摘要

Since the states of the system are not totally observable, particle filter cannot collect adequate measurement information as demanded while marginalized particle filter is applied to stationary self-alignment with large misalignment angles. To solve this problem, a filtering model of initial alignment with large misalignment angles is constructed. A marginalized particle filtering method assisted by the gyro measurement information is proposed. The simulation results show that the proposed approach can greatly lighten the heavy computational load with high dimensions and large initial errors, quicken the alignment speed of the azimuth misalignment angle, and get higher initial alignment precision than the conventional method with ordinary particle filter.