Strapdown Inertial Integrated Navigation Algorithm under Short-term Failure of Information Sources

In strapdown inertial integrated navigation, the short-term failure of auxiliary information source may lead to the reduction of integrated navigation accuracy. To solve this problem, a Gaussian process regression aided strapdown inertial navigation algorithm is proposed. In the proposed algorithm, the output speed of strapdown inertial navigation sysytem corrected by integrated navigation closed-loop is used as the training input set of Gaussian process regression, and the speed output of auxiliary information source is used as the training output set. Gaussian process regression is trained whithout failure of auxiliary information source. When the auxiliary information source fails, Gaussian process regression is used to predict the output of auxiliary information source to realize the measurement update of combined filter. In the experiment, the attitude, position and gyro constant drift estimated by the proposed algorithm are compared with those of the traditional algorithm in the case of failure by using the low precision MEMS/GPS integrated vehicle data. The results show that, in the case of 60 s short-term failure, the proposed algorithm improves the effects of pitch, roll and position estimations by about 90%, the effect of yaw estimation by 70.7%, and the effects of east and north gyro drift estimations by 10.3% and 10.8%, respectively. The results show that the proposed algorithm can effectively solve the problem of short-term failure of auxiliary information sources in integrated navigation. © 2020, Editorial Board of Acta Armamentarii. All right reserved.