Research On The Method Of Ship Tracking Prediction

With the rapid development of the maritime industry,the situation of maritime traffic is becoming more and more complicated,and the contradiction between the limited resources of the ocean and the increasing traffic flow has become increasingly prominent,and the burden on sea areas has become heavier.Under such circumstances,the problems existing in ships and human-induced maritime traffic accidents have gradually increased,causing huge economic losses and casualties.Therefore,relevant maritime departments and ship traffic service systems must improve decision-making standards and decision-making efficiency,and the core technology is ship tracking and prediction technology.This article takes "Study on ship track tracking prediction methods" as the topic,analyzes and researches common related algorithms,summarizes the advantages and disadvantages of each algorithm,and proposes new tracking prediction methods,with the aim of improving the accuracy and prediction of ship track tracking and prediction.Timeliness.The main work includes:1.Analyze the problems of the related algorithms of target tracking and prediction in practical engineering applications.2.A ship tracking method based on track credibility is proposed.This method quantifies the quality of point aggregation based on the coupling degree of the original point aggregation.Then,referring to the "forget gate" and "cell state" ideas of LSTM(Long Short-Term Memory)model,the concept of "temporary track growth value" is put forward,and the credibility of the temporary track is measured by it.At the same time,multiple wave gates are used to filter the track,eliminate false tracks,reduce the false alarm rate,and achieve the correct start of the track.3.A ship track prediction method is proposed.The data of the ship’s trajectory is a series with long-term characteristics,and it is susceptible to wind waves,clutter and noise,which makes its characteristic data change dynamically,and it is difficult to grasp the law of the trajectory.Based on the above problems,this paper uses the self-attention mechanism of the Transformer model for track prediction.At the same time,the Kalman filter is used to dynamically modify the output result to obtain the best track prediction value.4.Simulation.Aiming at the tracking algorithm,the experimental comparison in this paper shows that the proposed algorithm has better tracking effect than the traditional processing method,which verifies its feasibility and effectiveness.As for the prediction algorithm,this paper uses AIS(Automatic Identification System)data of Ningbo fishing boat for model training,and predicts the ship’s track.Experimental results show that the algorithm has higher prediction accuracy,and it is also superior in processing serial data in parallel.