State Estimation Based on Self-Triggered Measurements
Résumé
In this work, the problem of state estimation for nonlinear continuous-time systems from discrete data is tackled in a bounded error context. One assumes that all poorly-known system variables belong to a bounded set with known bounds. Then, a self-triggered algorithm is proposed to improve the performance of the classical set-membership state estimator based on the prediction-correction procedures. In order to cope with pessimism propagation linked to the bounding methods, this algorithm triggers the correction step whenever the size of a part of the estimated state enclosure becomes greater than a time-converging threshold a priori defined by the user. The effectiveness of the proposed self-triggered algorithm is illustrated through numerical simulations.