Abstract: The non-orthogonal angle of the single-axis rotational inertial navigation system can cause the attitude angle errors in horizontal plane. Aiming at the shortages that traditional least square method would bring out large approximation error when estimating the non-orthogonal angle of rotational axis and doesn’t consider the data’s statistic characteristics, a new correction method based on Kalman filter is introduced. On the basis of the traditional error model, the proposed method adds the accelerometer output as the observation, and then estimates the rotating-axis non-orthogonal angle and corrects the rotating-axis non-orthogonal error. Simulations results demonstrate that, with the new correction method, the horizontal attitude angles are reduced to 102 from 502, and the accuracy is improved by about 80%, which show that the precision of horizontal attitude angle is improved by the proposed method.

Key words: rotating modulation, non-orthogonal angle, least square, Kalman filter