Experimental monitoring of eccentric gears with different mechanical conditions

Abstract

The use of Condition Monitoring (CM) for engineering design, and industrial gears, has dramatically changed the gear design and manufacturing process in a short space of time. As CM deployment grows rapidly, the feasibility of CM for the diagnosis and prognosis of most common gear failure modes has well been investigated and documented. However, this is not the case for the CM of industrial gear eccentricity. Previous published work, in particular, focused only on the use of simulation models as a basis to investigate the gear eccentricity. Simulations cannot always ease the detection of complex problems and complications that are experienced in real operations. For instance, excessive eccentricity can be considered as one of manufacturing errors that may severely lead to direct effect on the overall dynamic performance of gears. Yet, it produces very high modulated mesh frequency rates; thus, making the detection of faulty machine components very difficult or even impossible. With this in mind, this paper presents the first known attempt at the diagnosis and prognosis of experimental vibration datasets from five different eccentric gear conditions. Datasets were first analysed using Signal Intensity Estimator (SIE) method in time and frequency domains. Then, the data was subjected to additional processing for the classification of gearbox status. Observations from the results showed that the proposed techniques could successfully discriminate the “good” and “bad” gears.