CORDIS fornisce collegamenti ai risultati finali pubblici e alle pubblicazioni dei progetti ORIZZONTE.
I link ai risultati e alle pubblicazioni dei progetti del 7° PQ, così come i link ad alcuni tipi di risultati specifici come dataset e software, sono recuperati dinamicamente da .OpenAIRE .
Risultati sfruttabili
The OBIDICOTE project has allowed to generate a common non-linear gas turbine jet engine simulation tool and to develop a common methodology in the field of diagnosis and engine monitoring. This project has created a common basis for future collaborations in modelling, diagnosis and controls.
A first important result has been demonstrated: the non-linear model is running satisfactorily (fast and accurate) on equivalent conditions corresponding to on-board computer operations. An improvement in the accuracy of diagnosis and monitoring system has been achieved.
On the basis of an a priori information obtained with the engine physical model, new developments of measurement validation method have been conducted: the multivariate statistical method, the pattern recognition method and the probabilistic neural network method.
The efficiency of these methods has been successfully demonstrated on the basis of several test cases including sensors corrupted with different types of faults (biases, slow drifts, and accuracy degradations).
Four diagnosis and fault (engine component characteristics) identification methods, based on the non-linear model, have been studied to:
-Update the model to predict current engine performance.
- Diagnose faults, which occur rapidly due to damage, breakage or failure.
- Identify gradual deterioration.
- Tune an average engine model, to represent a specific engine.
All the new methods out-performed the current techniques (Kalman Filter for instance).
When detecting deterioration, one method clearly outclasses the others; but, at the same time, this method was less efficient to fulfil the other requirements. Therefore, it should be convenient to use in parallel at least two methods, in order to cover all needs.
Innovative mathematical methods based on Neural Networks and Belief Networks have been analysed and implemented, in order to develop a tool able to perform an automatic diagnosis of a reference turbine engine, by analysing the gas path. On the basis of the above-mentioned tools and methods, various applications have been developed.
An optimised adaptive fault tolerance control based on the non-linear model has been designed and implemented. These studies have illustrated new capabilities to maximise and enhance engine performance (thrust, turbine temperature, surge margin, etc.), fault tolerance (by analytical estimation of missing measurements) and component life.
Methods using information provided by the onboard engine model have been studied:
-The use of engine model signals as "virtual" measurements ("model based adaptive control").
- New control structures using the onboard model to separate the demand following part of the control system from the disturbance rejection part ("model reference control").
- Fault tolerant adaptive systems.
Thanks to these results, all OBIDICOTE objectives can be considered as met. The studies conducted and the various simulations achieved have demonstrated that complex dynamical and non-linear models, running on small on- board computers, will be able to improve the accuracy and quality of engine control and component diagnosis.
È in corso la ricerca di dati su OpenAIRE...
Si è verificato un errore durante la ricerca dei dati su OpenAIRE
Nessun risultato disponibile