Control
Hossein Zahmatkesh; Hussein Eliasi
Abstract
State estimation of nuclear reactors often plays a crucial role in accomplishing load-following control. This study presents a novel approach that leverages a weighted particle filter to address the challenges associated with estimating these crucial parameters, including relative precursor concentration ...
Read More
State estimation of nuclear reactors often plays a crucial role in accomplishing load-following control. This study presents a novel approach that leverages a weighted particle filter to address the challenges associated with estimating these crucial parameters, including relative precursor concentration (C_r) and fuel temperature (T_f), under varying reactor power conditions. A high-fidelity nonlinear dynamic reactor model was developed, incorporating noises in both process and measurement models. The proposed method was evaluated by extensive simulations under a wide range of operational scenarios. The particle filter demonstrated exceptional performance in tracking the time-varying states of the nuclear reactor. Comparative analysis with a conventional Kalman filter and the extended Kalman filter revealed the superior robustness of the particle filter in handling nonlinearities inherent in nuclear systems. The proposed approach offers several advantages, including the ability to capture multimodal distributions, handle non-Gaussian noise, and provide probabilistic estimates. Despite the increased computational cost associated with particle filtering, the benefits in terms of estimation accuracy and reliability justify its application in nuclear power plant monitoring and control systems.
