Development of an experimental test bench for the validation of prognostic algorithms for electromechanical actuators

??In the recent times, the aerospace industry is moving toward the development of more electric aircraft. This shift involves a transition from an hydraulic to an electric type of actuation. Conversion that promises to reduce the overall weight of the aircraft while increasing the efficiency of the system. However, the use of electric-type actuators called EMAs is currently limited to non-safety critical functions given the still basic insight into the behavior of these systems. ??To compensate for the lack of confidence in the EMAs systems, the use of prognostic techniques has found ample opportunity. The model-based approach described is intended to reproduce the behavior of an actuation characterized by reaction times that fall somewhere between those of a primary and a secondary flight control system. The use of a model makes it possible to reproduce the performance of the system with high accuracy but renders necessary the verification of its fidelity to the actual system. ??To perform model verification, a test bench was therefore assembled, the general architecture of which can be divided into two segments: the one representing the EMA system and another one delegated to the generation of resistive torques. The control architecture selected consisted of closed-loop position control. To this purpose, the system was fitted with an encoder placed on the output shaft of the motor-gearbox assembly. The encoder installation had been designed in such a way that the variation in system backlash could be reproduced in a controlled manner in order to be able to observe its effects on the actuator response. ??The comparison of the data received from the model with those acquired from the test bench provided a means to verify the soundness of the model. The primary variables tested included position, velocity, and current. In all of these cases, a good degree of accuracy could be observed, both in terms of amplitude and timing. A notable aspect that emerged was the fact that the evolution of deviations between model and bench is concentrated at high values of play. However, the high capacity to simulate the behavior of the real system is counterbalanced by the associated computational cost.