Support Vector Machine based Gear Defect Classification using Wavelet Denoised Vibration Signal

In gearboxes and in all rotating elements in general, gear damage detection is often very critical and can lead to increased safety in aviation and in industry. Therefore the importance for their regular inspection and/or on-line health monitoring is growing and effective diagnostic techniques and methodologies are the objective of research efforts over the last 30 to 40 years. Gearbox is widely used in all industrial applications and hence there are many analytic techniques, which are used to prevent serious damage in a mechanical system. In general, the mechanical components produce abnormal transient signal when a fault condition occurs. These transient signals can be utilized to identify the fault in various conditions. These fault diagnosis techniques are mainly based on sound emission and vibration signals in time and frequency domains, some more advanced studies are continuous wavelet transform and discrete wavelet transform (DWT) can effectively detect weak impulse signals for fault conditions and hence wavelets can be used for the fault diagnosis of gearbox.