Development and quantitative effect estimation of an integrated decision support system to aid operator's cognitive activities for NPP advanced main control rooms

As digital and computer technologies have grown, human-machine interfaces (HMIs) have evolved. In safety critical systems, especially in nuclear power plants (NPPs), HMIs are important for reducing operational costs, for reducing the number of necessary operators, and for reducing the probability of accident occurrence. Efforts have been made to improve main control room (MCR) interface design and to develop automation or support systems to ensure convenient operation and maintenance. In this paper, an integrated decision support system to aid the cognitive activities of operators is proposed for advanced MCRs in future NPPs. The proposed system supports not merely a particular task, but also the entire operation process based on a human cognitive process model. It supports the operator's entire cognitive process by integrating decision support systems that support each cognitive activity. In this paper, the operator's operation processes are analyzed based on a human cognitive process model and appropriate support systems that support each activity of the human cognitive process are suggested. Two decision support systems were developed in this paper. The first one is the fault diagnosis advisory system (FDAS) which detects faults and diagnoses them. The FDAS provides a list of possible faults and expected causes to operators. It was implemented using two kinds of neural networks for more reliable diagnosis results. The second system is the multifunctional operator support system for operation guidance, which includes the FDAS and the operation guidance system. The operation guidance system is to prevent operator's commission errors and omission errors. Furthermore, the effect of the proposed system was estimated because to evaluate decision support systems in order to validate their efficiency is as important as to design highly reliable decision support systems. The effect estimations were performed theoretically and experimentally. The Bayesian belief network model was used in the theoretical effect estimation, and the model was constructed based on human reliability analysis event trees. For the experimental effect estimation, a prototype of the proposed system was implemented. The prototype has simplified functions for an alarm information system, a fault diagnosis system, a computerized procedure system, and an operation validation system. In the results of both theoretical and experimental evaluations, the proposed system showed better performance compared to independent decision support systems.