Gear fault diagnosis using energy-based features of acoustic emission signals

In this work, energy-based features are introduced for monitoring and diagnosis of machine conditions in spite of speed and load variations. The basic feature, termed here the energy index (EI), is a statistical measure of relative energy levels of segments of a time domain signal over a cycle. The properties of the EI are discussed and its different forms are derived. A procedure is presented for fault diagnosis of gears using the proposed features. As an illustration, time domain acoustic emission (AE) signals of a test gearbox have been processed to extract these features and to test the relative significance in the diagnostic process. The proposed technique is compared with some of the existing methods using the same AE data for early fault detection. The applicability of the proposed technique is also studied using a set of vibration data of a helicopter drivetrain system gearbox. The results show the effectiveness of the proposed features in monitoring and diagnosis of machine conditions, with the capability of early fault detection.