Estimation of reactivity and delayed neutron precursors' concentrations using a multiscale extended Kalman filter

A wavelet based multiscale extended Kalman filtering technique for estimation of reactivity and delayed neutron precursors' concentrations is presented in this paper. Reactivity which indicates the criticality status of the reactor core can only be measured in indirect way. Similarly delayed neutron precursors' concentrations, the source of the delayed neutrons which play important role in reactor control cannot be measured directly. Nuclear reactor is an example of multirate nonlinear system in which different state variables evolve with widely varying dynamics. The state estimation algorithm presented here is based on and preserves merits of Extended Kalman Filtering (EKF) technique. In addition, use of wavelet filters enables multiscale decomposition of the state variables that in turn, effectively captures the multirate nature of the system. Estimation has been carried out using reactor power as the only input. In order to justify effectiveness of the proposed method, simulation results are shown for completely known power variation dataset and experimental power variation datasets collected from one of the Indian research reactors.