Set-membership State Estimation for a Continuous Stirred-Tank Reactor

In this paper, the robust state estimation problem for a continuous stirred-tank reactor is addressed. Two different robust estimators are studied for the set-membership state estimation. Here, a strip-based observer is compared to a set-propagation observer. In the first case, we consider the classic recursive optimal parallelotopic outbounding algorithm. In the second case, we use a Zonotopic Kalman filter (ZKF). The comparison takes into consideration the CPU time and the estimation accuracy (in terms of the final volume of the estimation set). Our findings indicate that the differences in CPU time are insignificant, yet the strip-based method showed a major reduction in the final reachable set.