New decoupling matrix method for generalized likelihood fault detection and isolation of redundant IMU

Integrating multiple inertial sensors by using appropriate redundant configurations can greatly enhance the guidance system's reliability. Failure of the inertial device will pollute the navigation information, which needs to be detected and isolated. The existing generalized likelihood test (GLT) method with Potter decoupling matrix constructed cannot detect and isolate the faults of several strapdown inertial measurement units along a specific axial direction, because Potter method does not satisfy linear correlation conditions. This paper proposes construction of the decoupling matrix by selecting the maximal linearly independent systems of the orthogonal projection array and then orthogonalizing and unitizing the maximal uncorrelated set. Numerical simulation performs of several configurations of redundant strapdown inertial measurement unit to verify the fault detection and isolation effects of the improved generalized likelihood test method. The simulation results show that the improved generalized likelihood test method can detect and isolate the faults of several configurations of redundant strapdown inertial measurement unit. This method provides a new idea for the on-line fault monitoring technology of the launch vehicle guidance system.