Study on Gas Turbine Gas-Path Fault Diagnosis Method Based on Quadratic Entropy Feature Extraction

To avoid disrepair and over-repair and improve gas turbine reliability and availability, gas-path diagnosis is an effective technical means of disseminating early warning information for evolving or impending deterioration. Aiming at the problems of existing gas-path diagnosis methods (i.e., data-dr...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access 2019, Vol.7, p.89118-89127
Main Authors: Li, Jingchao, Ying, Yulong, Ji, Chunlei
Format: Article
Language:English
Subjects:
Degradation
Diagnostic systems
Dimensional changes
Entropy
Entropy (Information theory)
Fault diagnosis
Feature extraction
Gas turbine
Gas turbines
gas-path diagnosis
Genetic algorithms
Maintenance engineering
quadratic feature extraction
Turbines
visualization
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:To avoid disrepair and over-repair and improve gas turbine reliability and availability, gas-path diagnosis is an effective technical means of disseminating early warning information for evolving or impending deterioration. Aiming at the problems of existing gas-path diagnosis methods (i.e., data-driven based gas-path diagnosis and a model-based gas-path diagnosis), this paper proposes a novel gas-path diagnosis method based on model-data hybrid drive, which is a forward solving mathematical process, to ensure real-time monitoring performance. Through case analysis, the proposed diagnostic method is not limited by the intrinsic nonlinear shape change of the characteristic maps of the actual component, which has good diagnostic applicability. And after the quadratic feature extraction of the two-dimensional entropy features (i.e., Shannon entropy and exponential entropy features), it is convenient to obtain visualized gas turbine gas-path diagnosis results for the operation and maintenance personnel. Moreover, although the extracted two-dimensional entropy values will change slightly when the operating conditions change, it can maintain good inter-class separation and intra-class aggregation performance.
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2019.2927306