Aircraft APU Prognostic Health Monitoring Utilizing Physics-Based Model

While diagnostic solutions for an aircraft Auxiliary Power Unit (APU) turbine system have reached relatively high level of maturity, accurate prediction of Remaining Useful Life (RUL) remains a challenging task. This is mainly due to the high complexity of the system, limited number of measured variables and significant diversion of operating conditions. This paper deals with a hybrid approach to APU Prognostic Health Monitoring (PHM) system development employing both, a physics-based model of APU system available to Honeywell as the engine manufacturer and historical data from APUs operations and repair shop visits. Two different Condition Indicators (CIs), the model-based and the data-driven, are developed to estimate the level of degradation of APU core section. The CIs are fed into binary classification algorithm which is used to distinguish if the RUL of an asset is lower than a defined threshold; this representation of Prognostic Indicator (PI) can be deemed as a maintenance recommendation flag provided to APU operator to schedule a maintenance action