Reducing computational load in Moving-Horizon Observers using partial explicit map inversion

This paper proposes a novel approach that may help reducing the computational burden when using the Nonlinear Moving-Horizon Observer/ Estimator (MHE) technique. This technique is often used to reconstruct unmeasured quantities of a dynamic system. The latter may include both the state and the vector of parameters that are involved in the system model. The proposed approach is based on the use of partial explicit inversion maps that express a part of the problem unknowns as a function of the remaining ones. The Moving-Horizon Estimators can therefore concentrate on the latter reduced dimensional unknown vector. The paper shows how explicit inversion maps can be derived based on a recently developed graphical-signature-based technique. Two illustrative examples are proposed to show the efficiency of the proposed solution.