Incipient loose detection of hoops for pipeline based on ensemble empirical mode decomposition and multi-scale entropy and extreme learning machine

Xiaowei Li, Qin Wei, Yongzhi Qu, Lin Cai

Research output: Contribution to journalConference article

3 Scopus citations

Abstract

Hoops are very important fittings in hydraulic pipeline, incipient loose detection of hoops will help to prevent hydraulic piping system from breaking down. Since the vibration signals of fluid pipe are non-stationary and of great complexity, multi-scale entropy(MSE), a method characterized by evaluating complexity and irregularity of time series on multiple scales, is used for extracting feature vectors from the vibration signals. In order to obtain components related to system characteristics, ensemble empirical mode decomposition(EEMD) is applied to reconstruct the original signals before the procedure of MSE. Extreme learning machine(ELM) is a new machine learning algorithm characterized by high accuracy and efficiency. In this paper, ELM is introduced as a classifier to identify the different conditions of hoops according to feature vectors extracted by EEMD and MSE algorithms. Thus a novel loose detection method combining with EEMD-MSE and ELM is put forward. The analysis and experimental results demonstrate that the proposed loose detection and feature extraction method for hydraulic pipeline is effective with high performance.

Original languageEnglish (US)
Article number012011
JournalIOP Conference Series: Materials Science and Engineering
Volume211
Issue number1
DOIs
StatePublished - Jun 27 2017
Externally publishedYes
Event1st International Conference on Aerospace, Mechanical and Mechatronic Engineering, CAMME 2017 - Bangkok, Thailand
Duration: Apr 21 2017Apr 23 2017

Fingerprint Dive into the research topics of 'Incipient loose detection of hoops for pipeline based on ensemble empirical mode decomposition and multi-scale entropy and extreme learning machine'. Together they form a unique fingerprint.

  • Cite this