Fault tolerant inertial navigation system
Kevin Vanderwerf, KNUT WEFALD · 1988
This paper describes the sensor redundancy management concept employed in a fault tolerant inertial reference system built by Honeywell's Commercial Flight Systems Minneapolis Operation to be flight tested in 1988. vided. No attempt was made to use state of the art packaging. Existing circuits were used wherever feasible, and low voltage power is provided using external commercial power supplies. The hexad IRU is shown in figurel. The concept uses a sensor hexad and employs separate fault detection and isolation (FDI) algorithms for navigation computations and outputs to the flight control system. The flight control FDI employs large thresholds and short time constants consistent with flight control requirements. In order to achieve the low detection levels required to maintain navigation accuracy, the navigation FDI employs heavy filtering, low thresholds, Kalman filtering to correct the parity residuals for sensor misalignment and scale factor errors, and dynamic parity gains to minimize the effects of maneuvers. Two redundant nav solutions using sub-optimal sensor sets are also included to reset the primary solution after failures. The FDI algorithm implementation is based on the generalized likelihood ratio test (GLRT) that has been the subject of several papers in recent years. Enhancements to the failure isolation logic virtually eliminate the probabilities of false detection and wrong isolation inherent with the basic GLRT concept, thus making the GLRT scheme practical in the real world of high reliability requirements.