The Integrated Navigation System Based On The Federal Kalman Filtering And Machine Learning

With the development of the modern navigation system, it is stricter about the requirements of the navigation accuracy, reliability, fault tolerance and long hour continuous working. INS (Inertial Navigation System) and GPS (Navigation Satellite Timing and Ranging Global Position System) are the most widely used and the most mature navigation system in the international. They have highly complementary, so they can learn from each other, and combined into integrated navigation system, which can full play every navigation methods advantages, overcome the shortcoming and improve the overall system efficiency and fault tolerance.This paper designed the model of the integrated navigation system respectively based on Federal Kalman Filter and partial least squares support vector machine, and also considered the real-time and fault tolerance of these models. These models are effectiveness and feasibility through simulation used simulated trajectory. The concrete works of this paper are as follows:1) This paper made a brief analyze for the working principle and basic equation of SINS and GPS, and pointed out the deficiencies that each of the tow navigations in actual use. So Integrated Navigation is necessary and feasibility. And then, two simplified integrated navigation was build and explained their advantages and disadvantages;2) This paper introduced the Kalman filter theory and basic equations and then proposed the Federal Kalman Filter model, and analyzed the several combination modes and system structure. Chosen One of these as the basic model framework to design INS/GPS integrated navigation system. Theory flight trajectory simulation results showed the practicality and feasibility of the model;3) The fault-tolerant structure of the Federal Kalman Filter is designed using classical algorithm. Simulation results showed that this structure can detect and isolate system fault effectively;4) Finally, the paper discussed the application of least squares support vector machine used in the integrated navigation system. And then, designed an integrated navigation system structure based on this algorithm, and proved the stronger regularity in the data processing of navigation system, It point out the direction for further research of the future navigation system.