Association Rule Analysis of UAV Accident Causation Based on Text Mining

Given the increasing prevalence of UAV usage and the concomitant rise in accident rates, it is imperative to conduct a comprehensive investigation into the underlying causes of these incidents. This study aims to identify the principal causal factors and elucidate the association rules that govern UAV accidents. A total of 144 UAV accident investigation reports were analyzed using text mining techniques and the TF-IDF algorithm, resulting in the identification of 23 significant causative factors. Subsequently, the Apriori algorithm was employed to derive 25 significant association rules, which were then visualized and analyzed through a multidimensional network constructed using Gephi software. The findings indicate that UAV accidents are predominantly attributed to inherent system issues (e.g., datalink/power failures) and human factors (e.g., operator error, insufficient inspection). Moreover, the interplay among system failures, human factors, and environmental influences underscores their combined impact in precipitating UAV accidents. Overall, this study furnishes a robust theoretical framework and practical guidelines for UAV accident prevention.