Fuzzy Adaptive Interacting Multiple Model Algorithm for INS/GPS
Tianlai Xu, Cui Pingyuan · 2007
The integration of INS and GPS is usually achieved using a Kalman filter. The precision of INS/GPS system will be reduced in condition that a priori information used in Kalman filter does not accord with the actual environmental conditions. The problem of INS/GPS navigation system with uncertain noise is considered in this paper. Fuzzy adaptive Kalman filtering algorithm (FAKF) and adaptive interacting multiple model algorithm (AIMM) is combined, named FAIMM, to address this problem. In each cycle of FAIMM, FAKF is used firstly to determine rough statistical characteristics of noise, then the AIMM algorithm completes the integration of INS/GPS data, using a limited number of subfilters formed according to the rough values obtained from the FAKF. Simulations in INS/GPS integrated navigation system demonstrate that the FAIMM algorithm can obtain better statistical estimation of noise and provide better coverage of variable noise statistical characteristics than IMM when environmental conditions change, and the accuracy is improved compared with either Kalman filter or IMM algorithms.