with different dynamic characteristics operating at different steady-state speeds is also proposed. It is observed that the previously proposed methods are indeed able to separate the range of conditions tested on machines with different dynamic characteristics. Analysis done with newly proposed MSMF approach gives improved isolation of fault conditions tested compared to the previously proposed techniques.