New SVD method in FDI of redundant IMU

The SVD (singular value decomposition) method was proposed by Shim in FDI (fault detection and isolation), to improve the reliability of Redundant IMU (Inertial Measurement Units) in vehicles. But disadvantages existed in Shim's FDI strategy, such as the lack of normalization, the too large threshold, the wrong isolation strategy, false isolation of a negative fault. Here a new FDI strategy is presented based on Shim's SVD method, which overcomes the defects in Shim's method by using an improved decoupling matrix, and an improved isolation function involving normalization and module value. Monte Carlo numerical simulation results show that the new strategy is better than Shim's, which has higher Probability of Correct Detection and Probability of Correct Isolation. Especially, new strategy has a 99% Probability of Correct Isolation against the negative fault. The new method provides an approach in FDI of Redundant IMU.