A new characteristic analysis method for aero-engine rotor–stator rubbing

Abstract

Aero-engine fault diagnosis is often studied according to casing signal owing to the inconvenience of assembly and structural limitations. Satisfactory result by cyclic statistics alone is difficult to obtain because casing signal is weak and complex. Based on this, a cyclostationary theory combined with autocorrelation function and Hilbert transform is applied to extract the characteristics of aero-engine rotor–stator rubbing fault according to the aero-engine structure and rotating features with regard to of single-point rubbing and local rubbing. Signal autocorrelation function is analyzed by cyclic autocorrelation function and Hilbert transform, in order to extract rubbing characteristics based on casing signal of aero-engine and reduce inconsistent frequency in the frequency spectrum between time-delay signal of cyclic frequency slice position and original signal. Meanwhile, consideration is given to the influence of rubbing position and sensor installation position in the extraction of rubbing characteristics. Finally, the Hilbert envelope spectrum is compared and analyzed between normal running and occurrence of the rubbing faults of aero-engines. The result indicates that the different cyclic frequency positions slice signal's Hilbert envelop spectrum of signal autocorrelation function, but not original acceleration signal which has outstanding characteristic frequency for rubbing frequency (product between blade numbers and rotating frequency) and its twice frequency when rubbing fault occurs not only in single rubbing but also in local partial rubbing. Meanwhile, when compared with the original vibration signal slice, inconsistent frequency of the autocorrelation function slice in the Hilbert envelope spectrum is decreased considerably.