Initial Alignment Method of Human Soft Strapdown Base Based on NKF-FRKF

Abstract

Aiming at the problem of inaccurate statistical characteristics of system model noise within initial alignment of inertial pedestrian navigation, a novel heterogeneous hybrid correlation entropy Kalman filter (NHMCC-KF) alignment method is proposed considering inertial devices and human soft shortcuts. Firstly, the lever arm error is expanded as a state quantity to establish the initial alignment model of the inertial coordinate system. Then, on this basis, the covariance of the measurement noise is adjusted by combining the fast robust Kalman filter (FRKF) and the heterogenous mixture correntropy criterion, using a mixture of Laplace kernel and Gaussian kernel as the adjustment factor of the correntropy, and introducing the prior error covariance feedback adaptive Kalman filtering (NKF) to adjust the process noise. By comparing the alignment effect of NHMCC-KF and FRKF under different conditions through designed experiments, the azimuthal alignment accuracy is improved by more than 23%. The experimental findings demonstrate that the approach described in this research has superior alignment accuracy and speed.