Gear fault diagnostics using extended phase space topology

Abstract

This paper applies a novel feature extraction method calledExtended Phase Space Topology (EPST) in order to diagnosevarious faults in a gear-train system. The EPST method,that our research team has been developing, is based oncharacterizing the vibration data using the topology of phasespace, computing its density distribution and then expandedin a series of orthogonal functions. The resulting coefficientsare subsequently used in a machine learning algorithm. Forthis study, multiple test gears with different health conditions(such as a healthy gear and defective gears with root crackon one tooth, multiple cracks on five teeth and missing tooth)are studied. The vibration data of a gear-train is measured bya triaxial accelerometer installed on the test. Results indicatethat EPST is efficient in diagnosing the status of the healthof the gear system and characterizing the dynamic behavior.Moreover, the EPST procedure does not require a prioriknowledge about the dynamics of the system. EPST needsno noise reduction, signal prepossessing, feature ranking orselection, and therefore can easily be applied in a relativelyautomated process.