부산대학교 도서관

This paper focuses on the self-alignment problem for the strapdown inertial navigation system (SINS). The traditional self-alignment methods are vulnerable to the random noise of the inertial sensors. To solve this problem, an improved self-alignment method based on vector observation and truncated vectorized K-matrix is proposed. There are three innovations. First, this algorithm overcomes the shortcomings that the Optimal-REQUEST algorithm estimates the scalar gain and performance index conservatively. Second, considering the zero-trace and symmetry of K-matrix, the dimension of state-space is reduced from sixteen down to nine, which simplifies the calculation of self-alignment compared with the filter of the direction cosine matrix (DCM). Third, the proposed method combines the advantages of Kalman filter and Wahba problem to estimate the optimal quaternion. Therefore, the proposed method can effectively reduce the influence of measurement errors in the self-alignment process. The practical experiment results prove that the proposed method greatly improves the accuracy and speed of self-alignment in comparison with the traditional algorithm.