Intelligent fault diagnosis of rotating machinery using composite multivariate-based multi-scale symbolic dynamic entropy with multi-source monitoring data

Fault diagnosis of rotating machinery plays a significant role in the reliability and safety of modern industrial systems, which generally requires collaborative fault diagnosis by features extracted from multiple sensors since the multi-channel vibration signals carry a wealth of fault information. However, there is a remaining obstacle for fault diagnosis of multi-source monitoring data: integration of multisensory data. Hence, a novel framework is proposed for fault diagnosis of multi-source monitoring data. First, composite multivariate multi-scale symbolic dynamic entropy is proposed to extract fault features. Second, Laplacian score is introduced to select the distinguishing features with better clustering ability. Finally, the selected features are fed into a logistic regression classifier so that various faults of machinery are diagnosed. The simulation and two case studies using gearbox and pump data are performed to validate and demonstrate the superiority of the proposed method.