A novel approach for state estimation

A state estimator design, based on an identified linear model of a quasi-linear system, is presented for use in the design of a feedback controller. Without a priori information of the plant dynamics, the present method estimates the states to the actual shape of the true state trajectories using only measured input/output data. This novel state estimation method is developed from a selected best pseudo-observable form of multivariable system to obtain an estimation of a particular set of state variables for the identified system model. The results presented here provide a more appropriate discrete-time state estimator for feedback control in comparison with a Luenberger type state estimator.