Abstract:

The precision of Kalman filter is high, but it is not robust to unknown disturbance. The robustness of H∞ filter is good, but the precision is low. Hybrid Kalman/H∞ filter can obtain both robustness to noise and high precision. By evaluating the performance of Kalman filter, an adaptive hybrid Kalman/H∞ filter based on nonlinear mapping is proposed. Simulations are performed based on global positioning system/dead reckoning integrated navigation model to confirm the performance of the proposed method. As can be seen from simulation results, when statistics of the noise change and system model perturbation exists, the precision of the proposed hybrid Kalman/H∞ filter is improved as compared with Kalman filter and H∞ filter， thus it is more suitable for practical applications.