Physics-Based Model And Neural Network Model For Monitoring Starter Degradation Of Apu

An electric starter provides the initial power to run the Auxiliary Power Unit (APU) during the startup process. With the starter degradation, its output power declines, which affects the APU starting performance, and eventually leads to the starting failure. Previous works have attempted to estimate the starter degrading trend, however a clear symptom for the starter degradation is not provided, which can be instrumental for the preventive maintenance. This paper develops a physics-based transient model to assess the starter degradation using the gas-path measurements of the APU. To overcome shortcomings for the lack of component characteristics, a generic modeling approach is adopted. For the comparative study, a back-propagation, feedforward neural network model is structured, trained and tested. Both models are implemented in the nominal and degraded conditions, and their capabilities as monitoring tools for the starter degradation are verified. The physics-based approach provides more accurate results for the cases with degraded starters, whereas the neural network model shows superior results with the starters in healthy condition.