• model-based
• fault-tolerant
• FFT
• health-monitoring
• spalling
• prognostic algorithms

In recent years,the importance of predictive maintenance has been remarkably increased thanks to the application of real-time detection and prediction algorithms regarding future failures in different scenarios.
In this work, a model-based prognostic health-monitoring algorithms is developed aiming to evaluate the phenomenon of spalling damage presents in the PMSMs mechanical transmission for a UAV propulsion.
In order to achieve this goal, an analysis of the acceleration signals of the propeller mechanical shaft is performed via FFT signal processing technique by identifying the resonant peaks related to possible damages.
Firstly, the conceptual design of the algorithm is developed by the means of a reduced-order model system of the mechanical transmission with the injection of the spalling damage.Analysing this model,trends of the FFT resonant peaks are evaluated and limited operating velocities are defined.Secondly,applying the developed algorithms to the PMSMs validated model, FFT peaks maps are created as function of the spalling geometry at the limited operating velocities.Finally,a validation of the robustness algorithms is performed by the injection of electrical,electromagnetic,mechanical and sensory disturbances.
In conclusion,results are discussed underlying the strengths and limitations of the algorithms.