De-Noising Signal of The Quartz Flexural Accelerometer by Multiwavelet Shrinkage
De-Noising Signal of The Quartz Flexural Accelerometer by Multiwavelet Shrinkage
复制标题
DOI：
10.21307/ijssis-2017-535
复制
发表时间：
2013-01
影响因子：
1.2
通讯作者：
Chihang Zhao;Zhong Xin;Dang Qian;Zhao Li-ye
中科院分区：
文献类型：
--
作者：
Chihang Zhao;Zhong Xin;Dang Qian;Zhao Li-ye
摘要 对基于五种多小波的JSD-I/A石英弯曲加速度计输出信号去噪方法进行了比较研究。首先，分析了多小波变换的理论和广义交叉验证准则。其次，由于通过专用夹具固定在SCT-1两轴旋转平台上的JSD-I/A石英弯曲加速度计的启动过程为3分钟，因此在施加电压5分钟后对石英弯曲加速度计的输出信号进行采样。再次，基于软阈值函数和通用阈值，分别应用GHM正交多小波、SA4正交多小波、CL正交多小波、Cardbal2平衡多小波和BIGHM双正交多小波对4级分解级JSD-I/A石英弯曲加速度计的采样信号进行去噪。最后利用广义交叉验证准则评价上述5种多小波的去噪效果。实验结果表明，BIGHM双正交多小波的广义交叉验证值能够有效地对JSD-I/A石英弯曲加速度计输出信号进行去噪，并且比GHM正交多小波、SA4正交多小波、CL正交多小波和Cardbal2平衡多小波具有最佳的性能。
Abstract Methods of de-noising the output signal of the JSD-I/A quartz flexural accelerometer based on five types of multiwavelets are comparatively investigated in this paper. Firstly, the theory of multiwavelet transform and the generalized cross validation criterion are analyzed. Secondly, because the JSD-I/A quartz flexural accelerometer which is fixed in SCT-1 two-axis rotation platform by the appropriative clamp has a start-up procedure of 3 minutes, the output signal of the quartz flexural accelerometer are sampled after applying the voltage for 5 minutes. Thirdly, based on the soft threshold function and the universal threshold, GHM orthogonal multiwavelet, SA4 orthogonal multiwavelet, CL orthogonal multiwavelet, Cardbal2 balanced multiwavelet and BIGHM biorthogonal multiwavelet are applied to de-noise the sampled signal of the JSD-I/A quartz flexural accelerometer with 4 decomposition level, respectively. Lastly, the generalized cross validation criterion is used to evaluate the de-noising effects of the above five multiwavelets. The experimental results shows that the generalized cross validation value of BIGHM biorthogonal multiwavelet is effective in de-nosing the the output signal of the JSD-I/A quartz flexural accelerometer, and offer the best performance than GHM orthogonal multiwavelet, SA4 orthogonal multiwavelet, CL orthogonal multiwavelet and Cardbal2 balanced multiwavelet.