Research On SINS/GPS Tightly-coupled Integrated Navigation Based On Nonlinear Filter Algorithm

In various navigation systems, the strap-down inertial navigation system (SINS) and global positioning system (GPS) integrated navigation system is recognized as the optimal navigation mode since it combines the complementary characteristics of these two systems and overcomes the shortcomings of ethier alone. In aircraft, aerospace, voyage and weapon guidance applications, navigation takes over an important and unnecessary part. Due to the wide demands of navigation in various engineering applications, a deep and fine research on SINS/GPS integrated navigation system algorithm is very significant. In this dissertation,SINS/GPS tightly-coupled integration navigation system is taken as the background and the research stresses on the data fusion algorithm in navigation system, i.e., nonlinear filtering algorithm. Based on the working quality of SINS/GPS tightly-coupled integration navigation system, the research details are as follows:The principles of SINS, GPS and SINS/GPS tightly-coupled integration navigation system are analyzed respectively. The SINS kinematical equations are derived in local geographic coordinate system and its error model is established based on the concept of Euler error angle. For GPS, how it works about location and speed detection is introduced in detail and its various error sources are analyzed deeply. Then the research focuses on SINS/GPS integration navigation system. State equation and measurement equation are derived.Combining the specific characteristics of the system equation, some analyses have been done and it can be found that the nonlinear filtering in SINS/GPS tightly-coupled integration navigation system generally has the following problems: the correlation between process noise and measurement noise introduced by discretization of the continuous state equation,colored noises in GPS receiver's measurement and linear measurement updation in traditional Kalman type filters.Aiming at addressing the correlation between process noise and measurement noise, a new nonlinear filtering algorithm is proposed. According to the time correlation occurs, the correlationship can be classified in two types: the process noise and measurement noise are correlated at the same epoch and one time step apart. Then two algorithms are designed from the perspective of conditionally Gaussian noise distribution, based on the framework of Gaussian filter. Comparing with existing methods at present for these two types noise correlation, merits of the new algorithms are that: both of them give out a general framework of nonlinear filter for different kinds of noise correlations such that various numerical integration methods can be adopted. What's more, the proposed method can achieve a better performance than existing method for a specified noise correlation.The problem of colored noise in state estimation in nonlinear system is studied. The Kalman type filters all work on the prerequisite that the process noise and measurement noise are zero-mean Gaussian white noise, which is a difficult requirement in engineering applications. In some practical systems,the measurement noise exits temporal correlation- In the case of strong temporal correlation, the effect of colored noise couldn't be ignored.Therefore,a new algorithm, based on new state augmentation approach, suitable for linear/nonlinear system with colored measurement noise, is proposed in this dissertation.How to get a better measurement updation on state estimation in nonlinear system is also considered. In traditional nonlinear Gaussian filters, the measurement updation obeys the minimum variance rule and assumes that the state and measurement are joint Gaussian distributed which results in a linear updation on state. The detail analyses show that these couldn't obtain an optimal state estimation because of the ignorance of nonlinearity of measurement function. So a recursive update Gaussian filter algorithm is proposed which stresses on the "recursive update". The proposed algorithm has a general framework which is easy to be implemented with various numerical integration methods. Simulation results demonstrate that the proposed method can achieve a better filtering performance.The basic principles of trajectory simulation are studied. Based on the detailed derived carrier's kinematical equations under different states of motion, the true trajectory emulator,SINS emulator and SINS/GPS tightly-coupled integration navigation simulation system are designed by Matlab. Using these emulators, problems of correlated noise, colored noise and measurement updation of nonlinear filter of SINS/GPS tightly-coupled integration navigation system are testified. Simulation results show that the proposed method is effective for the corresponding problem which improves the navigation accuracy of system.