Interval Estimation for Bounded Jacobian Nonlinear Systems by Zonotope Analysis

Abstract: This paper introduces an interval state estimation method for discrete-time bounded Jacobian nonlinear systems allying Luenberger-like observer with zonotope set computation. First, a robust observer is designed to obtain bounded-error and point-estimation with a peak-to-peak performance. This allows one to cope with the unknown but bounded process disturbance and measurement noise. Then, based on the stable dynamics of the observation error, tight interval estimation is obtained by applying zonotope set computation and analysis. To sum up, a comprehensive interval estimation algorithm is proposed by integrating the robust (in the sense of peak-to-peak performance index) point-valued estimate with the feasible zonotope set of the estimation error. Numerical simulation tests are conducted to assess the effectiveness of the proposed approach.