Development of a prototype knowledge-based system for estimating human error probabilities~1

摘要

Human reliability analysis (HRA) is characterized as the prediction and evaluation of work-oriented human performance in quantitative terms, using indices like error likelihood or probability of task accomplishment. Today, the HRA practitioner is confronted with a bewildering set of approaches, where a poor selection or improper application of methodology can produce invalid HEP estimates leading to severe consequences relatd to accidents causing property damage, injury, or death. To address these situations, Los Alamos National Laboratory is developing a human reliability centered, expert system to help with the quantification of human erro actions. This system will improve HRA validity by disseminating problem solving knowledge to minimally trained end users, by standardizing conclusions for a given set of data, and by freeing the human expert from various repetitive routine HRA quantification jobs.