Geometric Analysis and Optimization for FDI Performance of Redundant Inertial Sensors

Abstract

This thesis suggests optimal configurations for redundant inertial sensors with analysis of geometric parameters with respect to Fault Detection and Identification (FDI). To define FDI performance of each configuration, a performance index for FDI method based on Parity Space Approach (PSA) is applied. Even though this index is dependent on the geometry of sensor configurations, however, it is hard to analyze the performance index directly since it is expressed in the null space of Direction Cosine Matrix (DCM) for the configurations. To solve this limitation, a modified form of the FDI performance index is presented as a function of geometric parameter of the configurations. It makes the FDI performance analysis and optimization of the configurations much easier. Additionally, the optimizations of configurations such as platonic solids, single cones and dual cones are conducted by the modified performance index. Finally, the FDI performance of each configuration is compared with others by the FDI performance index. The comparison result shows that the optimized dual conic configurations achieve FDI performance superior to the one of other configurations. The same results are also confirmed by simulations and experiments on each configuration.