Human Error Analysis of Nuclear Power Plant Operational Events Based on the Modified HFACS-BN Model

Human error is a critical factor affecting the operational safety of nuclear power plants. This paper proposes a human factors modeling approach that integrates the Human Factors Analysis and Classification System (HFACS) with Bayesian Networks (BN), based on 190 historical operational event reports from Chinese nuclear power plants. A modified four-level HFACS framework is used to identify human and organizational factors, and their multi-level relationships are mapped into the structure of a Bayesian Network. The model’s rationality was verified through the axioms of probability monotonicity and interval smoothness. The study identified key high-sensitivity factors with unsafe acts as the target node and quantified their influence strength using the Hellinger distance, thereby revealing the main evolution paths of human errors. This research provides a quantitative analytical method for nuclear power plant safety management, helping to identify the propagation relationships and critical influence chains of human errors within the system.