Estimation of continuous-time nonlinear systems by using the Unscented Kalman Filter

Abstract

This paper proposes a continuous-time model estimation method by using the Unscented Kalman Filter (UKF) from the sampled I/O data, in which the plant parameters as well as the initial state are estimated. The initial state is estimated based on a backward system of the plant, and the parameters are estimated by using the iterated UKF, which repeats the estimation of the forward system and the backward system alternately. In order to demonstrate the effectiveness of the proposed method, the rotary pendulum is provided to estimate the parameters of the continuous-time nonlinear system.