Marine Target Behavior Recognition System And Realization Based On Machine Learning

In the 21 st century,the oceans have once again become the key to global contention,and the strategic status of countries with coastlines has increased rapidly.As a maritime power,due to historical reasons,there are maritime disputes with many neighboring countries.The relevant countries frequently violate China's territorial sea rights through island occupation,oil and gas stealing,illegal mapping,illegal fishing,and illegal intrusion.Due to the lack of effective control methods,nearly 90% of China's jurisdictional waters have not been fully monitored and can only be dealt with after the fact by law enforcement cruises,satellite remote sensing and other methods.The methods are single,time-efficient,and lack of early warning identification and continuous tracking and monitoring capabilities.,Seriously restricting the in-depth implementation of China's maritime power strategy.The research of this project is mainly targeted at three business scenarios: sea area development and utilization activity monitoring and detecting,island and reef change monitoring,and marine moving target behavior recognition.This thesis is mainly related to the behavior recognition business of maritime mobile targets,that is,to establish radar data,AIS data,electromagnetic signal detection data,visible light and infrared image data and other data,establish multi-source data association rules,and integrate target attributes based on machine learning.Therefore,the traditional decentralized and independent identification method is changed to a cooperative and non-cooperative combination mode.Then,based on the rule template technology,the target abnormal behavior is modeled to achieve targeted abnormal behavior recognition.This thesis aims at the behavior recognition and monitoring needs of typical targets in the jurisdictional waters of our country.It makes full use of the team's existing equipment,systems and research foundations,and targets application scenarios such as specific sea surveillance,maritime mobile target behavior recognition,and uses information fusion,machine learning,and rule templates.And other technologies,design the perception system,communication system and processing system under the integrated monitoring system of "sky,air,ship,and shore",build a marine target behavior recognition system with a wide range of early warning and continuous tracking capabilities,and achieve multi-device collaboration Surveillance,and effectiveidentification and continuous tracking of dynamic targets at sea,solve the problems of scattered recognition methods,and low target recognition rate.The research results of this thesis have been applied in the normalized demonstration areas of the South China Sea,providing a more effective tool for improving China's maritime sovereignty.