Implementation and Validation of Simultaneous State and Parameter Moving Horizon Estimation of a Pressurized Water Reactor

This paper presents the implementation of a Moving Horizon Estimation (MHE) approach for the simultaneous estimation of state and parameters within Pressurized Water Reactors (PWR) used in Nuclear Power Plants (NPPs). Addressing the inherent model stiffness, we leverage collocation as the integration method, making direct collocation a natural choice for transcription of the optimization problem into a nonlinear programming problem (NLP). The implementation benefits from state-of-the-art tools for modeling, expressing, and solving optimization problems, specifically CasADI and IPOPT. In a comparative analysis with a standard Extended Kalman Filter (EKF), our proposed MHE method exhibits superior performance and accuracy, even with rather classical tuning parameters.