Abstract: A Bayesian network-based model is proposed to predict the density of passengers in the baggage carousel of a single flight, combining with baggage claiming process analysis and machine learning of historical data to establish Bayesian network. Incremental learning characteristics of Bayesian network is used to realize dynamic adjustment of BN model, so that it can better adapt the new data and get more accurate prediction of passenger density. Taking extracted data from a large hub airport in China as instance, expectation maximization (EM) method is used to conduct model training. Results show that the proposed model can effectively predict the density of passengers in baggage claim area with high accuracy.

Key words: airport operation, passenger density, baggage claim, Bayesian network, dynamic prediction