A New Audio Signal-Based Methodology for Fault Detection in Gears

Gears are one of the popular mechanical components, which are used for power and motion transmission and change of rotational speed. The harsh environment, where gears work, usually leads to the presence of defects. Consequently, the development of these defects will probably cause unplanned down time. Hence, monitoring the condition of gearbox and early detecting of defects are of great importance. In the present study, a new audio signals-based methodology is proposed for the purpose of fault diagnosis in gears. The methodology suggests analyzing the audio signals, recorded from gears during working, as time series and representing them using an accurate and suitable model. Then, the model coefficients are used as features for indicating the condition of the gears during the monitoring process. Two conditions (healthy and single broken teeth) are considered in the analysis. The results showed that the present method is able to detect the presence of faults effectively. The method is validated for a number of rotational speeds.