Topic identification of air traffic control unsafe events based on structural topic model
In order to explore the occurrence regularity and characteristics of unsafe events in civil aviation control operation,the structural topic model ( STM) was employed to capture the semantic lexical co-occurrences and hierarchical structures in the reports.The optimal number of topics for operation events control was identified by considering the indexes of structural topics such as semantic coherence and exclusivity.The severity,happening date and other covariates of events were selected to construct the multivariate topic analysis,and the severity difference of different types of events and their evolution trend in the time dimension were mined.The empirical analysis was conducted based on 4,346 control event reports from the Aviation Safety Voluntary Reporting System (ASRS) from 2018 to 2022.Unsafe events related to air traffic control operations can be categorized into 15 types:air traffic control handover,adverse weather conditions,human factors incidents,night flights,ground taxiing,takeoff and departure,interagency coordination,radar separation,equipment failure,approach procedures,radio teleph-ony,complex airspace environments,flight conflicts,flight altitude,runway overrun/excursion,and runway incursions.The events such as flight area taxiing,approach procedure,and flight conflict are more likely to lead to severe accidents,and the unsafe events related to flight altitude present an increasing trend annually.The research results can provide theoretical sup-port for enhancing the level of safety risk in control operation,and offer reference for management personnel to implement the targeted measures.
air traffic managementair traffic control operationsafety levelunsafe eventstructural topic model
万方数据
维普
