A Modified GLT Double Faults Isolation Approach Based on MLE and RPV for Six-Gyro Redundant SINS

In this paper, a modified generalized likelihood test (GLT) double faults isolation approach based on maximum likelihood estimation (MLE) and reduced-order parity vector (RPV) has been proposed. To improve the reliability of the strapdown inertial navigation system (SINS), redundant inertial sensors (gyros and accelerometers) are applied to SINS, which is called redundant SINS. The six-gyro SINS has been widely used because of its high reliability, low cost, and small volume. Using fault isolation, faulty sensors can be isolated when they are faulty in redundant SINS. The GLT double faults isolation approach based on MLE effectively isolates double faults due to its high sensitivity, small calculation, and easy implementation. However, this approach cannot correctly recognize detailed double faults in six-gyro SINS. Additionally, this approach cannot work during SINS motion. In this paper, first, based on MLE, double faults isolation function and faults isolation threshold are redesigned, and then, the application range of double faults isolation function is analyzed. Second, a modified RPV isolation approach is proposed to correctly recognize the detailed double faults. Third, the Kalman filter is employed to compensate the dynamic errors of sensors, making a new approach applicable to dynamic application. The simulation shows that the new approach is able to isolate double faults effectively for six-gyro SINS during SINS motion.