Stability of a nonlinear Moving Horizon Estimator with pre-estimation

In this paper, a Moving Horizon Estimator with pre-estimation (MHE-PE) is proposed for discrete-time nonlinear systems under bounded noise. While the classical Moving Horizon Estimator (MHE) compensates for model errors by estimating the process noise sequence over the horizon via optimization, the MHE-PE does it using an auxiliary estimator. The MHE-PE is shown to require significantly less computation time compared to the MHE, while providing the same order of magnitude of estimation errors. The stability of the estimation errors of the MHE-PE is also proven and an upper bound on its estimation errors is derived. Performances of the MHE-PE is illustrated via a simulation example of pressure estimation in a gas-phase reversible reaction.