Vibration Region Analysis for Condition Monitoring of Gearboxes Using Image Processing and Neural Networks

Abstract

This paper focuses on extracting vibration region of gearboxes, monitoring parameters of the condition and detecting faults. When the gear contact surface tension exceeds the fatigue limit of the gear material because of excessive loading, inadequate lubrication, oil contamination and so on, tooth surface failures such as pitting, wear, tooth breakage, etc. occur. These types of defects in gear systems cause deterioration in the performance of gears, and thus significant operational problems may occur in the industry. In addition to existing studies related to fault detection, this study proposes extracting of vibration region to map the accelerations of the gearboxes. The proposed method consists of the following steps: the two-dimensional vibration region of the gearbox is created by using instantaneous accelerations taken from gearboxes in horizontal and vertical directions. Features of vibration region give valuable detail about the condition of gearboxes. Therefore, it is transformed into a binary image, and the features are extracted by using image processing algorithms. Finally, these features are used as inputs to the artificial neural network for classification of fault severity. To validate the proposed method, a two-stage helical gearbox and a worm gearbox are utilized as an experimental setup. As a result of the experiments carried out, it has been seen that the proposed method can accurately monitor the condition of the gearboxes and classify the severity of faults.