Kalman filter, particle filter, and extended state observer for linear state estimation under perturbation (or noise) of MHTGR

Abstract

States estimation of the nuclear reactor often plays a critical role in accomplishing load-following control and operation monitoring. This paper investigates the states (relative density of delayed neutron precursor (Cr), average temperature of reactor core (TR), and average temperature of helium in the primary loop (TH)) estimation of the modular high-temperature gas-cooled reactor (MHTGR) in the absence and presence of noise (continuous, random Gaussian white, random non-Gaussian colored). A real-time comparison of the Kalman filter (KF), particle filter (PF), and linear extended state observer (ESO) is performed under the load-following operation of the MHTGR. To make the comparison reasonable, the estimation performance comparison of the KF, the PF, and the linear ESO under the same conditions is realized, and four different simulation cases are taken into account to compare their estimation performance. Finally, numerical simulation results show that the KF provides better estimation performance for states Cr, TR and TH in comparison with the PF and the linear ESO.