Use of neural networks for in-core fuel management

Artificial neural networks (ANNs) are an adaptive nonalgorithmic approach to information processing that provides an almost instantaneous capability for finding solutions to very complex problems. Although work in this area was deemphasized more than a decade ago following several critical reviews, there has been a resurgence of ANN-related research with recent developments in the underlying methodology and in the applicable technology. The Defense Advanced Research Projects Agency (DARPA), as well as other organizations, is sponsoring an extensive research and development effort in this area. Opportunities for applications of ANNs span essentially all areas of commerce, and the area of nuclear science and technology is well suited for the use of neural networks. The University of Tennessee, Department of Nuclear Engineering, is conducting ANN-related research in signal validation, spectral characterization, plant modeling, plant diagnostics, load projection, and in-core fuel management. This paper presents a brief overview of various neural networks, and it includes results associated with their application to in-core fuel management.