Free Download Engineering | Science | Technology Books

File : pdf, 695 KB, 60 pages

by David Clifton (eng.ox.ac.uk)

Summary

Condition monitoring assesses the operational health of gas-turbine engines, in order to provide early warning of potential failure such that preventative maintenance action may be taken. Gas-turbine engine manufacturers are increasingly offering a “service-based” approach to marketing their products, in which their customers are guaranteed certain availability of the engine after purchase. To achieve this, manufacturers take on the responsibilities of engine condition monitoring, by embedding health monitoring systems within each engine unit and prompting maintenance actions when necessary.

This report describes preliminary research into condition monitoring approaches for modern gas-turbine aircraft engines, and outlines plans for novel research to contribute to machine learning techniques in the condition monitoring of such systems, leading to the D.Phil. degree.
A framework for condition monitoring of aircraft engines is introduced, using signatures of engine vibration across a range of engine speeds to assess engine health. Inter- and intra-engine monitoring approaches are presented, in which a model of engine normality is constructed using vibration data from other engines of its class, or from the test engine itself, respectively.
Results of inter-engine analysis of  final engine vibration tests prior to their release into service are presented, showing that the approach described within this report provides a more reliable estimate of engine condition than manufacturers’ conventional engine vibration tests, leading to better discrimination between “good” and “bad” engines.

Intra-engine analysis of an engine undergoing cyclic endurance testing, in which a set operational manoeuvre is performed repeatedly, shows that the method described in this report provides early warning of engine failure that eventually resulted in a hazardous engine fire, undetected by engine developers until it had to be shut down manually.

Future research is planned in application of this condition monitoring framework to an engine currently under development, improving upon existing methods and investigating new approaches, ultimately leading to the formulation of a general “black box” monitoring approach that can learn a model of system normality without prior knowledge of that system.

TOC

1 The Need for Condition Monitoring
1.1 Introduction
1.2 Aerospace Gas-turbine Engines
1.3 Existing Techniques for Condition Monitoring of Aerospace Gas-turbine Engines
1.4 Overview of this Report

2 Novelty Detection for Gas-turbine Engines
2.1 Introduction
2.2 Modelling Normality
2.3 Novelty Detection using TORCH

3 TORCH for Inter-engine Analysis
3.1 Introduction
3.2 Data Set of Vibration Signatures
3.3 Vectorisation
3.4 Normalisation
3.5 Visualisation
3.6 Clustering
3.7 Choosing a Model of Normality
3.8 Calculation of Novelty
3.9 Conclusion

4 TORCH for Intra-Engine Analysis
4.1 Introduction
4.2 Constructing TORCH Signatures of Vibration Phase
4.3 Intra-engine Analysis

5 Future Work 50
5.1 Flight Summary Compression
5.2 Improved Novelty Detection Methods
5.3 Time- and Time-Frequency Domain Analysis
5.4 Data Fusion, and Oil Debris Monitoring Systems (ODMS)
5.5 On-Line Learning
5.6 DPhil Project Plan

Download      :      pdf1