Analysis of Construction Safety Education Format Impact on Accident Occurrence Using Agent-Based Modeling

The construction industry has the highest occupational fatality rate, with over 75% of accidents attributed to workers' unsafe behaviors. Traditional top-down safety management approaches have limitations in reflecting the complex and dynamic characteristics of construction sites. This study developed an agent-based model based on construction workers' cognitive processes to analyze the effects of safety education cycles and scales. The model incorporates individual factors (risk attitude) and social-organizational factors (group norms and management norms) that influence workers' risk acceptance and behavioral decisions. Through simulation of 100 agents over 100 time units, various education scenarios were analyzed. Results showed that shortening education cycles from 20 days to 1 day reduced accident occurrence by 73% (from 3.18 to 0.86 incidents). An effective threshold was identified at education cycles of 2 days or shorter, with significant effects when targeting 50% or more of workers. This research provides theoretical foundations for safety education approaches through agent-based modeling and offers insights for improving current quarterly group education methods.