Research On Algorithm Of MIMU/GNSS Vehicle Integrated Navigation System Based On Improved AKF

With the fast developing of High-Tech since 21st century, microelectronics technology and micro mechanical technology have been improved and they promote the fast improvement of MEMS inertial sensors technology. More and more scholars began to join into the study of MEMS inertial sensors. The MIMU systems, based on the MEMS inertial sensors, have been widely used in mini robots, cars navigation system, IPhone, and unmanned aerial vehicles, due to its advantages, including miniaturization, low cost, batch production and high reliability.Since the MEMS inertial sensors are with low precision and its error accumulates over time, the error of MIMU system is obvious and easy to transpire. Therefore, MIMU system won't be able to complete the vehicle navigation independently. With MIMU the main system and the satellite navigation system the aiding-system, this paper takes advantage of the information complementarity of the MIMU system and the satellite navigation system, and focus on the study of MEMS gyroscope noise reduction program and adaptive linear information fusion method of the MIMU/GNSS integrated navigation system.This paper introduces the current research and development status of MIMU/GNSS integrated navigation, the basic combination and the classical information fusion algorithm of car integrated navigation system. Based on above, the paper studies the MEMS gyroscope noise reduction technology. Firstly, analyze the random error of MIMU devices, use the Allan variance and autoregressive process to identify error, and build MEMS gyroscope error model.Secondly, design MEMS gyroscope noise reduction program based on model, to reduce random error, and to lay a solid foundation for the subsequent studyIn the study of MIMU/GNSS integrated navigation fusion Algorithm, AKF filter Algorithm is built, for improving observer status variance R value, to solve the optimal estimation information output instability problem of the car integrated navigation system.Based on above, square root filter is used, to solve the state quantity calculation error in the process of information fusion. Combining advantages of AKF and square root filter, S-AKF algorithm (the AKF based on square root) is promoted, and simulation method is used to verify its practicability.In the End, the paper conducted static test and vehicle dynamic test, using MTi-G and GNSS receiver. The test result shows that, S-AKF is better than AKF in vehicle speed,location and station error fusion effect. The error of speed is reduced by 0.5m/s,the error of location is reduced by 5m, the error of attitude is reduced by 0.1° .S-AKF well processed the random noise during vehicle moving, and compensated the calculation error, which proved the practicability of S-AKF algorithm.