Utilization of expert systems in severe accident management

Human factors have played a central role in all major nuclear accidents. The most recent, Fukushima accident, highlighted the need to reanalyse current human factor engineering concepts. A number of actions were initiated by the international community to reduce the contribution of human factors to nuclear accidents. An extensive effort is directed towards design considerations of HMI, training, human reliability analysis and other organizational aspects. However, less attention is being paid to increasing human reliability and performance through advanced expert systems for accident management. Management of severe accidents is a complex process where stress and lack of information significantly affect performance of individuals and teams. The paper presents two novel concepts for advanced computerized expert systems for severe accident management support in real accident conditions. The first is focused on the support for evaluation of plant status and subsequent decision making based on advanced machine learning algorithms. The expert guidance is being developed based on current plant data measured during an accident and an extensive set of pre - calculated analytical simulations. The second approach, which is based on applied graph theory and hydraulics, provides technical support centre staff with practical guidance for optimal equipment line-up for the implementation of the selected strategy.