Gear Wear Fault Diagnosis in Tail Gearbox of Helicopter Using K-Nearest Neighbor Recognition Pattern

Application of rotary systems in aerospace, power stations, automotive industries and many others is prevalent. System failures and damaging always force a lot of costs to owners of these industries. In addition, maintenance of rotating system based on traditional logics is very expensive. Therefore, intelligent fault detection of engineering systems, especially mechanical ones, is important and a growing issue in industries all around the world. This paper presents a new intelligent approach to fault detection and diagnosis of gear wear. The proposed method is applied to a power transmission system in tail gearbox of a helicopter. Vibration signals of helicopter tail gearbox are measured at some convenient locations and further processed to determine the significance of these signals. A discrete wavelet analysis is used for feature extraction in addition to conventional statistical features extracted from vibration signals. The principal component analysis is used to obtain dominant features. Then, these features are fed as input for training and testing a K-nearest neighbor classifier. The results of the proposed intelligent diagnosis system are found to be encouraging. Review History: Received: 23 December 2012 Revised: 15 December 2015 Accepted: 2 January 2016 Available Online: 26 October 2016