Based on automatic identification system（AIS）big data，this paper analyzes the AIS original message，separates the keywords，and matches other ship data sources after data cleaning and filtering to further identify the ship type and divides by tonnage. Applying big data to the prediction of ship flow，this paper tracks the ship’s trajectory，summarizes the ship’s navigation law and operation，combines historical data with the development of the macro situation to comprehensively forecast the volume of waterway freight，predicts the ship flow in multi-branch channel by allocation model，and takes the south passage channel of the Yangtze Estuary as an example to predict the ship flow. The results show that AIS big data can be used to analyze the complicated and changeable ship routes，so as to accurately predict the ship flow in the multi-branch channel.