Condition Monitoring of Rolling Element Bearing Based on Moving Average Cross-Correlation of Power Spectral Density

The purpose of bearing condition monitoring (CM) is to monitor the health of bearings and to provide the information, by means of suitable indices, for prognostics, which eventually estimates the remaining useful life (RUL) of the bearing. The reliability of bearing CM and RUL estimation depends on the consistency and on the performances, in wide sense, of the health index used. The health indices based on machine learning require a large amount of historical data, which are not available in many cases. Other health indices, which are characterized by monotonical growth, may be affected by the difficulty to determine the correct failure threshold in different operating conditions. A new health index, called ‘MACPSD’, is proposed in this paper, which is the moving average cross-correlation of power spectral density (PSD) of signals. The rationale of MACPSD is that different health condition would have different energy distribution in the frequency domain. MACPSD has the further advantage that it ranges from zero to one. MACPSD can track the changes of the energy distribution as the bearing damage develops. The performance of MACPSD are shown by using a set of data of bearing run-to-failure, which was generated by the NSF I/UCR Center for Intelligent Maintenance System.