Spur Gear Fault Detection Using Design of Experiments and Support Vector Machine (SVM) Algorithm

In this research, the primary objective is to ensure the appropriate functioning of transmission components, particularly the gearbox, which is highly prone to wear due to carrying the load directly. Condition monitoring and predictive maintenance of the gearbox are essential to prevent failures that can result in downtime and costly repairs. To simulate the wear in a controlled manner, tooth breakage and pitting were artificially induced using EDM. The raw vibration data obtained from an accelerometer sensor were then imported to LabVIEW software via a data acquisition system and analyzed in time and frequency domains at varying speeds and loads. The time-domain analysis included metrics such as “RMS and kurtosis,” while the frequency-domain analysis involved features such as "order spectrum." Additionally, time–frequency domains, such as "DWT and CWT," were utilized to provide a more comprehensive analysis of the gearbox's health. To classify the results obtained, support vector machining was used. The results obtained from the analysis provide a more in-depth understanding of the predominant types of wear in gearboxes and can be used to develop effective condition monitoring and predictive maintenance strategies to improve the reliability and lifespan of transmission systems.