Rolling Bearing Fault Diagnosis using Support Vector Machine Algorithm with K-Fold Cross Validation and Gaussian Kernel

Rolling Ball bearings are vital components in any machine ranging from motors, pumps to heavy equipment like wind turbines. Specifically in Induction motors, bearings play a crucial role in supporting shaft movement, reducing friction, and ensuring smooth rotation, even under heavy loads. Consequently, any fault in the bearings can lead to interruptions or breakdowns in the entire machine process. Thus, predictive health monitoring of the rolling ball-bearing element can help prevent this. The primary objective of predictive health monitoring is to develop proactive measures that identify machinery issues and rectify them before they cause any disruptions. Different parameters, including current, speed, and acoustic emission, have all been considered to detect rotating machinery faults in advance. However, vibrational analysis has emerged as a significant parameter in diagnosing bearing faults in induction motors. The present approach to maintenance includes ball-bearing lubrication and replacement at consistent intervals. However, this approach may not be optimal because if that particular machine is underused, the bearing might be over-lubricated, and when overused, that can lead to the bearing being under-lubricated. Predictive monitoring offers a more efficient solution, automating the diagnosis of rotating equipment and enabling lubrication and replacement only when necessary.