Bearing Real-Time Condition Monitoring and Fault Feature Extraction Using SAX-LZC

Aiming at the need for real-time monitoring of bearing vibration, the monitoring parameter which fused symbolic aggregation approximation and Lempel-ziv complexity is proposed from symbolic dynamical point of view. Using Logistic map and Duffing equations as objects, the accuracy of SAX-LZC for dynamic structural characterization isverified from theoretical perspective, and the anti-noise ability and computational efficiency are verified also. Based on this, a comprehensive comparison of SAX-LZC with dynamic parameters such as information entropy, sample entropy, and multi-segment Lempel-ziv complexity is performed. The weak early abnormality of the bearing is monitored and the faults characteristic of the bearing are extracted from the experimental perspective. Theoretical research results show that SAX-LZC has the advantages of accurate dynamic structural characterization, good anti-noise ability, and high computational efficiency. It overcomes the weak application of conventional dynamic parameters. The experimental results show that the SAX-LZC accurately monitors early weak anomalies, and has good distinguishing ability for different types of faults. It remedies the shortcomings of time and frequency domains' insufficient representation ability to characterize bearing weak abnormalities. Therefore, it is an effective parameter for real-time monitoring and fault feature extraction of bearing vibration. © 2020, Editorial Department of JVMD. All right reserved.