摘要
惯性传感器信号的去噪效果通常都是用Allan方差来评价,针对这个特点,提出了一种基于Allan方差噪声系数的小波阈值改进算法。先用提升小波将信号分解到不同频段,再用Allan方差计算各频带上的噪声系数,根据各频带噪声系数的大小确定其滤波阈值,再结合参数可调的阈值函数对各尺度信号进行滤波去噪。改进算法以去噪效果评价参数来修正阈值,具有很强的针对性。通过同其它常用的惯性传感器去噪方法做比较,证明该算法可以有效的随机噪声,提取出更加准确真实的惯性传感器动态输出。
The noise reduction effect of inertial sensor is usually evaluated by the Allan variance. Therefore,an im-proved wavelet threshold algorithm based on Allan variance noise coefficients is proposed. Firstly,the signal is de-composed into different frequency bands by lifting wavelet transform. And then the Allan variance noise coefficientsof are calculated,their values are used in determining the filtering threshold. The improved algorithm has strong per-tinence,because the evaluation parameters of the de-noising effect are used to modify the threshold. Compare to oth-er inertial sensor de-noising methods. The proposed method can effectively suppress random noise to extract moreaccurate inertial sensor dynamic output.
出处
《传感技术学报》
CAS
CSCD
北大核心
2016年第11期1718-1723,共6页
Chinese Journal of Sensors and Actuators
基金
国家自然科学基金项目(61503390)
