Research On Geomagnetic Measurement Error Compensation For Underwater Geomagnetic Navigation

True autonomous navigation technology of long duration and high-precision becomes one of the main factors that restrict the development of autonomous underwater vehicle(AUV). Inertial navigation system(INS) has been core navigation equipment for AUV because of its excellences such as independence and whole navigation information. However, due to the existing shortcomings that the navigation error increases with time, the INS cannot meet the AUV’s mission. Geomagnetic matching technology can provide passive external resource of correction information for AUV, which improves the INS’s long voyage navigation accuracy. Hence this technology has become a research hot point of navigation. This thesis is focused on key techniques of underwater geomagnetic navigation, including measurement error compensation of geomagnetic field under strong uncertainty, analysis on the space distribution of measurements impacting the effect of measurement error compensation, and INS/Geomagnetic matching method for underwater navigation. The performance of the proposed techniques and methods are preliminarily verified by simulations and real experiments. Main contents and contributions of this thesis are as follows:(1) Based on a deep analysis of propagation characteristics of geomagnetic measurement error, an integration model is built including various of error sources. A motion constraint is suggested for maintaining the stability of vehicle magnetic field and easily compensation based on the analysis and tests of the dynamic changes of the vehicle magnetic field, which simplifies the problem of online geomagnetic measurement errors compensation. The principle of geomagnetic measurement error compensation algorithms is analyzed, and a framework of the algorithms is built, which provides strong conditions in applications.(2) A geomagnetic measurement error compensation algorithm based on the constrained total least squares estimator is proposed. The traditional compensation method was analyzed and corrected. Since the charactistics of measurement noises vary and become correlated, the constrained total least squares method is used to suppress the influence of measurement noises in both sides. Besides, the noises correlation was utilized to further improve parameters estimation accuracy. The numerical solutions are given by the derived real Newton method. Results of simulations and vehicular experiments verify the effectiveness and feasibility of this method.(3) A geomagnetic measurement error compensation algorithm based on the particle swarm optimization strategy is proposed. Since the parameterized model is strong nonlinear and the expression is complicated, the particle swarm optimization strategy is used for searching the global solutions, which avoid the drawback of good initial solutions dependent for the analytical optimization algorithms. The stretching function technology is embedded into the particle swarm optimization strategy for providing the capability of jumping out of local solutions, which increases the ability of global convergence, and improves the robustness and accuracy of the algorithm. Results of lab experiments and field experiments show that the estimates accuracy is better than the existing analytical optimization algorithms.(4) An algorithm based on D optimal design method was proposed for magnetometer maneuver arrangement optimization under the space constraint. As for the process of magnetometer calibration, multiple attitude maneuver of the sensor can encourage geomagnetic measurement errors and improve the estimation accuracy of model parameters. However, this is difficult to realize as its maneuverability is constrained by the strapdown navigation vehicle. The D optimal design is started from the point of experiment design and finds an optimal arrangement for the magnetometer in an allowed space. Simulations show that the condition number of geomagnetic measurement equation is sufficiently reduced, as well as the improvement of the geomagnetic measurement error compensation algorithm in terms of stability. The method can guide for operating the planning of AUV maneuver in the calibration spot.(5) To solve the ill-posed problem in shipboard three component magnetometer(STCM)calibration, a regularization method based on the truncated total least square technique combined with L curve method. A deep analysis is given on the ill-posed cause in the STCM calibration problem. Considering the mounting environment and the maneuverability of the sensor, the above algorithm was therefore proposed. It firstly has singular value decomposition about the total least square estimates, and truncates the sections of solutions that are impacted by measurement noises to get a more stable estimate. Results of simulations and experiments show that the proposed method can effectively reduce the influence of the ill-posed problem and get high-precision and stable estimates. Based on the compensation of the geomagnetic measurements of the water filed, a high accurate geomagnetic contour map was built.(6) An INS/Geomagnetic integrated localization algorithm based on tree searching strategy was proposed. Based on the analysis of the existing geomagnetic matching algorithms, a novel geomagnetic localization method was presented by considering the compensated geomagnetic measurements and the motion characteristics of the AUV. The method described the geomagnetic measurement errors and the INS navigation errors by using the bounder principle, and converts the localization problem into a tree searching problem. The breadth-first search and correlation matching criterion are combined to determine the best matching path. The results of the underwater experiment show that the true geomagnetic field signature can be effectively abstracted, the disturbing magnetic magnitude is reduced to several tens of nT. The proposed localization method can sufficiently fix the position error of the AUV, since the root mean square error value of the position error was reduced from 2.5 km to 139 m. These results provide the preliminary feasibility of the INS/Geomagnetic integrated matching navigation technique.