Diagnostics of gear faults using ensemble empirical mode decomposition, hybrid binary bat algorithm and machine learning algorithms

Early fault detection is a challenge in gear fault diagnosis. In particular, efficient feature extraction and feature selection is a key issue to automatic condition monitoring and fault diagnosis processes. In order to focus on those issues, this paper presents a study that uses ensemble empirical mode decomposition (EEMD) to extract features and hybrid binary bat algorithm (HBBA) hybridized with machine learning algorithm to reduce the dimensionality as well to select the predominant features which contains the necessary discriminative information. Efficiency of the approaches are evaluated using standard classification metrics such as Nearest neighbours, C4.5, DTNB, K star and JRip. The gear fault experiments were conducted, acquired the vibration signals for different gear states such as normal, frosting, pitting and crack, under constant motor speed and constant load. The proposed method is applied to identify the different gear faults at early stage and the results demonstrate its effectiveness