General method for empirical data decomposition filtering design

Abstract

Based on the research on Empirical Data Decomposition (EDD), the structure for Empirical Data Decomposition is proposed in which the high pass filter is composed of a predictor and an adder. In terms of reconstructing requirement, the filter design rule is presented when the EDD analysis filter and synthesis filter are restricted as FIR filter. The relationship between equivalent synthesis filter and analysis filter is also presented. Finally the structure for the synthesis filter is discussed. Except for the FIR requirement, there is no additional restriction for the predictor, so the filter can be easily designed to satisfy different requirements. EDD is suitable not only for stationary data analysis or piece-wise stationary data analysis but also for non-stationary data analysis.