R&D Study on on-line criticality surveillance system (V)

Yamada, Susumu*

In view of necessity and importance of criticality surveillance systems for ensuring the safety of nuclear fuel manufacturing and reprocessing plants, 5-year basic studies and 4-year R&D studies on an on-line criticality surveillance system were carried out since 1991. This report is a summary of these series of studies. Noticing that the signal from a neutron detector is random in principle, these series of studies aimed to accumulate knowledge for developing an inexpensive criticality surveillance system with quick response based on the Auto-Regressive Moving Average (ARMA) model identification algorithm. During five-year basic studies on criticality surveillance system since 1991, we obtained knowledge required for developing a criticality surveillance system based on the ARMA model identification algorithm through (1)studies on recursive ARMA model identification algorithms most appropriate for estimating subcriticality form time series data under a steady state condition, (2)studies on pre-processing of signal from neutron detectors, (3) developing a new recursive ARMA model identification algolithm with small time delay to estimate time-dependent subcriticality, (4) proposing a basic concept for the elements required for an on-line criticality surveillance system, and (5) numerical analysis of data from the DCA experiments. During next four-year R&D studies on a criticality surveillance system since 1996, we (1) proposed modules required for a no-line criticality surveillance system, (2) revealed effectiveness of a adaptive digital filter (ADF) algorithm, as an important redundancy to the recursive ARMA model identification algorithm to be used in the signal processing module through numerical analysis of real data, (3) proposed a module of the Feynman- method over ray signal and a fast signal processing module for ray signal, (4)devdoped a line-noise removal filter(Notch filter) and revealed its effectiveness for the DCA ...