A human error mechanism for pilot based on fault tree analysis and Bayesian network

Abstract

At present, aircraft accidents caused by human factors occur frequently. As the flight crew is the last line of defense for aviation safety, how to identify and quantitatively evaluate human errors in flight and reduce the error rate has become an important issue for the civil aviation, and it is also the most effective way to control flight operation risks. Thus, the human error mechanism for pilot was investigated by the combine of the fault tree and the Bayesian network theories based on historical research data of typical unsafe events caused by flight human errors. Firstly, the fault tree was used to identify and qualitatively analyze the risk of systems, and then transformed into a Bayesian network model to obtain the relative probability of intermediate events and top events. Finally, the system hierarchy of unsafe events caused by flight human error was quantitatively evaluated. The results showed that there were 96 failure modes in the system, and the flight human error was caused by the coupling of multiple risk factors. The probability of non-technical skill loss is the highest, followed by that of the lack of technical skills and violations. The basic events in the organization, environment and equipment factors have a great impact on the flight human error, which is a weak link in the system. The results provide some theoretical basis for developing preventive measures of flight human error and improving the level of flight safety.