Trend extraction of gyro's drift based on modified empirical mode decomposition

Nonlinear and nonstationary time series are decomposed into a series of instrinsic mode functions and a residual trend item by the empirical mode decomposition (EMD). A modified EMD method to extract the trend item of Gyro's drifts is introduced. The method modifies local mean estimation method of EMD by the two-tap adaptive time-varying filter, and restrains end effects using the adaptive border estimation. The concrete steps of the proposed method are presented and applied to a Gyro's drift time series. The experiment result indicates this method is efficacious.