ePrints@IIScePrints@IISc Home | About | Browse | Latest Additions | Advanced Search | Contact | Help

Jet Engine Health Signal Denoising Using Optimally Weighted Recursive Median Filters

Uday, Payuna and Ganguli, Ranjan (2010) Jet Engine Health Signal Denoising Using Optimally Weighted Recursive Median Filters. In: Journal of Engineering for Gas Turbines & Power, 132 (4).

[img] PDF
GetPDFServlet.pdf - Published Version
Restricted to Registered users only

Download (480kB) | Request a copy
Official URL: http://asmedl.aip.org/getabs/servlet/GetabsServlet...


The removal of noise and outliers from health signals is an important problem in jet engine health monitoring. Typically, health signals are time series of damage indicators, which can be sensor measurements or features derived from such measurements. Sharp or sudden changes in health signals can represent abrupt faults and long term deterioration in the system is typical of gradual faults. Simple linear filters tend to smooth out the sharp trend shifts in jet engine signals and are also not good for outlier removal. We propose new optimally designed nonlinear weighted recursive median filters for noise removal from typical health signals of jet engines. Signals for abrupt and gradual faults and with transient data are considered. Numerical results are obtained for a jet engine and show that preprocessing of health signals using the proposed filter significantly removes Gaussian noise and outliers and could therefore greatly improve the accuracy of diagnostic systems. [DOI: 10.1115/1.3200907].

Item Type: Journal Article
Publication: Journal of Engineering for Gas Turbines & Power
Publisher: The American Society of Mechanical Engineers
Additional Information: Copyright of this article belongs to The American Society of Mechanical Engineers.
Department/Centre: Division of Mechanical Sciences > Aerospace Engineering(Formerly Aeronautical Engineering)
Date Deposited: 22 Feb 2010 09:54
Last Modified: 19 Sep 2010 05:55
URI: http://eprints.iisc.ac.in/id/eprint/25832

Actions (login required)

View Item View Item