Aeroengine Diagnosis Using a New Robust Gradient-like Methodology

Abstract

A new gradient-like methodology has been developed for aeroengine diagnosis, determining the engine health condition, which is defined by the engine degradation from an undegraded state and uses measurements at various sensors distributed along the engine. The developed tools are able to accurately compute, not only the engine degradation, but also the turbine inlet temperature, which is very important and novel in the field. The quality of a given sensors set is evaluated, and a method is developed to guide in the improvement of deficient sensors sets. The methodology is tested in a representative two-spool turbofan engine, obtaining consistent results in a computationally inexpensive way. Moreover, results are robust in connection with the random noise added to the sensors data.

Keywords

• aeroengine
• predictive maintenance
• model-based diagnosis
• health monitoring
• gradient-like optimization
• adapted Newton method