Research On The Feature Model Of Abnormal Ship Track Based On Neural Network

In the supervision of maritime traffic accidents,intelligent traffic management is one of the important means of maritime safety prevention and control.Among them,the research on abnormal behavior of ships is an important part of the scientific theoretical research on maritime safety,which is used in the supervision of illegal activities such as frequent traffic accidents and violations at sea.According to the ship’s trajectory,it analyzes the ship’s navigation status and the normativeness of its trajectory,studies the characteristics of the ship’s behavior,and then identifies and judges the abnormal behavior of various ships with specific purposes to ensure the safety and smooth movement of the sea.This article focuses on the research of ship behavior based on the correlation between trajectory data.This research is of great significance to ship supervision and safe passage at sea.Through data mining,this thesis analyzes the relationship between the data in the track information,and creates a feature database to realize the identification of abnormal ship behavior.The abnormality of ship behavior is mainly judged from two aspects,the rationality of ship motion law,and the matching of ship track data association relationship.Four models were created for different aspects of the anomaly,based on the analysis of the overall trend of the actual sailing trajectory,the ship trajectory smoothness measurement algorithm was proposed,and the Radial Basis Function(RBF)was used to smoothly fit the trajectory,create a smoothness judgment model to realize the judgment of abnormal smoothness of the trajectory trend;By considering the impact of the actual offshore environment,a wind flow pressure difference judgment algorithm is proposed,which realizes the identification of the abnormal wind flow pressure difference information contained in the trajectory data,and constructs the wind flow pressure difference analysis model;On the basis of comprehensively considering the overall trend of the actual sailing trajectory,by analyzing the relationship between the ship’s track direction and the data direction,a dual feature factor association matching algorithm is proposed,and a dual feature factor abnormality judgment model is created by using a multi-layer back Propagation neural network(BP)to realize the recognition of abnormal association matching degree;Combined with the ship navigation path prediction method,the overall relevance analysis of the trajectory information is carried out,by using the mid-latitude algorithm,a multi-feature factor path matching algorithm is proposed.Based on the idealized predicted trajectory,the BP neural network is used to create a multi-feature factor abnormality judgment,the model realizes the recognition of the abnormal matching degree of the actual trajectory data.Based on the four models created above,a ship abnormal trajectory judgment system was designed and implemented,which is used to comprehensively judge ship abnormal behaviors for specific purposes.The normal trajectory database and the abnormal trajectory database were created by statistical analysis method,and the distribution of abnormal types and various abnormal degrees of the two trajectories in the case of big data were counted,which was used to analyze and verify the judgment results of the model and the system.The results show that the abnormal ship trajectory feature model based on neural network can effectively judge the abnormal behavior of the trajectory,and the abnormality judgment system built on this basis can provide relevant monitoring basis for multiple abnormal situations at sea,improve maritime supervision ability,and guarantee The safety of maritime navigation and the security of maritime defense.