An Improved Kalman Filter Is Applied Research In The Ins / Gps Integrated Navigation

Kalman Filter has become the mainstream technique of navigational data processing up till now. Various new algorithms based on Kalman Filter have constantly welled up. For the actualization of conventional Kalman Filter, the models of dynamic system and measurement system as well as statistical characteristics of driving noise and measurement noise must be known exactly. The models established do not agree with actual dynamic system owing to lopsided understanding dynamic system or simplifying models. And it is difficult to obtain exact statistical characteristics of noises. Thus filtering noises out of the dynamic models established sometimes appears divergence by conventional Kalman algorithm:1. In this thesis, by modeling and simulating the Inertial Navigation System and Global Positioning System, concrete reasons for filtering divergence of the Inertial Navigation System models are analysed and it points that some ready Adaptive Kalman Filter algorithms can not always avoid filtering divergence. This thesis presents an algorithm based on nonstationary modeling and lucubrates relative problems.2. This thesis mainly researches how to inhibit filtering divergence of the Inertial Navigation System models using algorithm based on nonstationary modeling. At the beginning, an nonstationary linear model is established in terms of the sorts of dynamic systems and a new Kalman Filter algorithm is created based on this model. Afterward, factor of filtering divergence is found from this algorithm and emphasized in this algorithm created. On these bases, an expected and improved nonstationary linear model is established which is of stationarity. At last, simulating results prove that the improved Kalman Filter algorithm can surely inhibit filtering divergence of the Inertial Navigation System models.3. In addition, this new algorithm is applied to data fusion of INS/GPS Integrated Navigation Systems. Concrete simulating results prove that improved filtering algorithm can not only inhibit filtering divergence of the Inertial Navigation System models and increase its weight to optimized Federal Filtering estimation but also improve precision of optimized Federal Filtering estimation.