Contraction of Information and Its Inverse Problem in Reactor System Identification and Stochastic Diagnosis

Abstract

Research concerning power reactor noise analysis makes rapid progress in the areas of the system identification, prediction and diagnosis. Keywords in these studies are artificial intelligence, neural network, fuzzy, and chaos. Nonlinear, nonstationary, or non-Gaussian processes as well as linear and steady processes are also studied in fluctuation analysis. However, we have not enough time to study a fundamental theory, since we are urged to obtain results or applications in power reactor fluctuations. Furthermore, we have no systematic approach to handle observed time series data in the linear process, since power reactor noise phenomena are complicated. Hence, it is important to study it from the fundamental viewpoint. It is a main aim of the present review paper to describe a unified formalism for reactor system identification and stochastic diagnosis.