An Extended $H_{i}/H_\infty $ Optimization Approach to Fault Detection of INS/GPS-Integrated System

This paper deals with the problem of robust fault detection (FD) for an integrated system of inertial navigation system and global position system (INS/GPS). Under the assumption of process disturbance and measurement noise being l
₂
-norm bounded, a state-space model with potential additive fault is first introduced to describe the error dynamics of the INS/GPS integrated system. For the purpose of the FD, an observerbased FD filter (FDF) is used as a residual generator and the design of the FDF is formulated so as to find an observer gain matrix and a postfilter in the framework of H
_i
/H
_∞
optimization. By utilizing the techniques of innovation analysis and orthogonal projection, a solution to the extended H
_i
/H
_∞
-FDF is given in terms of Riccati recursions. Furthermore, a strategy of residual evaluation is also considered, which concerns the selection of an evaluation function and a threshold. Finally, a flight experiment is implemented to show the effectiveness of the proposed method.