Ship trajectory prediction method based on machine learning and AIS data

Abstract

The invention discloses a ship trajectory prediction method based on machine learning and AIS data. The method comprises the following steps: preprocessing a historical AIS data set of a ship; dividing the pre-processed AIS data set according to tracks to obtain multiple pieces of trajectory feature data; detecting and deleting error data in the trajectory feature data; performing gridding processing on each piece of trajectory feature data, and dividing prediction targets of trajectory points in each piece of trajectory feature data into eight neighborhood grid directions; and establishing a trajectory prediction model based on an xgboost algorithm, and predicting the trajectory points through a trajectory prediction model. Therefore, the accuracy of ship motion direction prediction is improved. Meanwhile, when the data set is selected, the data can be considered to be valid as long as returned AIS data position information is required to be within a grid precision allowable range, so that the accuracy requirement on the AIS data is reduced.

Description

technical field [0001] The invention relates to the technical field of ship shipping safety, in particular to a ship track prediction method based on machine learning and AIS data. Background technique [0002] AIS is an automatic tracking system installed on ships, which can send characteristic information of ships under navigation in real time, such as speed, course, heading, etc. By analyzing the AIS data obtained in the past, combined with the current voyage trajectory of the ship, the voyage trajectory can be predicted, and it provides assistance for the monitoring and dispatching of the port ships. [0003] At present, the methods for predicting ship voyage trajectories based on AIS data mainly include two types: based on ship kinematics model and based on machine learning model. The former only considers the current trajectory of the ship, but does not consider the historical trajectory, ignoring the prior information of the current waters. , while the latter has hig...

AI Technical Summary

Problems solved by technology

[0003] At present, the methods for predicting ship voyage trajectories based on AIS data mainly include two types: based on ship kinematics model and based on machine learning model. The former only considers the current trajectory of the ship, but does not consider the historical trajectory, ignoring the prior information of the current waters. , while the latter has higher requirements on the trajectory data, which requires the trajectory data to be continuously and evenly distributed in time, but the original AIS data is often affected by the equipment and the marine environment, and various missing values ​​often appear. Interpolation correction is performed in the data to fill the non-existing sampling point data. The degree of deviation between the interpolation result and the true value has a great impact on the accuracy of the prediction.

Images