摘要
光纤陀螺输出信号中主要含有白噪声和具有非平稳、长程相关、自相似性及具有1/fγ类型谱密度的分形噪声,传统的滤波方法难以去除这类噪声,小波分析的多分辨分析特性,使之成为分析分形噪声的有力工具.文中分析了光纤陀螺输出信号误差特性,建立了输出信号的数学模型,在分析了分形噪声的特性的基础上,采用小波变换方法生成分形噪声.在实验室环境下采集了某型号光纤陀螺仪输出信号,对其进行多种小波滤波方法滤波对比实验,并进行了实验结果分析,实验结果表明采用平移不变离散小波变换半软阈值滤波方法可以有效地去除分形噪声和白噪声,克服了硬阈值法有较大的方差和软阈值法有较大的偏差的缺点.
In a FOG (Fiber Optic Gyro), there mainly exist white noises and fractal noises characterized by non-stationary, long-term dependence, self-similarity and 1/fγ spectrum density. It is difficult to eliminate all FOG noises by using traditional filtering methods. Wavelet analysis is effective in dealing with the FOG noises for its multi-resolution analysis. The FOG error characteristics and mathematic model were given, the characteristics of fractal noises were analyzed and fractal noises were generated in the wavelet transform method. FOG signals were acquired and filtered with many wavelet filtering methods in a lab environment. TIDWT (Translation Invariant Discrete Wavelet Transform) filter based on semi-soft threshold is effective in removing the noises existing in FOG proved by the simulation experiments, overcoming the big variance of hard threshold method and big biases of soft-threshold method.
出处
《传感技术学报》
CAS
CSCD
北大核心
2007年第10期2276-2279,共4页
Chinese Journal of Sensors and Actuators
基金
"十一五"总装备部预研项目资助(51309020503)
国防973项目资助(973-61334)
国家自然科学基金项目资助(50575042)
教育部博士点专项科研基金项目资助(20050286026)
