The objective of this methodology is to create a model capable of analysing and predicting the identification of potential conflicts between aircraft during en-route flights by air traffic control operators (ATCOs). This aspect is crucial to prevent Separation Minima Infractions (SMIs) and ensure safety. Initially, a Bayesian Network Model (BNM) is developed to define the safety event and its key dimensions. Using descriptive analysis and statistical knowledge, appropriate models are selected to represent the frequency of safety events. In addition, a data-driven approach is used to enhance model development. The BNM is calibrated, adjusted and subjected to sensitivity analysis using real data. The explanatory power and mixed effects of models and independent variables are assessed. By considering mixed effects in the sensitivity analysis, safety performance is characterised not only in terms of individual dimensions, but also in terms of their combinations. This allows the identification of threshold values that would minimise the occurrence of safety incidents.
Read full abstract