Research On Navigation And Data Fusion Technology For Autonomous Underwater Vehicle

The underwater vehicle is applied widely as improtant device for exploring ocean inmilitary use and in civil use. With the deepening of ocean research and the sophistication ofmilitary requirement, the multiple underwater vehicle cooperation system is a hot issue inunderwater vehicle domain research.The high precision navigation and positioning istechnical support for underwater vehicle working securely and reliably under water. So thisdissertation investigates two aspects including single underwater vehicle navigation andpositioning and multiple underwater vehicle navigation and positioning. The major work is asfollow:1、From the Bayesian estimation viewpoint, Bayesian estimation theory under gaussiandomain is analysied. Baesd on the gussian Bayesian estimation framwork, the celebratedKalman filter is derived. With the nonlinear system under guassian domain, traditionalnonlinear approximate filtering algorithms including the extended kalman filter andunsecented kalman filter, and a new nonlinear filter, namely cubature kalman filter, arediscussed. The filter performance is compared for all kinds of nonllinear filtering algorithmsthrough simulation experiment.2、Based on the concept of Enler platform error angle, the nonlinear error model isdeduced. In view of initial alignment of strapdown inertial navigation system with largealignment on dynamic base, the velocity of Doopler Velocity Log is used to help implementinitial alignment in motion. Meanwhile, due to nonlinear of error model of strapdown inertialnavigation system, a new cubature kalman filter based on nonlinear prediction filter, namelyNPF-CKF, is put forward, and is applied for alignment of underwater vehicle on dynamicbase. The simulation is carried out and shows that the method is efficient.3、The integrated navigation of underwater vehicle is investigated. The mathematicalmodel of the integrated navigation system of SINS/DVL is established. The revising methodsare discussed, and the mixed revising method of output revising and feedback revising isproposed and is applied SINS/DVL integrated navigation. The Sage-Husa adaptive filteralgorithm and strong tracking kalman filter are studied. In view of unkown or time-varyingfor measurement noise of Doppler Velocity Log in actual working environment, The strongtracking adaptive filter algorithm of time-varying measurement noise is proposed, andapplied SINS/DVL integrated navigation. The simulation is carried out through test data.4、In order to enhance the navaigation and positioning precision of underwater vehicle,mutiple underwater navigation sensor information fusion technology is discussed. The federal filter is deduced and analyzed theoretically. The mathematical model of integrated navigationof underwater vehicle is established. In view of the susceptibility to measurement noiseunkown or time-varying of local filter in traditional federal filter, the adaptive federal filter isproposed in which strong tracking adaptive filter algorithm is applied in local filter. Then thefederal filter proposed is applied in mutiple navigation sensor information fusion.Finally, thesimulation is carried out.5、In view of requirement of mutiple underwater vehicles collboration, the cooperativenavigation of multiple underwater vehicles is discussed. The fundamental principle ofcooperative navigation of mutiple underwater vehicles based on many leaders is analyzed,andthe mathematical model of cooperative navigation of mutiple underwater vehicles based onmany leaders is established. The cooperative navigation algorithm based on EKF is proposed.In order to simply the system structure, the cooperative navigation of multiple underwatervehicles is discussed. The fundamental principle of cooperative navigation based on singleleader is analyzed,and the mathematical model of cooperative navigation based on singleleader is established. Based on the observability theory of nonlinear system, observabilityanalysis of cooperative navigation based on single leader is done under Cartesian coordinate.The observability condition is obtained. Finally, the simulation is carried out and proves thatthe validity of algorithm of cooperative navigation based on single leader.