由于机场航班之间存在前后衔接关系,每个航班的延误会波及到下游机场及其航班,因此需要一种有效手段来分析航班衔接时的延误传播。贝叶斯网络(BN)是一种有效的传播分析方法,从某个枢纽机场航班延误出发,对其关联机场的衔接航班的延误影响进行分析,提出了基于BN的航班延误传播模型。结合某航空公司实际数据,通过最大期望值算法对模型进行训练,给出了测试结果。实验表明,所提出的方法能有效地分析航班延误从局部到全局的传播。 Due to the convergence of airport flights, the delay of each flight will affect the downstream airports and their flights. Therefore, an effective means is needed to analyze the delay propagation in flight convergence. Bayesian Network (BN) is an effective propagation analysis method. Starting from the delay of a hub airport flight, the delay impact of the linked flight at its associated airport is analyzed. A BN-based flight delay propagation model is proposed. Combined with the actual data of an airline, the model is trained by the maximum expected value algorithm, and the test results are given. Experiments show that the proposed method can effectively analyze flight delays from local to global.