以SINSiGPS组合导航系统为背景,在对Kalman滤波原理和工程应用进行深入分析的基础上,总结了该方法的不足,提出了应用神经网络和模糊推理技术对系统噪声、观测噪声和其相关阵进行直接调控的方法。该方法根据新息和新息方差的变化,实时调整自适应因子,间接改变Kalman滤波器的当前观测量和过去信息的比例关系。仿真结果表明,该算法对模型和噪声干扰有较强的自适应性,能够有效抑制滤波发散,在不损失原有精度的前提下,提高了系统的鲁棒性。
This paper, which bases on GPS/SINS integrated navigation, analysis the theory of Kalman filtering and concludes its shortcomings. It brings forward the method to control the related matrix about system model and noise model. This method uses fuzzy neural network and it adjusts the adaptive gene directly according to the variety and variance of new-information, and at them same time, it adjusts the proportion to use current observation and last message indirectly. Simulation results indicates that the adaptive Kalman filtering algorithm bases on fuzzy controller can adapt to the disturb of system and noise model, restrain filtering failure, improve the system robustness.