Research On The Multi-position-reference-system Information Fusion Methods Of Dynamic Positioning Vessel

Ship's dynamic positioning system is an indispensable key equipment of marine development,job search and rescue and resource mining,as well,it is the technical support of deep sea research.To keep the vessel in a fixed position or path in the process of lifting and laying work,it is necessary to feedback the position of vessel to the controller which is measured by position reference system.It constitutes a close-loop control system which controls the propeller of vessel against the environmental interference.In general,more than one different or the same types of position reference system are installed in vessels.How to obtain the precise position information from multiple position reference systems is a key technology of precise vessel positioning.As dynamic positioning system is nonlinear,process noise and measurement noise may have a correlation(correlation I),as well as among measurement noise(correlation II).Therefore,aiming at above problems,this paper uses the "offshore oil 201" ship as the research object,adopts a way of combining theory and simulation,proposes a decentralized cubature information fusion method,measured data of multiple position reference systems is fused accurately and effectively.The main research work of this paper is as following:(1)Analyzing the vessel motion in the marine environment,establish motion mathematical model of the dynamic positioning vessel and environmental interference model,and establish position reference system model of GPS,acoustic,tension cable,Radius.(2)Aiming at the problem that measured data of position reference system is based on different sampling frequency and the coordinate of reference system itself,and it is not precise as the phenomenon of freezing,outliers,the slow drift will come out due to system itself or interference of environment.In order to guarantee the reliability of the subsequent fusion,time registration,space registration and quality detection are required for the measured data of position reference system before the data fusion.(3)As process noise and measurement noise are nonlinear in this system,cubature Kalman filter is proposed which is suitable for high dimension and more precise than extended Kalman and boundless Kalman.With deduced framework of extension information filter in addition of noise correlation,embedding cubature Kalman filter of noise correlation in it,cubature information filter is deduced aiming at process noise and measurement noise.(4)Using decentralized fusion structure,the filtered results of the cubature of information filter are fused.By comparing simulation results with the real position value,analyzes the effectiveness of proposed fusion algorithm.The results show that information fusion function of position reference system realizes well,which is a method of decentralized cubature information aiming at of process noise and measurement noise.