The dissertation focuses on the application of nonlinear filtering method in AUV navigation systems. The coarse alignment methods in the inertial frame of SINS, the indirect nonlinear initial alignment approach and the direct nonlinear filtering approach for AUV navigation systems are studied. The main research work and results are as follows:
(1) With or without the consideration of longitude, two coarse alignment methods in the inertial frame of SINS are investigated. Since the earth’s angular velocity is very small, the error of traditional coarse alignment method is very large and not even applicable when the AUV is swinging in disturbing or moving environment. We theoretically analyze the influence of the linear acceleration on the alignment accuracy of the coarse alignment methods in the inertial frame. The coarse alignment methods in the inertial frame are compared with the traditional analytical coarse alignment by means of the mathematical simulation in the condition of various motions. The correctness of the theoretical analysis is verified through the mathematical simulation of various motions, and the coarse alignment method in which the longitude is considered can achieve better performance, and it can adapt to a variety of environments. The coarse alignment performs well not only when the AUV has a large angular movement, but also when the AUV has the constant linear velocity.
(2) The SINS indirect nonlinear initial alignment method using GHQKF is presented. When AUV is influenced by the waves, especially in the poor sea conditions, the coarse alignment accuracy is too low to use the linear filters, and the nonlinear filters is considered to achieve the SINS initial alignment on the moving base. The configuration method for one dimensional Gauss points and their corresponding coefficients in Gauss-Hermite quadrature is derived, which is then extended to form the multidimensional Gauss points and their coefficients using the direct tensor method. A GHQKF is presented, and the SINS nonlinear error model with large azimuth misalignment angle is used to verify the new algorithm. The new algorithm based on the GHQKF is better than the non-linear initial alignment based on the UKF. With the increase of the number of one dimensional Gauss-Hermite quadrature points, alignment accuracy can be further improved. However, the calculation load also increases a lot, so we need to take into account the performance and the calculation load.
(3) In order to solve the problem that the initial alignment method based on the GHQKF has big calculation load, the SINS nonlinear initial alignment using the SGQKF is presented. The algorithm based on SGQKF can keep the alignment performance, and can also reduce the calculation load and improve the practicability. The theory of sparse-grid and Gauss-Hermite quadrature is adopted to generate multidimensional quadrature points and weights. The semi-physical simulation experiment verifies the superiority of the new algorithm. When the initial azimuth errors are small, the new algorithm can achieve better alignment performance than Kalman. If the initial azimuth errors become larger, the new algorithm shows similar alignment performance with GHQKF. The calculation of the new algorithm is much smaller. Considering the alignment accuracy, convergence speed, computation and applicable conditions, the SGQKF at the level-3 quadrature accuracy can be used in SINS initial alignment.
(4) A direct nonlinear approach for the information filtering is presented, using the measured velocity in the body frame as the filter measurement and Euler angle kinematics as the system model for AUV navigation systems. Comparing with the indirect approach based on the error models, the direct approach can combine the SINS navigation calculation and the filtering, and it also can better track the system nonlinearity. The AUV navigation system is usually equipped with DVL, providing the velocity in the body frame to assist the SINS. The velocity in the body frame is used as the filter measurements without the velocity projection from the body frame to the local-level frame as in traditional approaches, so the cross-noise problem is avoided. The proposed approach is based on Euler angle kinematics equations instead of the quaternion kinematics equations, so the constraint in the quaternion no longer exists. The model, using the measured velocity in the body frame as the filter measurement and Euler angle kinematics as the system model, is established in detail. The direct approach uses an UKF because the system model and measurement model are nonlinear. The performance of the proposed approach is verified by road test data from a land vehicle equipped with an odometer and GPS. The proposed approach can achieve higher navigation accuracy than the commonly-used indirect approach. As an alternative, it can be considered as a supplement to the conventional integration navigation algorithm.
(5) The problem how to avoid the singularities when using Euler angles in the direct approach is discussed, and a non-singular and nonlinear approach for the AUV initial alignment based on the dual-Euler method is presented. The new algorithm is established to solve the singularity problem which exists in the direct approach using the single Euler angle set. The characteristics that the singularity is complementary of the positive and passive Euler angle differential equations are analyzed. The direct approach using dual-Euler method is applied for the nonlinear SINS initial alignment, the alignment flow is presented, and the semi-physical simulation is carried out. The singularity problem, which exists in the Euler angles, can be efficiently solved by means of the dual-Euler method. The proposed approach is non-singular because the positive and passive Euler angle direct approaches operate alternately according to the switch flag to remove the singularities, and the non-singular direct approach can be applied to the case of large maneuver for the AUVs.
(6) The indirect and direct approaches for the SINS initial alignment in the inertial frame are investigated, which further extends the idea of the coarse alignment in the inertial frame to the fine alignment. The indirect error model and the direct model in the geographic frame are modified, and inspired by the direct approach in the geographic frame the direct SINS initial alignment in the inertial frame is presented. The indirect error models are established by means of two different measurements, and their corresponding fine alignment flow charts are presented. The semi-physical simulation test is conducted to verify the proposed indirect approach. The steps to establish the nonlinear direct filter model in the inertial frame are presented in detail. The UKF is applied to estimate the states. The validity of the proposed direct approach is verified by mathematical simulation, the heading convergence speed is improved, and its convergence time is also shortened.