Lifetime control of electromechanical actuators

A modeling framework for deteriorating control system and predictive maintenance of actuators

“…It is assumed hereafter that the control input u(t) and its variations participate in the evolution of the actuator's degradation. This viewpoint agrees with [5,19] where it is explained that there is a physical relation between u(t) and the level of actuator's degradation. In these papers, the studied system is an electrical motor where the current input u(t) both impacts the winding temperature and the bearing load.…”

“…This indicator is the result output of a feature fusion process of the vibration signals such as principal component analysis used in [34]. Finally, acting on the motor angular speed u(t) controls the bearing's lifetime [5].…”

“…A second step is to consider stochastic degradation models. In [5,7] for example, a physical and deterministic relation between the control law and the actuator's degradation level is pointed out, but without taking into account any uncertainty. Moreover, the degradation does not imply a loss of effectiveness to fully implement control.…”

“…In general, required performance is obtained by expending high actuator energy with no-control on the actuator lifetime. Only some research has already investigated this problem and give partial answers [4][5][6][7][8]: the degradation of the actuator is modeled by a loss of effectiveness to fully implement the control input but this kind of degradation is (i) independent on the control law, (ii) independent on any kind of physical degradation process, (iii) governed by deterministic fault occurrence dates. So it is difficult to use this type of model to take into account random usage or operational condition variations in the future.…”

EMA remaining useful life prediction with weighted bagging GPR algorithm

“…Prognostic and Health Management (PHM) can provide advantages for the applications of EMAs [15,16,17]. Furthermore, it can also make the adoption of EMAs in vital applications more feasible due to the aspiration of minimizing failures [18,19]. RUL prediction is a crucial part of PHM [20].…”