State estimation of external neutron source driven sub-critical core using adaptive Kalman filter

•EKF and CKF are proposed to estimate the state parameters of sub-critical reactor core.•The NRE and ERV adaptive algorithms are proposed to adjust the covariance matrix online.•For estimation of hidden variables, CKF has divergence and EKF achieves good results. Extended Kalman filter (EKF) and cubature Kalman filter (CKF) are proposed to estimate the state parameters of an external neutron source driven sub-critical reactor, including power level, reactivity, external neutron source, six-groups of delayed neutrons precursor densities, equivalent fuel temperature, average coolant temperature and nuclear densities of iodine, xenon, promethium, samarium nuclides. Parameter settings and matters needed attention in EKF and CKF are also analyzed, especially the relationship between model prediction covariance matrix and measurement covariance matrix. In order to effectively identify the maneuvering of the external neutron source and reactivity on the uncertainty of the prediction model, two adaptive algorithms are proposed to adjust the covariance matrix of the prediction model online. The results show that these two adaptive algorithms can effectively detect various maneuvering such as neutron source variation and reactivity insertion, and realize the optimal estimation of the reactor state using EKF method. However, CKF has divergence and non-convergence. EKF achieves good results in all parameters estimation.