Use this URL to cite or link to this record in EThOS:
Title: Performance diagnostics and measurement selection for on-line monitoring of gas turbine engines
Author: Bechini, Giovanni
ISNI:       0000 0004 2712 8098
Awarding Body: Cranfield University
Current Institution: Cranfield University
Date of Award: 2007
Availability of Full Text:
Access from EThOS:
Access from Institution:
The increasing importance of maintenance planning and optimization in the current and future scenario of gas turbine aftermarket makes the gas turbine analyst aware of the benefits associated with an effective health monitoring system. This thesis reviews today’s gas-path diagnostic methods in order to investigate the shortcomings and limitations regarding their capability of reducing downtime, increasing availability and minimizing life cycle costs of the engine. Having identified drawbacks in the implementation of existing approaches, a novel design procedure is proposed for an innovative diagnostic method, aimed to close the gaps left by current technologies. This procedure is based on a pattern recognition process supported by a non-linear observability analysis for measurement selection. The importance of providing the diagnostic system with the necessary information to perform an accurate diagnosis is emphasized, and the impact of different measurement set on the accuracy of the diagnosis is studied, resulting in the isolation of the optimal set for monitoring purpose. Different from previous studies, this diagnostic method features an innovative fusion between probabilistic-stochastic algorithms (Bayesian Probability and Probability Density Estimation) and Artificial Intelligence (Fuzzy Logic). These tools are embedded within a logical frame similar to a Bayesian Belief Network, where a performance model of the engine plays a role in the set-up phase. Gas turbine users and manufacturers require enhanced levels of accuracy (for multiple faults isolation), speed (for on-line monitoring) and data-fusion capability (to integrate the diagnostic system with external sources of information), and this method is specifically designed to meet those requirements to a higher extent. The robustness of the analysis is demonstrated through extensive numerical tests using simulated data from two different engines for aero and industrial applications. The gas turbine community will benefit from the novelty of this work which has resulted in the submission of a patent application to the UK Patent Office.
Supervisor: Singh, R. Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available