摘要
为了有效地去除光纤陀螺信号中的分形噪声,提出了全频域小波多尺度阈值降噪方法。从分形信号的分数微分形式出发,建立了分形信号与高斯白噪声之间的联系,结合小波分析的高阶消失矩特性,实现了分形信号在小波空间的去相关。当分形信号被具有高阶消失矩的小波分解后,其在小波域具有白噪声的特点,可以采用阈值处理的方法予以去除。同时,根据陀螺信号噪声的宽频带特点,对小波分解的低频近似系数和高频细节系数进行阈值处理,有效地抑制了噪声成分。光纤陀螺信号的降噪实例表明:相比传统的小波阈值降噪方法、卡尔曼滤波方法和滑动平均滤波方法,该方法具有较好的降噪效果。
In order to remove the fractal noise of fiber optic gyroscope signal, a method of full frequency-domain multi scale wavelet threshold denoising was proposed. Based on the differential form of fractal noise of fiber optic gyroscope signal, the relation between fractal noise and white ganssian noise was revealed. Combing with the high order vanishing moments of wavelet function, decorrelation of fractal noise was achieved in the wavelet domain. When fractal signal was decomposed by wavelet with high-order vanishing moments, the decomposed signal was similar to white gaussian noise and apt to be removed through threshold processing. In addition,according to the wide-band characteristic of the whitened fiber optic gyroscope signal, noise was eliminated through threshold processing to approximation coefficients and detail coefficients. Experiments of fiber optic gyroscope signal denoising demonstrated that the proposed method of better than the classic wavelet threshold deonising, kalman filtering and moving average.
出处
《红外与激光工程》
EI
CSCD
北大核心
2012年第12期3401-3405,共5页
Infrared and Laser Engineering
关键词
全频域降噪
光纤陀螺
分形噪声
小波分析
去相关
full frequency-domain denoising
optic gyroscope
fractal noise
wavelet analysis
decorrelation
