Trend Extraction in Vibration Signal Based on EMD

Abstract

This paper proposed a de-trend method for vibration signal of telemetry based on the empirical mode decomposition (EMD) by correlation coefficient matrix. The signal is decomposed to a series of intrinsic mode component and the remainder item by EMD. It mainly distinguishes between the remainder item and the signal trend, according to the correlation coefficient matrix to determine whether some intrinsic mode component belongs to the trend item or not. The results show that signal trends can be extracted accurately through the effective combination EMD with correlation coefficient matrix and the proposed method has good applicability for different signals and different trends.

A trend extraction method based on logistic functions and envelopes

“…Some methods have been studied in trend extraction, such as the five-spot triple smoothing method 3 , least-square method 4 , wavelet transform method 5 and empirical mode decompositions (EMD) 6 . The wavelet transform method is a commonly used signal-processing method in the frequency domain.…”