Long-time-sequence prediction method for ship trajectory

Abstract

The invention provides a long-time-sequence prediction method for a ship trajectory. The method comprises the following steps: firstly, preprocessing track data; then, performing track clustering on the track data by using a K-means algorithm to obtain a ship motion behavior mode set; secondly, training and testing samples are constructed based on ship motion behavior pattern similarity matching, an improved pix2pix network model is constructed, and iterative training is carried out; and finally, inputting a ship trajectory needing to be predicted and matched trajectory information thereof into the trained network to obtain a ship trajectory long-time-sequence prediction result. The method can effectively solve the problems of error accumulation, low efficiency and the like in long-time-sequence prediction of the ship trajectory, and effectively realizes long-time-sequence, high-precision and high-efficiency prediction of the ship trajectory.

Description

Technical field [0001] The present invention belongs to the field of intelligent prediction technology of ship trajectory, in particular to a long time series prediction method of ship trajectory. Background [0002] Traditional offshore ship trajectory prediction methods, such as ARMA, Kalman filtering, BP neural network, etc., can complete the accurate prediction task of ship trajectory in a short period of time, but when the prediction time is increased, the traditional method often leads to poor prediction accuracy due to problems such as error accumulation. With the continuous development of artificial intelligence, deep learning technology has made a huge contribution to related prediction problems in various fields. At the same time, with the rapid development of the automatic identification system (AIS), satellite radar, electronic reconnaissance system and other technologies, the trajectory data of maritime moving targets has been complete enough to provide sufficient da...

AI Technical Summary

Problems solved by technology

[0003] In recent years, due to the increasingly complex sea traffic environment, the demand for ship trajectory prediction at sea tends to be diversified. Traditional ship trajectory prediction methods are difficult to meet the actual application requirements in terms of accuracy, long-term sequence, and efficiency.

Although deep learning prediction methods such as CNN-based ship trajectory prediction model and LSTM-based ship trajectory prediction model have been widely used in recent years, and their prediction efficiency and accuracy have improved compared with traditional methods, but in the prediction of longer time series There is still the problem of excessive error

Images