Institute of Transportation, MOTC

        As the demand of the cruise industry increasing, there are the problem of crowding in cruise terminals. According to the feature of cruise passengers, the reason causes the bottleneck points is closely related to passenger arrival patterns, therefore, this study mainly discusses how to use real-time passenger arrival rates and improvement strategies to alleviate crowding in terminals and reduce passenger waiting times. In this study, a "reducing cruise station crowding model" is constructed, which includes a prediction module, a station waiting module, and a crowding reduction strategy module. In this study, a prediction module is used to predict the arrival rate of cruise passengers using a long short-term memory (LSTM) method; a station waiting module is established to calculate the station crowding degree and passenger waiting time; by adjusting the facilities in the terminal and the number of passengers arriving outside the station will achieve different effects of improving congestion.
        According to the classification data of multiple cruise ships and different cruise companies, the improvement results are evaluated by simulating two different data characteristics, and the number of people in the average station is reduced by 38.2% and 9.5%, and the average passenger waiting time can be reduced by 44.4% and 22.1%. In addition, the sensitivity analysis is used to adjust the congestion conditions and the proportion of affecting passengers, and analyze the improvement of the average number of passengers in the terminal and the average passenger waiting time before and after the model application. It can be seen from the empirical research of this study that different parameter settings have different magnitudes for improving the congestion problem, so the parameters and restrictions can be adjusted according to the station conditions and customs clearance conditions.