利用时频分析法精化GNSS/INS组合导航先验滤波模型

摘要

准确的先验信息对于经典Kalman滤波和自适应滤波都非常重要.应用Kalman滤波时,先验模型的确定往往具有一定的随意性和较强的经验性,缺乏理论基础.以GNSS/INS组合导航为例,利用时频特性分析的方法由惯性元件数据提取出误差的随机特性,得到较为准确的先验信息,避免了实际应用Kalman滤波时繁琐的调试.%Exact prior information is of great importance to both classical Kalman filtering and adaptive filtering. In applications concerned with Kalman filtering, prior model is generally determined arbitrarily and experien-tially, lacking methodology. Taking the GNSS/INS integrated navigation as an example,the stochastic character of errors, which is subsequently transformed into proper prior information, is obtained from inertial data based on time-frequency analysis. This method avoids the complicated tuning in utilizing Kalman filtering.