A hybrid approach to bearing defect diagnosis in rotary machines

This paper presents a computational method, termed optimized envelope order spectrum, for bearing defect diagnosis in rotary machines that operate under non-stationary conditions. The method takes a holistic approach to bearing vibration signal analysis by integrating wavelet enveloping, computed order tracking, and spectral analysis. A scale selection criterion based on the maximum relative energy to Renyi entropy ratio is proposed to determine optimal decomposition scale for wavelet analysis. To eliminate the effect of speed variations and facilitate envelope-order spectral analysis, computed order tracking is introduced. The effectiveness of the method for identifying and localizing bearing defect is evaluated numerically and confirmed by experimental studies.