Vibration feature extraction for hydropower units based on ensemble empirical mode decomposition and sample entropy

Read full text Summarise with VIRAT
Authors: Baiyun Qian, Chunyi Lu, Chunxu Yang, Xiang Lin, Xiaoxun Zhu, Zeqi Shi, Jia Gao, Qiaozhen Wang
Cited by: 0
References: 5

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Abstract Feature extraction from hydropower units’ non-stationary/nonlinear vibration signals is fundamental for condition monitoring and fault diagnosis, which is a key hydropower research focus. This paper proposes an EEMD-SampEn-GRNN method. First, the EEMD algorithm decomposes denoised signals. Then effective IMFs are screened using the correlation coefficient. The sample entropy of these IMFs is calculated as features and input into the GRNN model for recognition. Compared with EMD-SampEn, it achieves 100% maximum and 98.85% average accuracy (vs. 93.68% and 88.54%), with superior performance and stability, enabling fault pre-warning and ensuring safe unit operation.

Fields of Study & Subjects

Machine Fault Diagnosis Techniques Fault Detection and Control Systems Power Transformer Diagnostics and Insulation

Publication details

Journal / source
Journal of Physics Conference Series
Volume
3198
Issue
1
Pages
012051–012051
Publication date
1 March 2026
Language
EN