摘要
为了能有效地补偿MEMS(微电子机械系统)陀螺仪的随机漂移,提高载体姿态估计的精度,基于小波理论与多尺度分析方法,使用db4小波,将MEMS陀螺仪随机漂移进行深度为4的多尺度分解,得到5组小波系数。根据分解后的各尺度系数进行信号重建,得到5组多尺度陀螺仪漂移数据。对重建后的各尺度漂移数据进行时间序列建模,可以得到MEMS陀螺仪随机漂移的多尺度时间序列模型。在多尺度时间序列模型的基础之上,建立多尺度离散系统的系统模型,使用卡尔曼滤波方法,对个尺度陀螺随机噪声进行滤波,可以有效地滤除MEMS陀螺仪的随机漂移。试验结果表明本方法能有效降低信噪比。
Wavelet theory and multiscale analysis method were adopted to compensate the random drift from MEMS（Micro Electro Mechanical System） gyros. In order to estimate the carrier＇s attitudes more accurately, the drift of MEMS gyros were decomposed with depth of 4 using db4 wavelet, and 5 groups of wavelet data were derived. On each scale, the signal was rebuilt, so 5 groups of multiscale gyros drift data were got. After modeling the data, the multiscale timeseries model of MEMS gyros random drift could be built. A dispersed system model was built based on the multicale timeseries model. Finally a Kalman filter was applied to filter the gyros random drift from each scale. The result shows that the SNR is reduced by the method.
出处
《中国惯性技术学报》
EI
CSCD
2007年第2期229-232,共4页
Journal of Chinese Inertial Technology
作者简介
赵世峰（1979-），男，博士生，研究方向为微小型飞行器导航与控制。E-mail：zhaoshifeng@asee．buaa．edu．cn
范耀祖（1936-），男，教授，博导。E-mail：fanyuezu@126．com
