Spare optimistic based on improved ADMM and the minimum entropy de-convolution for the early weak fault diagnosis of bearings in marine systems

Yangde Gao; Mohammad Karimi; Aleksey A Kudreyko; Wanqing Song

doi:10.1016/j.isatra.2017.12.021

Spare optimistic based on improved ADMM and the minimum entropy de-convolution for the early weak fault diagnosis of bearings in marine systems

ISA Trans. 2018 Jul:78:98-104. doi: 10.1016/j.isatra.2017.12.021. Epub 2017 Dec 30.

Authors

Yangde Gao¹, Mohammad Karimi², Aleksey A Kudreyko³, Wanqing Song⁴

Affiliations

¹ The School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, 333, Long Teng Road, Shanghai, 201620, China.
² Young Researchers and Elite Club, South Tehran Branch, Islamic Azad University, Tehran, Iran.
³ Ufa State Petroleum Technological University, Department of Physics, Kosmonavtov 1, 450062, Ufa, Russia.
⁴ The School of Electronic and Electrical Engineering, Shanghai University of Engineering Science, 333, Long Teng Road, Shanghai, 201620, China. Electronic address: swqls@126.com.

PMID: 29295740
DOI: 10.1016/j.isatra.2017.12.021

Abstract

In the marine systems, engines represent the most important part of ships, the probability of the bearings fault is the highest in the engines, so in the bearing vibration analysis, early weak fault detection is very important for long term monitoring. In this paper, we propose a novel method to solve the early weak fault diagnosis of bearing. Firstly, we should improve the alternating direction method of multipliers (ADMM), structure of the traditional ADMM is changed, and then the improved ADMM is applied to the compressed sensing (CS) theory, which realizes the sparse optimization of bearing signal for a mount of data. After the sparse signal is reconstructed, the calculated signal is restored with the minimum entropy de-convolution (MED) to get clear fault information. Finally we adopt the sample entropy. Morphological mean square amplitude and the root mean square (RMS) to find the early fault diagnosis of bearing respectively, at the same time, we plot the Boxplot comparison chart to find the best of the three indicators. The experimental results prove that the proposed method can effectively identify the early weak fault diagnosis.

Keywords: Boxplot; CS; Early weak fault diagnosis; Improved ADMM; MED; Marine systems; Morphological mean square amplitude; RMS; Sample entropy; Spare optimistic.