Circular domain features based condition monitoring for low speed slewing bearing

Abstract

This paper presents a novel application of circular domain features calculation based condition monitoring method for low rotational speed slewing bearing. The method employs data reduction process using piecewise aggregate approximation (PAA) to detect frequency alteration in the bearing signal when the fault occurs. From the processed data, circular domain features such as circular mean, circular variance, circular skewness and circular kurtosis are calculated and monitored. It is shown that the slight changes of bearing condition during operation can be identified more clearly in circular domain analysis compared to time domain analysis and other advanced signal processing methods such as wavelet decomposition and empirical mode decomposition (EMD) allowing the engineer to better schedule the maintenance work. Four circular domain features were shown to consistently and clearly identify the onset (initiation) of fault from the peak feature value which is not clearly observable in time domain features. The application of the method is demonstrated with simulated data, laboratory slewing bearing data and industrial bearing data from Coal Bridge Reclaimer used in a local steel mill.

Integrated Condition Monitoring and Prognosis Method for Incipient Defect Detection and Remaining Life Prediction of Low Speed Slew Bearings

“…How these features were extracted has been described in Refs. [32,33]. It is apparent that the features do not display any clear trend.…”

“…Step (1) is the detection of the incipient slew bearing defect where combined MSET and SPRT is used to process circular-domain kurtosis [32], time-domain kurtosis [32], WD kurtosis [32], EMD kurtosis [32] and LLE feature [33].…”

“…Figure 1 illustrates the computational implementation of the method. kurtosis [32] and LLE feature [33].…”

“…In this paper, circular kurtosis, time-domain kurtosis, WD kurtosis and EMD kurtosis are employed to obtain the bearing degradation parameter for prognosis method. [32]; (c) wavelet kurtosis [32]; (d) EMD kurtosis [32]; (e) the LLE feature [33]. Note: t = dayth.…”

“…Consequently, the method only uses five of the most sensitive features as the input to the MSET. The five features were (1) kurtosis extracted from accelerometer signal, f1(t); (2) kurtosis extracted from the piecewise aggregate approximation (PAA) combined with circular analysis, f2(t); (3) kurtosis extracted from the wavelet decomposition, f3(t); (4) kurtosis extracted from the EMD, f4(t); and (5) LLE feature obtained from the largest Lyapunov exponent algorithm, f5(t) [32]. The five features were extracted over the 139 days test period.…”