Abstract:

Taking the investigation reports of foreign unsafe events in ATM in the past 20 years as data samples, the risk factors leading to human errors of controllers were studied. Based on the method of cognitive error backtracking analysis (TRACEr), the types and root causes of controller errors were divided into four cognitive levels: perception, memory, planning decision and response execution. The data mining method based on RST and the reverse reasoning method of BN were used to calculate the error risk representation value of each cognitive field. The results show that vigilance failure, information processing fault, environmental interference and unclear information transmission were the high-risk factors of unsafe events. The root causes of visual perception errors such as boundary vision limitation, visual recognition error, expectation tendency and unglobalized attention are the main influencing factors in the ground unsafe events. The root causes of memory impairment, insufficient learning, risk identification failure, and negative effects, account for the main influence in air insecurity incidents. The visual perception error and information transmission error obtained by RST are the main human error types. In the conclusion of BN analysis, judgment error is the most important human error next to visual perception error.

Key words: unsafe event in air traffic management(ATM), human error risk, rough set theory(RST), Bayesian network(BN), technique for retrospective analysis of cognitive errors(TRACEr)