Modelling and simulation of faulty electromechanical actuators

The PHM (Prognostics and Health Management) is one of the most innovative technique for the prediction of the fault/failure occurence in a component through the diagnosis and its Remaining Useful Life (RUL) by the prognosis. Recently, there has been growing interest in the improvement of the requirements in terms of system performance and safety. For this purpose, the Model Predictive Control method was presented here as an effective control strategy, which uses prognostic data to extend the RUL of the component. In this Thesis, the basis for the PHM routines are provided, performing the initial step of the diagnosis, that is the fault isolation and identification, on a PMSM linear electromechanical actuator (EMA). In this setting the fault presence, represented by the rotor static eccentricity, was simulated on a MATLAB/Simulink model of the system. Then, the Fast Fourier Transform (FFT) analysis was used for the fault identification, exploiting the frequency spectrum of the motor phase current. At the end, this work suggests a first attempt of feature extraction, which should characterize the fault behavior as a function of system parameters. This will lead to the fulfilment of the diagnosis phase and to the start of the prognosis one, which is not subject of this study.