为降低MEMS陀螺仪输出信号中低频噪声提高信号精度,提出一种采用基于EMD和分形高斯噪声的滤波方法.陀螺仪输出的横摆角速度信号使用滑动窗口法,对窗口数据进行聚合方差法估计Hurst参数,并通过EMD分解窗口数据获得各层IMF分量及余项,计算窗口阈值并进行阈值处理选择,逐步处理滑动窗口数据,将处理后的IMF分量和余项整合,得出滤波后的信号数据.通过仿真实验验证及实车数据验证,证明滤波方法对信号噪声精度提高的可行性.
In order to reduce the low-frequency noise in the output signal of MEMS accelerometer and improve the signal accuracy,a filtering method based on EMD and fractal Gaussian noise is proposed.The yaw rate signal output by the accelerometer uses the sliding window method to estimate Hurst parameters from the window data using the aggregate variance method,and obtains the IMF components and residuals of each layer through the EMD decomposition window data,calculates the window threshold and conducts threshold processing selection,gradually processes the sliding window data,integrates the processed IMF components and residuals,and obtains the filtered signal data.The feasibility of the filtering method to improve the precision of signal noise is proved by simulation experiment and real vehicle data verification.