Research On Algorithm Of Deep-Sea AUV Integrated Navigation System And Its Observability Analysis

As we more and more interest is directed to oceans, researches on Autonomous Underwater Vehicles (AUV) are receiving intensive attention from around the world. At present, the SINS/DVL navigation system is widely used in underwater navigation technology abroad. This paper is based on the SINS/DVL navigation system and focuses on the AUV integrated navigation system. The main research includes the basic principles of AUV Navigation calculating, online calibration, observability analysis and RBAUKF algorithm.The major work carried out in this research is as follows:1) The coordinate systems of the AUV integrated system is introduced, and their transforming relations are deduced. An error model of SINS is established. Error equations of the AUV integrated system, including attitude, velocity and location error equations are analyzed and deduced.2) Online calibration of the AUV integrated system is investigated. An error model of SINS is established. The online calibration is based on DVL and the path, and the calibration process is given. The results show that part of the error parameters can be estimated using a kalman online calibration method. Installation errors and scale factor errors of DVL are estimated, which are extended as the state vector in the kalman filter.3) The observability analysis of the AUV integrated system is investigated. The concept of observability and the significant measure in a linear system is introduced, and then the methods of PWCS are given. The method of Singular Value Decomposition(SVD) is proposed to be used in observability analysis. This method can help validate the observability of the error parameters and calculate singular value of all state variables.4) The nonlinear filtering algorithm of the AUV integrated system is investigated. Principles of the UKF algorithm are introduced. The indirect method of RBAUKF is used in the SINS/DVL navigation system. A direct method filtering model is also built, and simulation results are given, which show that both the direct method and the indirect method can get high precision while AUV has strong nonlinear. Compared to the indirect method, the output of the direct method filtering model is navigation parameter, which has a smaller amount of calculation.