مطالعه اثر محل نصب سنسور ارتعاش‌سنج در تشخیص عیوب چرخ‌دنده با استفاده از الگوریتم ماشین بردار پشتیبان

Gearboxes, due to their demanding and intermittent operating conditions, are highly prone to failure and making timely fault detection critically important. Vibration analysis is one of the most widely used methods for health monitoring of gearboxes. Through vibration sensors, the gearbox's vibration signals are measured, and its health condition is assessed via signal processing and analysis. As a result, determining the optimal location and direction for sensor placement is of paramount importance. In this study, vibration data from six different position-direction combinations in a laboratory gearbox setup were collected. Various features indicator of the health status of the gearbox and gears were extracted. To identify the best sensor placement, an SVM classifier with linear and nonlinear kernels was constructed for the extracted features of each sensor. It was demonstrated that nonlinear kernels tend to overfit, and the best position-direction identified across all three models, with varying accuracy, was the gearbox output in the vertical direction.