The work aims to improve the mining accuracy and safety efficiency of cargo logistics path trajectory information. In this study, the type, quantity and length of ships in the narrow channel of the strip Tiaozhoumen waters of Ningbo Zhoushan Port were statistically analyzed for the identification and repair of AIS logistics path trajectory abnormal points of cargo ships. Considering the environmental factors of actual ship navigation, a new date correction method was proposed. Considering the ship attributes and environmental factors, the abnormal data were identified with the navigation span and steering capability of the three-degree-of-freedom kinematics, and finally the Bi-LSTM method was applied to repair the screened shipping logistics path trajectory. The proposed screening method did not require trajectory clustering or building additional models for discrimination, the screened data accounted for 34.26% of the total quantity, and the repaired AIS cargo ship trajectory data were improved by 115.34% on the original basis. The method can effectively correct and repair abnormal shipping logistics path trajectory data, and provide some basic methods for shipping logistics trajectory data mining.