A Comparative Study of Short Time Fourier Transform and Continuous Wavelet Transform for Bearing Condtion Monitoring

摘要

Rolling-element bearings can be found on almost all rotating machines and their failure is one of the major causes of machine breakdown. This paper provides a review on joint time-frequency domain analysis. Since in the real machine monitoring environment, the monitored signal, such as from vibration measurement, can be transient events with abrupt changes in the waveform, traditional analyses conducted solely in either the time or frequency domain are not always capbable of revealing the occurrence of bearing faults. An approach is to utilise joint time-frequency domain methods such as the Continuous Wavelet Transform (CWT) or the Short-Time Fourier Transform (STFT). The transformed signals were represented as colour-coded images which might contain unique characteristic features relating to the various types of bearing faults. Simulated signals were used in this study to compare the performance between STFT and CWT for signal classification. In this case, classification was basically pattern recognition of the image. The similarity between images was quantified using the correlation matching method. Results showed that CWT was more effective than STFT and its superiority was further affirmed by tests conducted on real bearing vibration signals.