Research And Application Of State Monitoring Technology For Marine Fishing Vessels Based On Multi-source Information Fusion

In recent years,China's marine fishery informationization construction developed rapidly,more and more fishing vessels have been included in the Vessel Monitoring System(VMS),the monitoring data of fishing vessel has exploded and the trajectory data as the key information of it also got a lot of accumulation.How to monitor of fishing vessel state more efficient and accurate through the monitoring data is an urgent problem to be solved by the fishery regulatory authorities.Trajectory prediction and anomaly detection are two important components to the state monitoring of fishing vessel,and accurate trajectory prediction and anomaly detection have important sense to fishing vessel state monitoring.This dissertation has conduct some researches about the above issues.(1)In order to monitor fishing vessel more effectively,this dissertation implements a vessel monitoring system based on Beidou and AIS.The main functions of the system include chart display,ship query,track playback,trajectory prediction,anomaly alarm and so on,among them,trajectory prediction and anomaly alarm are two important components of the vessel monitoring system.(2)In order to monitor fishing vessel more accurately when vessel data is lost,this dissertation makes full use of multi-source trajectory data of fishing vessel and proposes a trajectory prediction algorithm based on multi-source information fusion.Firstly,the algorithm fuses the Beidou and AIS trajectory data to make the fishing vessel trajectory closer to its actual trajectory,then,the trajectory prediction is carried out to improve the accuracy of trajectory prediction.Experiments show that the algorithm can improve the accuracy of trajectory prediction significantly.(3)For the problem of traditional anomaly detection algorithms only focused on the location of trajectory,and ignore motion attributes of the moving object,such as direction and speed,this dissertation takes account of motion attributes into anomaly detection based on the TRAOD algorithm,proposes a fishing vessel trajectory multidimensional feature collaborative anomaly detection algorithm(FVT-MCAD).The algorithm uses six independent components to detect the anomaly of fishing vessel,including instantaneous angular acceleration,average angular acceleration,instantaneous speed,average speed and acceleration,then combines the anomaly trend scores obtained by each component and define anomaly condition of fishing vessel by comparing with the anomaly threshold.In addition,the algorithm can set the component weight according to the validity of different components,and avoid the problem that the final anomaly trend score is overly dependent on a certain component.Finally,the robustness and reliability of the FVT-MCAD algorithm are verified by experiments.