Inhibition Of The End Effect Of Emd Method

Hilbert-Huang transform is proposed as a new signal processing method in recent years, which gradually decomposes the real existence of fluctuations in different scales or trends out of the time signal via Empirical Mode Decomposition (EMD), to produce a series of intrinsic mode function (IMF) with different characteristics. Each IMF component can grasp the characteristics of the original data information more accurately and effectively, so EMD method is an adaptive time-frequency localization analysis method. However, there are many issues that must be resolved due to the method proposed is not long, especially the end effects of EMD.The end effects of EMD process is a kind if error which is caused by incomplete envelope because the signal at the endpoint is not extreme. Such error would be spread inward as the IMF gradually decomposes, further, pollution of the entire data series would be caused to make the results meaningless. The main tasks of this article include:(1) It has analyzed the causes on the end effects of EMD and found the most effective way to inhibit the end effects of EMD is of carrying out a more appropriate method of interpolation during the endpoint extension.(2) It has studied the shortcomings of the mirror continuation method and the slope-based method that can inhibit the end effects of EMD to a certain extent in other literatures. Based on such flaws, this paper has been classified the endpoints into four classes, and the previous two methods have been improved. Then, it has proposed the second slope-based method and further suggestions for changes. At last, the new idea of combining the mirror method and the improved first slope-based method as well as the improved second slope-based method separately to inhibit the end effects of EMD has been verified through the simulation results.(3) It has conducted the Ort and the instantaneous frequency as the standard to evaluate different methods for the characteristics of inhibiting end effects of EMD, and it has validated the Validity of the proposed method in this paper through different simulation experiments on different sampling data with different type endpoints. (4) It has obtained satisfactory results when the new method proposed in this paper was applied to SOI data.