Spur gear tooth root crack detection using time synchronous averaging under fluctuating speed

Filtering techniques are used to improve the signal to noise ratio (SNR) for better feature extraction. The time synchronous averaging (TSA) is one of such method that is based on averaging periodic sections. However, it fails to give significant results for an asynchronous or fluctuating speed condition. Moreover, most of the real life applications of gear are in asynchronous conditions. The aim of this paper is to develop a methodology which is robust for fault detection of gears under fluctuating load and speed conditions. A multiple-pulse individually rescaled-time synchronous averaging (MIR-TSA) technique in conjunction with conventional time synchronous averaging has been proposed. A 2-D finite element methodology based on principal or linear elastic fracture mechanics is adopted for predicting the crack propagation path at the root of gear tooth. The crack has been introduced using wire electrode discharge machining (WEDM). The vibration signals were recorded using drivetrain dynamic simulator (DDS) setup for various combination of load and crack length both for constant as well as fluctuating speed. Various time domain features such as root mean square, crest factor and kurtosis have been calculated using classical TSA and proposed MIR-TSA. A comparison of different extracted features between the proposed method and classic TSA has also been outlined. It has been observed that the proposed method enhances the fault detection under fluctuating speed conditions.