Scaling Analysis of Increment White-Noise Series

Noise is always resident in vibration data collected from rotating machinery and different noise levels may relate to different mechanical conditions. As a result, the noise level can seemingly be exploited to evaluate running conditions of rotating machinery. To this end, this paper examined scaling properties of a white-noise series and its increment series. Consequently, it was found that the increment white-noise series exhibits scaling properties sensitive to the change of strength of white noise, compared with the original white-noise series. Therefore, the scaling properties of the increment vibration series, whose noise levels differ between different running conditions, allow a separation between different mechanical conditions. Subsequently, the conclusion was confirmed numerically. Moreover, an interesting phenomenon was observed when extracting the fluctuation parameters of the starting and ending points of the scaling-law curves of the increment white-noise series as feature parameters. In addition, the interesting phenomenon, which may be investigated for diagnosis of mechanical faults, was explained concisely. Accordingly, a novel method for condition monitoring of rotating machines was presented. Finally, a gearbox experiment was conducted for confirming the feasibility of the proposed scheme.