A New Method of Modeling Gear Faults

Abstract

Detailed mathematical models of gear vibrations have appeared in recent years, but to utilize them requires a detailed knowledge of the gearbox components and their dynamic properties. This paper presents an alternative approach which is applicable to the monitoring and diagnosis of gearbox faults based on an analysis of changes in the vibration signal, showing how these can be related back to various classes of fault. Most emphasis is placed on the effects of the various types of faults on the spectrum, but the applicability of two other techniques, viz., synchronous signal averaging and cepstrum analysis, is also discussed comparatively, making use of practical examples.