Monitorización y predicción del estado en flotas de motores usando análisis inteligente de datos para información intervalo-valorada y posibilística

Abstract

Engine health monitoring data is at the current core of the civil aeronautical business. The use of engine health monitoring systems has existed for several years, however it is only now that the in-flight knowledge gathered through this means is being used to address not only safety and reliability but to also understand customer operation and the overall engine condition. The assessment of EHM data for optimized life cycle cost, this is, the extension of an engine¿s time on wing and reduction of maintenance costs, has not yet been fully exploited due to several reasons. The main reasons however have been the lack of available data and the time and material type of maintenance operation common until now. Modern technology and a change towards TotalCare have influenced the current importance of EHM and its detailed assessment developments. This thesis develops a new EHM assessment methodology and its associated prognosis with the main objective of improving the level of engine maintenance required detail for a given engine prior to its maintenance shop visit. In addition, the prognosis methodology provides a significant long term capability on the state of the engine at module level which enables trade studies, not possible today. The existing EHM assessment capabilities concentrate on the safety and reliability aspects of engine containment and its reactive capabilities. The EHM methods developed are a proactive approach towards interpreting EHM data in its full extent, without filtering, in order to determine the actual condition of an engine at a modular level. The application of existing methods as BSS, and subsequently a possibilistic filter together with a fuzzy classifier, have enabled a new approach at understanding the internal engine condition through the combined assessment of all of the available variables. A subsequent classification method which enables the association of this level of deterioration to a known state or level of deterioration allows for a prediction of the level of engine deterioration, expected cost of maintenance and main exchanged parts to be replaced, to be performed. The assessment of events or condition changes does not directly reflect the deterioration of the engine. A sequence mining approach to the previous results obtained above was carried out to establish if certain sequences may be associated to specific levels of deterioration, transitions or to no significant changes. The additional knowledge provided through these methodologies to the current business is already a significant step change. A prognosis however was developed associated to this engine condition assessment which further enables the detailed understanding of the engine remaining useful life based on the integral of the instantaneous engine deterioration speed and the life objective established on its release. The result from this assessment is a new set of methods, which allows the maintenance facilities to optimize their limited capacity and predict in detail the level of workscope and man-hours to be employed for specific engine refurbishments. These methods also allow trade studies to be performed to optimize time on-wing versus over all engine level of deterioration.