Compound feature selection and parameter optimization of ELM for fault diagnosis of rolling element bearings

Meng Luo; Chaoshun Li; Xiaoyuan Zhang; Ruhai Li; Xueli An

doi:10.1016/j.isatra.2016.08.022

Compound feature selection and parameter optimization of ELM for fault diagnosis of rolling element bearings

ISA Trans. 2016 Nov:65:556-566. doi: 10.1016/j.isatra.2016.08.022. Epub 2016 Sep 9.

Authors

Meng Luo¹, Chaoshun Li², Xiaoyuan Zhang³, Ruhai Li¹, Xueli An⁴

Affiliations

¹ School of Hydropower and Information Engineering, Huazhong University of Science and Technology, Wuhan 430074, PR China.
² School of Hydropower and Information Engineering, Huazhong University of Science and Technology, Wuhan 430074, PR China. Electronic address: csli@hust.edu.cn.
³ College of Electrical Engineering, Henan University of Technology, Zhengzhou 450001, PR China.
⁴ China Institute of Water Resources and Hydropower Research, Beijing 100044, PR China.

PMID: 27622428
DOI: 10.1016/j.isatra.2016.08.022

Abstract

This paper proposes a hybrid system named as HGSA-ELM for fault diagnosis of rolling element bearings, in which real-valued gravitational search algorithm (RGSA) is employed to optimize the input weights and bias of ELM, and the binary-valued of GSA (BGSA) is used to select important features from a compound feature set. Three types fault features, namely time and frequency features, energy features and singular value features, are extracted to compose the compound feature set by applying ensemble empirical mode decomposition (EEMD). For fault diagnosis of a typical rolling element bearing system with 56 working condition, comparative experiments were designed to evaluate the proposed method. And results show that HGSA-ELM achieves significant high classification accuracy compared with its original version and methods in literatures.

Keywords: Ensemble empirical mode decomposition; Extreme learning machine; Fault diagnosis; Feature selection; Gravitational search algorithm; Parameter optimization.