GNSS/MEMS INS Deeply Integrated Navigation And Its Integrity Monitoring

Precise and reliable navigation is essential for the operations of a broad range of modern automated systems such as driverless car and unmanned aerial vehicle.GNSS(Global Navigation Satellite System)and INS(Inertial Navigation System)are two prevalent navigation systems with plenty of complementary characteristics;deeply integrating them is conducive to the meeting of strict navigation requirements.With the rapid development of MEMS(Micro-Electro-Mechanical System)technology,the MEMS IMU(Inertial Measurement Unit)is increasingly employed to implement INS and the deeply integrated navigation system of GNSS and MEMS INS(MEMS IMU based INS)could be a prospective navigation approach in military and civil applications.The integrated navigation involves fusing multi-source navigation data via data fusion algorithms to obtain better navigation result compared to that provided by stand-alone navigation system.In order to improve system's fault tolerance and computational efficiency,etc,distributed data fusion algorithms are commonly applied in the integrated navigation with multiple sensors.Accuracy and consistency are two important indicators for the performance of data fusion algorithm.In this paper,the principle relationship of CI(Covariance Intersection),CC(Convex Combination),LE(Largest Ellipsoid)and EI(Ellipsoidal Intersection)fusion algorithms is analyzed,and the consistency proofs of CC,LE and EI algorithms are presented.A PLE(Parallel Largest Ellipsoid)data fusion method with parallel fusion structure is also proposed.Compared to the CI algorithm and the SLE(Sequential Largest Ellipsoid)data fusion method with sequential fusion structure,the proposed PLE method has better overall performance in terms of accuracy,consistency and computational efficiency.With mutual assistance among tracking channels,the vector tracking loop has superior performance over scalar tracking loop.In the deeply integrated navigation system of GNSS and INS,the receiver baseband and INS are deeply coupled into a vector tracking based closed-loop system.The stability of GNSS signal tracking and the accuracy of navigation error estimating are vitally important for deep GNSS/INS integration.By treating the short-term vector tracking based code loop as a constant linear system,the error-state space model of vector tracking based code loop containing receiver baseband and navigation filter is built in the complex domain,and its ability to track step code phase input is subsequently analyzed.Based on the inequality between the GNSS signal tracking error and the navigation solution derivation error,four dual-vector tracking based tracking loops are presented and their mathematical models are established.And the tracking performance of the phase discriminator output based centralized dual-vector tracking loop for typical carrier phase inputs is analyzed in the complex domain.Through applying additive quaternion and the federated dual-vector tracking loop based on phase discriminator output,a deep GNSS/MEMS INS integration scheme on the basis of discontinuous GNSS signal tracking is designed for the purpose of mitigating the heavy computational burden of traditional deep integration approaches.The designed deep integration scheme intermittently collects and tracks the GNSS signals,which thereby facilitates the realization of enduring positioning and navigation on micro navigation devices with poor computing capability and small battery capacity.In the safety or liability-critical field,the integrity of navigation system is particularly important.Most of current RAIM(Receiver Autonomous Integrity Monitoring)algorithms are developed on the basis of LS(Least Squares)method or SS(Solution Separation)method.In this paper,a calculation method is presented to improve the protection level accuracy of WLS(Weighted Least Squares)based RAIM algorithm,which converts the integrity risk into the product of two cumulative distribution functions of chi-square distribution and uses a designed algorithm for solving the extreme problem under quadratic constraint to calculate protection level in the feasible domain.The deep GNSS/MEMS INS integration is susceptible to a variety of sources of error and fault.Following the analysis of the navigation solution error propagation characteristic of the deep GNSS/MEMS INS integration with phase discriminator output based federated dual-vector tracking loop,an autonomous integrity monitoring algorithm for the deep GNSS/MEMS INS integration is then proposed.In which,the test statistic for fault detection is constructed using the code phase errors from all valid receiver channel pre-filters and the protection level calculation is formatted as the extreme problem under quadratic constraint.The proposed integrity monitoring algorithm can monitor multiple GNSS faults,MEMS INS fault and integration fault.In order to verify the effectiveness and performance of above proposed algorithms with respect to deep integration,a deep GPS/INS integration software based on Matlab is developed,which can navigate with the given GPS IF(Intermediate Frequency)data and IMU data.And a method of synchronizing GNSS data with IMU data is also presented.With the developed software,the proposed algorithms are verified and analyzed by using the navigation data generated by semi-physical simulation and the real navigation data collected in vehicle based experiment.The test results show that: the designed deep GNSS/INS integration can stably track the GNSS signals and obtain accurate navigation solution in high dynamic scenario;the discontinuous GNSS signal tracking based deep GNSS/MEMS INS integration scheme can stably navigate in low dynamic scenario;the proposed integrity monitoring algorithm can effectively detect GNSS fault and integration fault,and system's position error stays within the calculated protection level when no fault is detected.