Neural network application for radionuclide modelling and prediction of radioactivity levels
Existing applications of artificial neural networks in physics research and development have been analyzed as a basis for proposing new opportunities using that AI technology for data analysis in physics. A taxonomy was developed, based on an extensive literature search, for physics problems where neural network applications have been useful. Then, a particular use of neural networks was carried out to study ways to predict normal concentrations of radioactivity measured at monitoring stations in different geographic locations. The purpose of the data collection and analysis was to establish background levels that would serve as bases for detecting unusual levels of radioactivity, for example due to nuclear weapons testing, in these physical environments. Useful data sets were developed in this area and a process was discovered for modeling the background levels.