Robust state estimation in networked control systems under linear hard equality constraints

This paper considers state estimation with networked control systems. This problem is addressed with regards to combined issues of hard equality constrained state, incomplete delayed observations and unknown but bounded varying process noise. The packet arrival process is modelled with a stochastic process supposed known. An analytic relationship between the filter convergence and the packet arrival process is provided based on the a priori and the a posteriori system model forms. The corresponding convergence problem is solved on condition that there are positive definite solutions to algebraic Kalman-Riccati-like inequalities or corresponding linear matrix inequalities. Moreover, using a dropping over-delayed packet strategy, algorithms to find the trade-off between the process noise with admissible intensity and the estimation performance in terms of a given restriction of estimation accuracy are proposed. The proposed method is tested on the battery state-of-charge estimation problem.