Application of minimum entropy deconvolution on enhancement of gear tooth fault detection

Abstract

Gearboxes are important parts of mechanical transmission systems, so that it is very pivotal for avoiding future catastrophic failures to detect faults in gearboxes at the early stage. Various vibration analysis techniques have been proposed and successfully applied to detect faults in gearboxes. The outputs of these techniques are used to identify the signal changes caused by defects. Because of modulations, the low-energy sidebands associated with partial gear teeth failures have also been proved to be notable. However, modulation sidebands are usually very weak and they would be easily submerged in inherent signal produced by mesh vibration. This paper presents an enhanced program based on frequency band analysis and minimum entropy deconvolution (MED) to identify the difference between healthy gear and gear with faulty tooth. With this program, the measured signal is first filtered with the MED filter. Then, spectrum analysis technique is applied to the enhanced-signal to find the modulation sidebands around mesh frequency. The enhanced program is validated using the test data from a two-stage wind turbine gearbox. The results show that the MED enhanced program is an effective tool to detect the pitting fault of gear teeth.