关键词: 端点效应, 经验模式分解, 延拓, 正弦函数, 指数函数

Abstract: The end effect of empirical mode decomposition that appears when using spline interpolation to get two envelops of the data is discussed. The shortage of ordinary end extending methods is pointed out. An extending method based on trigonometric functions is improved, and the influence of different extending periods, length and value of end point of the data is discussed. A comparison with mirror extending method is made and analyzed. On the basis of the improved method, a new method with exponent sine function extending approach is proposed, which decreases the envelop emanation of end points of extending data, so it can improve the speed and precision of EMD, and simulation is used to prove the effectiveness. Results show that the proposed improved extending method can restrain the end effect effectively.

Key words: Empirical mode decomposition, End effect, Exponent function, Extending, Sine function