非对称冗余惯导的在线标定与容错估计反馈策略

To guarantee the reliability of inertial navigation information, an asymmetric redundant inertial navigation system is designed under the condition of limited size and cost, including two sets of inertial navigation systems, FOG-INS and MEMS-INS. In normal case, the FOG-INS outputs navigation information and calibrates the MEMS inertial device errors online using sequential variational Bayesian Kalman filter. The exponentially weighted convergence degree of each estimation is mapped into a unified closed interval for standardized evaluation. When the FOG-INS fault is diagnosed, the estimation feedback strategy will compensate the errors of the MEMS inertial device based on the standardized evaluation result, and the corrected MEMS-INS will output the navigation information. The simulation results show that after a detection delay of 0.8 seconds, the gyro drift estimations of y-axis and z-axis of MEMS-INS based on variational Bayesian Kalman filter after the fault are improved by 80.80% and 67.76% respectively compared with Kalman filter, and the horizontal position error of corrected MEMS-INS after 60 seconds of failure is reduced by 33.59% compared with the non-corrected situation, which effectively ensures the reliability of the inertial navigation information.