Identification of Gear Faults in an Industrial Fan Prototype Using Spectrum Analysis and Continuous Wavelet Transform

Abstract

Gear is one of the power transmission systems in industrial fans used to reduce the shaft rotation. The spectrum analysis can be used to detect gear fault advancement. However, the spectrum only applies to stationary signals. In industrial fans, changes in fan workload cause signals to become non-stationary. In addition, heavy noise often occurs in the gear vibration which immerses the gear vibration. The continuous wavelet transforms (CWT) is effective for this condition. The purpose of this research is to apply the CWT for the gear fault advancement in industrial fans. The time synchronous averaging (TSA) is used as a pre-processing signal to reduce the unwanted vibration signal including noise. In this research, three different gear conditions are used i.e. normal gear (no fault), 50% (level 1) and 100% (level 2) gear fault advancement. The results of CWT are plotted into scalogram which gives both time and frequency information. The scalogram produced from vibration signals before and after TSA is compared and analyzed. The scalogram shows that the CWT after TSA gives prominent result to identify gear fault advancement. An increase in gear mesh frequency (GMF) amplitude is clearly observed and also the periodicity of frequency content is evident. The GMF amplitude of 50% and 100 % gear fault increases 2.5 times and 4 times to that of normal gear respectively. The comparison between the CWT and spectrum is also carried out where the CWT is proven superior.