Two-shaft stationary gas turbine engine gas path diagnostics using fuzzy logic

Our objective was to develop a Fuzzy logic (FL) based industrial two-shaft gas turbine gas path diagnostic method based on gas path measurement deviations. Unlike most of the available FL based diagnostic techniques, the proposed method focused on a quantitative analysis of both single and multiple...

Full description

Saved in:
Bibliographic Details
Published in:Journal of mechanical science and technology 2017, 31(11), , pp.5593-5602
Main Authors: Amare, F. D., Gilani, S. I., Aklilu, B. T., Mojahid, A.
Format: Article
Language:English
Subjects:
Control
Diagnostic systems
Dynamical Systems
Engineering
Erosion mechanisms
Faults
Fuzzy logic
Gas turbine engines
GE engines
Industrial and Production Engineering
Mechanical Engineering
Natural gas
Quantitative analysis
Random noise
Vibration
기계공학
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:Our objective was to develop a Fuzzy logic (FL) based industrial two-shaft gas turbine gas path diagnostic method based on gas path measurement deviations. Unlike most of the available FL based diagnostic techniques, the proposed method focused on a quantitative analysis of both single and multiple component faults. The data required to demonstrate and verify the method was generated from a simulation program, tuned to represent a GE LM2500 engine running at an existing oil & gas plant, taking into account the two most common engine degradation causes, fouling and erosion. Gaussian noise is superimposed into the data to account measurement uncertainty. Finally, the fault isolation and quantification effectiveness of the proposed method was tested for single, double and triple component fault scenarios. The test results show that the implanted single, double and triple component fault case patterns are isolated with an average success rate of 96 %, 92 % and 89 % and quantified with an average accuracy of 83 %, 80 % and 78.5 %, respectively.
ISSN:1738-494X
1976-3824
DOI:10.1007/s12206-017-1053-9