Predictive Mathematical Modeling for Ex-core Neutron Detectors Using A Neural Network

A methodology to calculate the reactor axial power distribution using ex-core detector signals with an algorithm based on a neural network is presented. The fundamental basis of the algorithm is that the detector response can be fairly accurately estimated using numerically generated parameters. In other words, the training set for the neural network representing the relationship between detector signals and axial power distributions can be obtained through calculations instead of measurements. Application of the developed method to Yonggwang nuclear power plant unit 3 (YGN-3) shows that it is superior to the current algorithm in place. (author)