Paper Title
Intelligent Gear Faults Diagnosis Based on Angular Measurements

The appearance of a defect in the rotating machinery generally causes a modification of vibratory signature. The use of vibration monitoring techniques by pattern recognition enables to automate the diagnosis. The performances of these methods are closely related to the relevance of fault indicators making up the Feature Vector (FV). This FV must be able to describe the different modes of operation or damage of the system, and also reflect the precise definition of the classes that represent the different modes of operation. These indicators must also be constructed automatically to make the most robust analysis possible. Contrary to time Sampled Acceleration signals (TA), angular measurements like Instantaneous Angular Speed (IAS), Transmission Error (TE) and Angular Sampled Acceleration (ASA) represents all potential sources of relevant information in fault detection and diagnosis systems, but also to construct FV to make the methods of classification robust and effective even for different running speed or load conditions. In this paper, several different signatures of nature of sampling (angular and temporal) are determined to monitor several different operating modes. For this purpose, features are extracted from time and frequency domains. Then, the Sequential Backward Selection algorithm (SBS) is applied in order to select the most relevant features. The classification is performed by Support Vector Machines (SVM) in order to improve the detection and identification of gear defects. The methodology is applied in normal conditions, then with five pinion faults for different running speed and load conditions. The experimental results prove the efficiency of angular indicators by increasing performance of the classification. Keywords - Fault diagnosis, Gearbox, Angular Measurements, Support Vector Machines