The purpose of the present paper is to establish a new moderator temperature coefficient (MTC) estimation method based on gray box modeling concept. The gray box model consists of a point kinetics model as the first principle model and a fitting model of moderator temperature kinetics. Applying Kalman filter and maximum likelihood estimation algorithms to the gray box model, MTC can be estimated. The verification test is done by Monte Carlo simulation, and, it is shown that the present method gives the best estimation results comparing with the conventional methods from the viewpoints of non-biased and smallest scattering estimation performance. Furthermore, the method is verified via real plant data analysis. The reason of good performance of the present method is explained by proper definition of likelihood function based on explicit expression of observation and system noise in the gray box model.
