A novel identification method of Volterra series in rotor-bearing system for fault diagnosis. (January 2016)

• Record Type: Journal Article
• Authors: Xia, Xin; Zhou, Jianzhong; Xiao, Jian; Xiao, Han
• Is Part Of: Mechanical systems and signal processing. Volume 66/67(2016)
• Page Start: 557
• Page End: 567
• Publication Date: 2016-01
• Subjects: Volterra series -- Fault diagnosis -- KK-PSO -- Rotor-bearing system; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2015.05.006
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: Volterra series is widely employed in the fault diagnosis of rotor-bearing system to prevent dangerous accidents and improve economic efficiency. The identification of the Volterra series involves the infinite-solution problems which is caused by the periodic characteristic of the excitation signal of rotor-bearing system. But this problem has not been considered in the current identification methods of the Volterra series. In this paper, a key kernels-PSO (KK-PSO) method is proposed for Volterra series identification. Instead of identifying the Volterra series directly, the key kernels of Volterra are found out to simply the Volterra model firstly. Then, the Volterra series with the simplest formation is identified by the PSO method. Next, simulation verification is utilized to verify the feasibility and effectiveness of the KK-PSO method by comparison to the least square (LS) method and traditional PSO method. Finally, experimental tests have been done to get the Volterra series of a rotor-bearing test rig in different states, and a fault diagnosis system is built with a neural network to classify different fault conditions by the kernels of the Volterra series. The analysis results indicate that the KK-PSO method performs good capability on the identification of Volterra series of rotor-bearing system, and the proposed method can further improve the accuracy of fault diagnosis. Highlights: The KK-PSO method can get the Volterra model effectively under sinusoidal input. The Volterra model is predigested which make it more effectively. Different Volterra models are built for rotor-bearing system in different states. Kernels of Volterra model are used for fault diagnosis with a neural network.