Human Factors Association Mining for Controlled Flight into Terrain Based on QAR Data.

Background: The International Civil Aviation Organization (ICAO) has identified Controlled Flight Into Terrain (CFIT), Runway Excursion (RE), Loss of Control In-flight (LOC-I), Mid-Air Collision (MAC), and Runway Incursion (RI) as the five core risks in the 2020–2022 Global Aviation Safety Plan. Airbus mentioned in the 2022 Commercial Aviation Accident Statistical Analysis Report that the top three causes of fatal accidents in the past 20 years were Loss of Control In-flight (LOC-I), Controlled Flight Into Terrain (CFIT), and Runway Excursion (RE), with CFIT accounting for 17% of the total, which is the second most fatal type of accident. Therefore, it is crucial to study the causation of CFIT accidents to reduce the fatal accident rate of airlines and improve the safety management level. However, the existing accident causation analysis methods are highly generalized and lack human factor causation analysis for CFIT accidents; moreover, the existing human factor causation analysis indexes are mostly descriptive indexes and lack data indexes. Purpose: The purpose of this study is to summarize the human factor causation pattern of CFIT accidents by analyzing historical CFIT accident investigation reports; combine with the flight data monitoring indicators, further refine the analysis level on the basis of the existing causation analysis methods, and put forward an optimized human factor analysis method based on the flight data of CFIT, so as to provide assistance for the safe and efficient operation of the safety company. Methods: First, a total of 73 CFIT accidents from 2008 to 2021 were collected and counted, and the investigation reports of the 73 CFIT accidents were analyzed in two dimensions, horizontal and vertical, using the SHELL model and the HFACS model. The horizontal dimension refers to the human-aircraft-environmental-management causal factors based on the SHELL model, and the vertical dimension refers to the human factors causal factors based on the HFACS model. Second, based on CFIT accidents investigation report, a combination of literature research and correlation analysis was used to correlate the indicators of the HFACS model with the Quick Access Recorder (QAR) data monitoring program and to refine the analysis level of the HFACS model. In particular, the levels of unsafe behaviors in the HFACS model are further refined on the basis of errors and violations to establish the correlation between human factors and QAR data monitoring items. Finally, based on the above findings, an optimized human factors analysis method for Controlled Flight Into Terrain (CFIT) based on flight data is proposed. Results: 1) The optimized HFACS model can refine the analysis of crew error violations in CFIT accidents and provide specific behavioral level qualitative analysis for CFIT accidents; 2) The correlation between crew error violations and QAR data monitoring items can mine out the behaviors that may lead to the occurrence of CFIT accidents and the QAR data monitoring items that need to be paid attention to, and provide specific behavioral level qualitative analysis for CFIT accidents. The correlation between crew errors and violations and QAR data monitoring items can be used to identify the behaviors that may lead to CFIT accidents and the QAR data monitoring items that need to be focused on, so as to provide a specific behavioral level quantitative analysis for CFIT accidents and provide support for the airlines to formulate flight training plans.