Classification of the Severity Level of Breakage Failure in Spur Gearboxes Through Frequency Domain Vibration Signal Analysis

In the industry, gearboxes used for their efficiency in power transmission are critical components, making early failure detection essential. This work aims to determine the ranking of condition indicators (CIs) to extract information from the vibration signal in the frequency domain for a spur gearbox and to assess the accuracy of the classification model for the failure severity level. In laboratory conditions, the tooth breakage of a pinion with different severity levels was simulated in a spur gearbox. Four accelerometers were installed in the gearbox in a vertical position to obtain the vibration signal. First, information was extracted from the vibration signal through 25 CIs in the accelerometers. Using artificial intelligence, the ranking of 10 CIs was carried out. Subsequently, the random forest (RF) algorithm was used to determine the accuracy in classifying the failure severity level. Finally, an ANOVA test was conducted to determine if there were significant differences in classification accuracy for the four accelerometers. The results concluded that the CIs selected through the ranking are optimal for determining the failure severity level of tooth breakage in a spur gearbox and that the sensor placement affects the classification accuracy.