• predictive maintenance
• EMA
• fault-tolerant
• spalling
• prognostic algorithm
• Rul

The main purpose of this thesis is the development of a model-based prognostic algorithm aiming to evaluate the spalling of metallic contacts in the ball-screw mechanism of a fault-tolerant electromechanical actuator designed and manufactured by Umbragroup SPA.
The basic idea underlying the monitoring technique is to perform maintenance built-in tests on the actuator, in which the accelerations at different points in the mechanical transmission are measured while the system is commanded at constant speed. The accelerometric signals are then analysed via FFT, taking into account the limitations related to sampling errors, by identifying the resonant peaks related to possible damages. The monitoring algorithm has been designed to provide an estimation of the Remaining Useful Life of the actuator by performing four tests, two ones at low speed sand two ones at high speeds.
The prognostic algorithm has been developed in two steps: in the first one, the general strategy of the algorithm (generation of FFT peaks map as function of damage characteristics, test type, identification technique) is defined by using a LTI model of the mechanical transmission with spalling damage; in the second step, the algorithm is verified by using an experimentally-validated model of the actuator, including nonlinearities and sensors’ errors. Actually, the quality of sensors plays a key role for the applicability of the technique, especially for the detrimental effects of noise and linearity errors. Once the damage is injected in the detailed nonlinear model, the RUL accuracy and the limitations of the algorithm are characterised and discussed.