Research On SINS/GPS And SINS/CNS Integrated Navigation Systems

This dissertation investigates the loosely and tightly coupled SINS/GPS integration architectures and SINS aided by small and large field of view star sensors.Stand-alone SINS cannot provide accurate navigation solution for long time duration without external aiding,especially when utilizing the low cost MEMS based inertial sensors.GPS offers promising low cost aiding for SINS.In environment with four or more active and visible satellites,the loosely coupled model provides accurate and continuous time navigation solutions.However,in difficult GPS areas such as tunnels,tall buildings,urban canyon,and forest canopy,a tightly coupled integration scheme outperforms and bounds the SINS errors to acceptable accuracy level.Therefore,mathematical modeling and comparison of loosely and tightly coupled models for different GPS environments is one of the main objectives of this dissertation.Lever arm effect and GPS clock errors are compensated to improve the accuracy of the integrated systems.Practically,in SINS/GPS integration,the system noises are not known correctly.Therefore,an Adaptive Kalman filter is proposed in this study to further improve the performance of the integration models.A field test is conducted to collect the real data from MEMS-IMU(STIM300)and an inexpensive single frequency NovAtel GPS receiver.The test systems are installed on a land navigation vehicle.A reference system(POSLV)is also mounted on the same vehicle for collecting the reference data for comparison purpose.The data from STIM300 and GPS is fused through SKF and AKF and compared with the reference data.The integration results demonstrate the performance,and effectiveness of the two models for compensation of low cost MEMS based SINS errors in different GPS environments.The second major objective of this dissertation is to design and investigate the SINS/CNS integration algorithms based on the measurements from small and large field of view star sensors.In low altitude flights such as UAVs,guided missiles and aviation aircrafts,the small field of view star tracker outperforms.In this study,the strapdown configuration is adopted and the integrated algorithms for both CNS systems are designed in a local level navigation frame.For small FOV star tracker,the azimuth,elevation and platform position error angles are utilized to develop the integrated model.For extended time flights mechanized in a local level navigation frame,the gyro and accelerometer errors make extensive contribution towards the velocity and position errors and gyro errors contributes more than accelerometer errors.The traditional SINS/CNS algorithms are not efficient to provide accurate positioning in navigation frame in long time flights.Therefore,some non-traditional algorithms are required to design for compensating the velocity and position errors.Consequently,thisdissertation aims to design effective SINS/CNS algorithms that can compensate and correct for the attitude and unbounded velocity and position errors through measurements from small and large FOV star sensors.A trajectory simulator is designed for generating flight data with different flight actions and maneuvers.A small star catalog of 50 bright stars is also developed utilizing the sky2000 online star database.The star catalog provides the data in J2000.0 coordinate system which needs to be transformed in to the local level azimuth and elevations angles.Therefore,one more novel algorithm is proposed and analyzed in this study for precise transformation of star coordinates between J2000.0 and WGS84 utilizing SOFA functions and subroutines.Simulation results are produced to demonstrate the integrity and validity of the proposed integrated methods for extended time flights.Position accuracy of 54 m in latitude and 26 m in longitude is achieved with one star observation.With two stars observations,the position is estimated with precision of 23 m in latitude and 21 m in longitude.The integration results based on a psi-angle measurement are more accurate.It estimates position with precision of21 m in latitude and 16 m in longitude.The results of SINS/CNS using large FOV star sensor measurements are corrected to 33 m and 22 m in latitude and longitude.Hence,the designed SINS/CNS integrated systems are validated and qualified to employ for reliable and accurate long duration autonomous navigation missions without GPS aiding.