Research on algorithm of adaptive interacting multiple model for integrated navigation system

The interacting multiple model (IMM) estimator has the defect that the more sub-models, the worse real-time performance. An adaptive interacting multiple model (AIMM) algorithm is presented, which combines the IMM algorithm with the simplified Sage-Husa adaptive filtering algorithm. The Sage-Husa adaptive filter is used to estimate the rough value of measurement noise statistical characteristics. The parameters of sub-models are calculated by the rough value and then are fed into the sub-filters in IMM algorithm, where the sub-filters have different parameters of noise. Simulation results of land vehicle integrated navigation show that the AIMM algorithm can achieve the coverage of real situation through a few sub-models, and the accuracy is higher than the IMM algorithm.