A novel procedure for diagnosing multiple faults in rotating machinery

Zhijian Wang; Zhennan Han; Fengshou Gu; James Xi Gu; Shaohui Ning

doi:10.1016/j.isatra.2014.09.006

A novel procedure for diagnosing multiple faults in rotating machinery

ISA Trans. 2015 Mar:55:208-18. doi: 10.1016/j.isatra.2014.09.006. Epub 2014 Nov 11.

Authors

Zhijian Wang¹, Zhennan Han², Fengshou Gu³, James Xi Gu⁴, Shaohui Ning⁵

Affiliations

¹ College of Mechanical Engineering, Taiyuan University of Technology, Taiyuan, Shanxi 030024, PR China.
² College of Mechanical Engineering, Taiyuan University of Technology, Taiyuan, Shanxi 030024, PR China. Electronic address: wangzhijian1013@163.com.
³ College of Mechanical Engineering, Taiyuan University of Technology, Taiyuan, Shanxi 030024, PR China; School of Computing and Engineering, University of Huddersfield, Huddersfield HD1 3DH, UK.
⁴ Signal Processing and Algorithms Group, School of Engineering, Manchester Metropolitan University, All Saints Building, All Saints, Manchester M15 6BH, UK.
⁵ College of Mechanical Engineering, Taiyuan University of Science and Technology, Taiyuan, Shanxi 030024, PR China.

PMID: 25440946
DOI: 10.1016/j.isatra.2014.09.006

Abstract

In analyzing signals from a wind turbine gearbox this paper suggests a new signal processing procedure named as CMF-EEMD method which is formed by applying conventional EEMD to a new type of combined mode function (CMF). This CMF consists of a low frequency CMF, denoted as CL, and a high frequency CMF, denoted as Ch. Then it optimizes the amplitude of the added noise in decomposing Ch and CL using EEMD. Finally, it calculates cyclic autocorrelation function (CAF) for every characteristic IMF from EEMD. The proposed procedure is applied to analyze the multi-faults of a wind turbine gearbox and the results confirm better performances in resolving different signal components by the proposed method than that from the cyclic autocorrelation function (CAF) of a direct EEMD analysis.

Keywords: CAF; CMF; EEMD; Fault diagnosis.

Publication types

Research Support, Non-U.S. Gov't