Marine target trajectory prediction method and device based on multi-source data

Abstract

The invention provides a maritime target trajectory prediction method and device based on multi-source data. The prediction method comprises the following steps: acquiring multi-source sensing data of a maritime target; on the basis of the multi-source sensing data, track grid mapping coding is carried out on the marine target; performing data fusion on the multi-source sensing data in the track grid within a preset time period; obtaining an input sequence based on the sensing data after data fusion, and inputting the input sequence into a prediction model; and calculating through a preset model to obtain an output sequence of marine target trajectory prediction. Aiming at the problem that a ship target cannot be continuously tracked under incomplete observation data information expansion, complementarity of multi-source data in ship target monitoring is fully utilized, confidence of multi-source means is considered, and through key technologies such as track grid mapping coding, multi-source data fusion and a prediction model, the ship target can be continuously tracked. The existence probability mapping of the marine target trajectory in a long-time sequence and a large-scale space under global grid subdivision is realized, and the prediction analysis of the marine target trajectory is realized.

Description

technical field [0001] The invention relates to the technical field of maritime target trajectory prediction, in particular to a method and device for maritime target trajectory prediction based on multi-source data. Background technique [0002] In order to realize the prediction of ship trajectory, in related technologies, a research method of ship trajectory prediction based on recurrent neural network is proposed. In order to ensure the regularity of trajectory prediction, commercial or civil ships with non-confrontational behavior are selected as experimental objects. This method constructs two prediction models based on recurrent neural network, namely Long Short-Term Memory Model (LongShort-Term Memory, LSTM) and Gated Recurrent Unit Model (Gated Recurrent Unit, GRU). The AIS data set is used as the input of the model. Training, and finally realize the accurate and efficient prediction of the longitude and latitude coordinates of the ship in the future. A comparative...

AI Technical Summary

Problems solved by technology

[0005] (1) Single data source:

[0006] Existing ship trajectory prediction algorithms only use AIS data. Ships with special purposes will turn off the AIS signal after entering and leaving the port, resulting in sparse and discontinuous ship trajectories, while remote sensing, public opinion, radar, radio and other data will effectively increase the density of ship trajectories. Algorithms that simply use AIS data as a data source will lead to inability to accurately predict ship trajectories

[0007] (2) The prediction space-time range is limited:

[0008] Existing ship trajectory prediction algorithms use neural networks to accurately predict ships in short time series and small spatial ranges, but cannot predict long time series and large-scale space ship trajectories.

[0009] (3) Application scenarios are limited:

[0010] The existing technology can only be applied on the premise that the monitoring object turns on the AIS device, and it is not applicable to the scene where the monitoring object in a special field turns off the AIS or turns on the AIS device intermittently, and the latter is in the application scenario of actively turning off the AIS behavior field more common

Images