Monitoring Defects in Attention Allocations of Nuclear Power Plant Operators through Operational Knowledge Representation

During nuclear power plant (NPP) operations, operators selectively focus their attention on alarms, indicators, and control switches. Failures in allocating attention at the right time on the right parts of the workspace have been observed to cause 35% of situation assessment errors. Existing methods are inefficient in preventing such failures due to a lack of comprehensive attention allocation processes in various NPP operations. Such characterization is critical for the real-Time identification of gaps in human attention for monitoring operational hazards. This paper presents formal knowledge representations for characterizing visual attention allocation processes of NPP operators. The proposed knowledge representations systematically mapped out attention drivers such as task structures, objects, and action types in operating procedures. Such a formalized representation of NPP operation knowledge can support automatic reasoning methods to identify critical objects and safety-related properties during reactor startup processes. Capturing this operational knowledge and understanding operators' visual attention allocation principles would help establish a human-centric computing framework for ensuring the safety and efficiency of NPP operations.