Spur Gear Failure Detection using Support Vector Machine

Torsional vibration generated by the spur gear as result the imperfection may generate geometrical error that can lead to spur failure. In order to prevent error propagation, preventive maintenance by detecting and analyzing vibration should be performed. Existing work has been dealt with torsional vibration detection either using tools and/or employing intelligent system. This paper proposed evaluation prototype with support vector machine to differentiate normal and broken spur gears. The assessment results show that gear states were easily determined by using some features. The accuracy achieves 100% in several trials with average accuracy 99.35%.