Abstract: In order to accurately perceive the flight departure process and estimate the flight departure time,a method for estimating flight departure time based on dynamic Bayesian network was designed.Firstly,the factors affecting the flight departure process are analyzed based on different flight attributes.According to the influencing factors,the data are classified and processed,and the Monte Carlo simulation method is combined with the historical data classification to obtain the joint and prior distribution of each link.The Kolmogorov test is used to determine the joint distribution model of each link,and the parameters of the dynamic Bayesian network modelare obtained.Secondly,the Bayesian network architecture and conditional probability are used to infer the dynamic estimation of the departure time and the completion time of each link.Finally,the single-destination operation data of an airport in the middle of the country are selected for simulation verification.The research results show that with the progress of the process,the propagation error will gradually increase,and the estimated accuracy of the departure time will be over 80%,what’s more,the stability of the dynamic estimation result will be better,which can fully reflect the actual situation of the key node in the departure process of the flights.

Key words: Air transportation, Conditional pro-bability estimation, Dynamic Bayesian network, Flight departure process, Kolmogorov test, Propagated error