中文摘要：

为了减少梳状音叉微机电系统（MEMS）陀螺的随机漂移误差,提出了一种小波域上的Karhunen-Loeve变换（KLT）的MEMS陀螺漂移信号的去噪方法。其主要思想是：先对分段的MEMS陀螺漂移信号进行小波分解;然后对各个中高频子带进行6抽头滤波,插值成和最高频带相同长度的样本点后,利用小波各尺度间的相似性进行高频分量的KLT变换,在一定程度上去除不相干噪声;最后对KLT降噪后的信号再进行小波阈值处理完成进一步的降噪。实验结果表明,所提方法相对于基于小波变换的各种阈值方法,陀螺输出信号的方差、零偏稳定性和随机游走误差都有了明显的改善。

英文摘要：

For effectively reducing the random drift error in micro-electro-mechanical system（MEMS） gyro signals,a novel denoising method is presented,namely the Karhunen-Loeve transformation（KLT） in the wavelet domain.The wavelet decomposition was performed on the segmented MEMS gyro signals;the 6-tap filter was applied to the corresponding medium and high frequency components;the KLT was carried out by the similarity among the wavelet scales for the interpolated high-frequency components to eliminate the extraneous noise;the threshold processing was further implemented on the KLT-denoised signal for better denoising.The calculated results show that the proposed denoising method can make variance,null deflection stability and random drift error improved obviously in comparison with the wavelet threshold based denoising methods.