| As China gradually promotes the construction of a strong marine state,the domestic marine fishing industry is developing rapidly and the number of fishing vessels is also rising rapidly,but collisions caused by fishing vessels are also occurring more frequently.The main reason is that China’s fishing vessels are still mainly small and medium-sized fishing vessels,which have fewer operators and backward navigation equipment,and the overall intelligence level is low,while most of the existing research work is aimed at automatic collision avoidance of large ships or developed as part of the bridge system,which requires large instruments for a large number of complex calculations and high costs,and cannot be promoted in small and medium-sized fishing vessels.Therefore,it is an important part of the solution to the collision avoidance problem of small and medium-sized fishing vessels to obtain the information of the surrounding environment of the vessel quickly and accurately in a low-cost way.Radar and Automatic Identification System(AIS)are the two most commonly used devices for fishing vessels to obtain collision avoidance information.The information obtained by these two sensors is different.Fusing the radar and AIS information can help ensure the safe navigation of fishing vessels in fishing production activities.This thesis focuses on the research of radar and AIS information fusion technology and the development of a comprehensive display system for fishing vessel collision avoidance information.The system utilizes the existing equipment on fishing vessels to improve their collision avoidance capabilities.The main research contents of this thesis are as follows:(1)A track association model based on improved deep forest is proposed to address the problem of radar and AIS track association.The e Xtreme Gradient Boosting are introduced during the Cascaded Forest learning phase of the Deep Forest to improve the performance of model.After training the forest,a threshold method is used to determine whether radar and AIS targets are associated.Simulation results show that the proposed method can achieve the track association of radar and AIS targets,and has a higher association accuracy compared to the original deep forest and traditional algorithms.(2)A track fusion method based on Sparrow Search Algorithm(SSA)optimized Back Propagation Neural Network(BPNN)is proposed to address the radar and AIS track fusion problem.The SSA algorithm is used to optimize the weights of the BPNN,thereby improving the network’s performance.Simulation results show that the proposed method can effectively fuse radar and AIS information.Compared with Particle Swarm Optimization(PSO)-BPNN,the convergence speed of SSA-BPNN is faster,and the track fusion accuracy is higher.(3)Based on C++ language and Qt5.7.1 development tool,a comprehensive display system for collision avoidance information of fishing vessels was designed.The track association and fusion algorithms studied in this thesis were embedded into the system to achieve radar and AIS information fusion,providing accurate surrounding vessel information for fishing vessels and improving their safety. |
