ABSTRACT
A computational modeling approach to human performance assessment in nuclear power plants is proposed. The approach takes into consideration three main processes involved in a human’s cognitive process: information perception, reasoning and response. Both the saliency and activation levels of a sensor signal in the control room are used to determine whether the signal can be actually perceived by the operator. The corresponding nodes in the operator’s knowledge base are activated by the perceived signals and the activation is propagated in the knowledge base to reach diagnoses and develop responses. Uncertainties in these processes are dealt with using sampling methods. The interplay between an operator’s stress and fatigue levels and the cognitive process are considered. A case study derived from the Three Mile Island accident is conducted to demonstrate the proposed approach.
