An artificial neural network position estimator for a variable reluctance linear actuator

A neural network approach to the position estimation problem for a variable reluctance linear actuator is investigated. The inputs of the neural network are current ripple measurements and the switching pattern of the power converter. The neural network approach allows to account for iron saturation and iron losses effects resulting in an accurate position estimation. Numerical simulations are developed to show the feasibility of the proposed method, as well as the neural robustness. Finally, experimental measurements on a prototype are presented to validate the proposed approach.