Research On In-motion Initial Alignment And Real-time Information Fusion Methods For GNSS/MIMU Integrated Navigation System

GNSS/MIMU integrated navigation system, as the research object, is studied in this dissertation. When it comes to the theory and algorithm of initial alignment for integrated system, our research is focused on GNSS/MIMU in-motion initial alignment scheme, observability analysis, improved Kalman filter based on observability or devised for real-time applications.As a first point, it is generally difficult for low-cost MEMS-IMU with low accuracy to achieve effective initial alignment. In order to address this problem, a novel and fast coarse alignment approach is adopted to obtain MEMS-IMU’s initial attitude information analytically. Vehicle and fight results show that this method is accurate enough to be the initial value for a fine alignment with little restrictions on the motion. This is called in-motion coarse alignment assisted by dual-epoch GNSS information.Subsequently, the fine initial alignment method for GNSS/MIMU integrated navigation system based on the coarse value is studied. A transfer alignment scheme called ‘velocity matching + location updating’ is adopted to realize fine initial alignment. Experiments show the fine alignment has achieved high convergence precision(<2deg) and short convergence time(<100s), in accordance with GNSS/FIMU results and considering poor performance of MIMU.Besides, as for problems such as variables selection and performance improvement, the observability theory has been studied. The effectiveness is demonstrated through tests with two different algorithms, which shows the one proposed by our team has advantages over the traditional ones in high-dynamic environment. The amount of calculation is much smaller and results are more suitable to actual circumstances. Motivated by this, an improved Kalman filter based on observability is proposed in hope of higher accuracy, shorter alignment time and improved MIMU performance. Experiments show that it is equivalent with classic EKF when system is observable; it can improve filtering precision and stability when system not completely observable.Finally, in view of navigation output delay owing to a delay in the transmission of GNSS information, a real-time correction algorithm is devised and analyzed by simulation to tackle with the problem which is encountered in the construction of GNSS/MIMU integrated system. The result shows that the method can realize the real-time navigation output in high accuracy.