Download files
Abstract
Integration of Global Navigation Satellite Systems (GNSS) with Inertial Navigation System (INS) has been considered as the best technique that can provide high accuracy and reliable positioning and navigation results. This is because the characteristics of errors in both sensors complement each other. Despite significant progress over the last two decades, it is necessary to ensure that the navigation solution provided by such integration meets the requirements of the navigation system performance particularly in terms of accuracy and integrity. While various quality control techniques (e.g., extended Receiver Autonomous Integrity Monitoring procedures) have been adopted in integrated navigation systems, their performances are limited to detect single fault at a time. However, this is not always the case because the system may include more than one failure simultaneously. As a result, the need of integrity monitoring systems that provide satisfaction performance should take the highest priority.
The aim of this research is to analyze reliability and integrity monitoring under the presence of multiple faults in different GNSS/INS integration modes. In this research four main GNSS systems, integrated multi constellations and different GNSS observations including pseudorange and carrier phase and hence different GNSS/INS integration scenarios have been investigated with respect to the quality control procedures. More specifically, the contributions of this research are:
a) Adaptive Kalman filtering
As Kalman filtering is normally used in integration GNSS/INS systems, the integration filter should be optimal in order to get optimal integration. Optimal integration filter requires realistic assumption of the measurement model and system covariance matrices R and Q respectively otherwise the filter will be suboptimal or sometimes diverge. This thesis has evaluated the performances of adaptive Kalman filtering methods in integrated navigation systems. This was through comparing adaptive Kalman filtering methods based on covariance analysis, and innovation and residual analysis. In addition, the Influence of moving window sizes on the performance of adaptive filtering has also been investigated. This allows selecting realistic values of both Q and R that give the best stable state estimates.
b) Multi-faults detection and identification
It is necessary to evaluate the capability of the fault test in terms of detection and identification of single and multiple faults in integrated GNSS/INS systems with different types of GNSS observations. In this research two GNSS/INS integration scenarios based on satellite pseudorange and carrier phase with respect to multi-faults detection and identification in measurement and dynamic models have been investigated. Besides, the internal reliability measures have been employed to measure the capability of the system to detect and identify faults in both GNSS/INS models. This leads to highlight the problems that are associated with detection and identification of multiple faults in measurement and dynamic models.
c) Integrity and reliability of integrated GNSS/INS systems
Horizontal and vertical protection levels for single and multiple faults as a function to the number of visible satellites and the satellites geometry have been analyzed in GNSS and integrated GNSS/INS systems. In addition, the influences of the dynamics and vehicle trajectory on the reliability and integrity levels have also been investigated.
d) Analysis of separability of multiple simultaneous faults
The influence of number of redundant satellites, satellites geometry and different structures of dynamic on the capability of the system to detect and identify faults in integrated GNSS/INS systems has been analyzed. In addition, multi-dimensional correlation coefficients between fault detection statistics have been used as an indicator of the capability of the system to separate multiple faults.