Abstract: In order to improve the dynamic error coefficient speed of the calibration strapdown inertial group, a dynamic error model is established in the inertial navigation system. The Kalman filter is used to reflect the observation degree of the error term to determine the objective function of the optimal algorithm. The penalty function for the reduced space search algorithm is improved by genetic algorithm. At the end, it not only gets rid of the interference of the local pseudo-optimal solution, but also improves the recognition speed when it is close to the optimal solution. Compared with the solution length of the conjugate gradient method, the efficiency of the improved reduced space search algorithm for dynamic coefficient identification is verified, and the problem that the dynamic error of the strapdown inertial group cannot be quickly compensated is effectively solved.

Key words: stapdown inertial navigation systems, conjugate gradient algorithm, subtractive space search algorithm, dynamic error calibration