Weight value ellipsoid based estimation algorithm with finite data window

This paper proposes an outer bounding ellipsoid estimation algorithm for a system with unknown but bounded noise. The finite data window idea is adopted to form the rolling data matrix by the limited data. The iteration on the ellipsoid can tighten up the outside of the parameter estimates that makes it converge to the real value. By the end of the metabolic step, the minimized ellipsoid is obtained finally and the center of the ellipsoid is regarded as the parameter estimates. The simulation results show the better accuracy and effectiveness compared with other similar estimation algorithm, as well as illustrates its performance on different finite data lengths.