中文摘要：

根据误差信号的频谱特性将MIMU（Micro Inertial Measurement Unit）误差信号分为长期误差和短期误差两类，分析了两类误差的来源和传统组合导航系统的降噪性能。利用Allan方差法对MIMU长期误差进行分析建模；针对短期误差的频谱特性，使用最优低通滤波器滤除高于运动特性频带上限的部分；通过小波阈值降噪方法减小混叠在运动特性频带中的短期误差。GPS／MIMU组合导航仿真结果表明，经过MIMU信号预处理的组合系统，在GPS信号有效时定位误差减小了20％，而当GPS信号丢失时，累积误差降低了近80％。

英文摘要：

On the time-frequency view, the errors of MIMU signal were classified into long-term errors and short-term errors according to their spectral signature, and the error resources and performance of de-noising in GPS/MIMU integrated system were analyzed. Allan variance method was used to model and compersate the long-term errors, and the short term errors whose frequency higher than the up band of motion dynamic was eliminated by optimal low pass filter, the wavelet threshold filter was used to reduce the remained short-term errors. Simulation results show that the performance of integrated GPS/MIMU system has been improved by 20% when GPS is available and 80% while GPS blockages after pre processing for MIMU signal.