Instability localization with artificial neural networks (ANNs)

The aim of this piece of research is to investigate the potential of artifi cial neural networks (ANNs) for tackling the problem of instability localiz ation. The instability is modeled by a variable strength absorber (point-so urce) in a two-dimensional bare reactor model with a one neutron-energy gro up. The proposed approach constitutes an exercise in simplicity in that: (1 ) an arbitrarily simplified model is employed for ANN training and validati on; (2) few training and validation patterns of low complexity are utilized ; (3) the ANN inputs are derived directly from the neutron noise signals, t he proposed location of instability is given on-line via an uncomplicated c ombination of ANN outputs; (4) the ANN architecture is independent of the n umber of possible locations of instability. In fact, unlike previous approa ches which employ hundreds of outputs (one for each fuel assembly), only tw o ANN outputs are employed representing the X- and Y-coordinates (location) of instability; (5) the responses of only a. few detectors are employed; ( 6) a measure of confidence in the prediction is assigned. The results of AN N testing, which is performed on patterns from both actual arid simplified models, are reported and analyzed. (C) 2001 Elsevier Science Ltd. All right s reserved.