Rauch-Tung-Striebel Smoother for Position Estimation of Short-Stroke Reluctance Actuators

This article presents a novel state estimator for short-stroke reluctance actuators, intended for soft-landing control applications in which the position cannot be measured in real time. One of the most important contributions regards the system modeling for the estimator. The discrete state of the hybrid system is treated as an input. Moreover, the model is simplified to facilitate the identification of parameters and the implementation of the estimator. Thus, auxiliary variables are added to the state vector in order to indirectly account for modeling errors. Another important contribution is the state estimation approach. It is based on the Rauch–Tung–Striebel fixed-interval smoother, which allows refining past data from later observations. Numerous simulations are performed to analyze and compare the proposal and several alternatives. In addition, experimental testing is presented to evaluate and validate the estimator. As the simulated and experimental analyses demonstrate, the combined effect of the novel additions results in significantly smaller estimation errors of position and velocity.