| 异常事件对EMD方法的影响及其解决方法研究 |
| |
| 引用本文： | 赵进平.异常事件对EMD方法的影响及其解决方法研究[J].中国海洋大学学报(自然科学版),2001,31(6):805-814. |
| |
| 作者姓名： | 赵进平 |
| |
| 作者单位： | 国家海洋局第二海洋研究所, |
| |
| 基金项目： | 国家攀登计划项目 ( 970 2 31 0 0 3)，国家自然科学基金委员会重点基金项目 ( 4 96 34 1 4 0 )资助，国家海洋局海洋动力过程与卫星海洋学重点实验室科技成果第 2 0 0 1 B0 0 6号 |
| |
| 摘 要： | 作者指出异常事件在数据中形成局部的高频信号 ,运用经验模态分解 (EMD)方法分析这种存在异常事件干扰的数据 ,就会产生本征模函数 (IMF)的频率混叠现象 ,而造成物理过程的重叠 ,使得难以用时间过程曲线表现特定的物理过程。这一问题是 EMD方法中尚未妥善解决的问题。为解决这一问题 ,作者利用干扰信号极值及其两边的极大与极小值位置与原始数据有明显对应关系的特征 ,将相关 IMF中的异常信息直接滤除 ,再用 Spline插值方法弥补滤除时段的数据 ,得到重新拟合的该 IMF数据。采用这种方法可以提取出异常信号 ,提取的精度与异常信号的时段长度有关。而且 ,拟合结果消除了异常干扰 ,可以将该 IMF与其余 IMF一起叠加成没有异常干扰的数据。将滤除了异常干扰的数据再次进行 EMD分解 ,可以得到新的 IMF系列 ,而它与不加校正的分解结果有相当大的差别 ,可靠地反映了真实物理过程。结果表明 ,只有在有效滤除异常干扰的情况下才能获得可靠的 IMF系列 ,并准确地描述各种尺度的现象 ;消除了异常干扰的 IMF可以任意单独或组合使用 ,表现各种时间尺度的变化与过程 ;所讨论的方法只适合异常时段较小的情形。对于异常时段接近或大于正常变化周期的干扰还需要探讨其他方法
|
| 关 键 词： | 经验模态分解(EMD)方法 异常事件 本征模函数(IMF) 高频信号 |
| 文章编号： | 1001-1862(2001)06-805-10 |
| 修稿时间： | 2001年6月29日 |
Study on the Effects of Abnormal Events to Empirical Mode De-composition Method and the Removal Method for Abnormal Signal |
| |
| Zhao Jinping.Study on the Effects of Abnormal Events to Empirical Mode De-composition Method and the Removal Method for Abnormal Signal[J].Periodical of Ocean University of China,2001,31(6):805-814. |
| |
| Authors: | Zhao Jinping |
| |
| Abstract: | Empirical Mode Decomposition (EMD) is a newly developed method used to analyzing non steady and non linear temporal process data.By EMD method,the original data can be decomposed into several Intrinsic Mode Functions (IMF). In time series data,local signals with higher frequency can often appear,which is usually caused by abnormal events.If EMD method is adopted to analyze this kind of data with abnormal signal,different frequencies will be mixed in each IMF.The frequency mixing causes the difficulty in understanding temporal processes.This problem has not been solved in EMD method. In this paper,three corresponding characteristics for abnormal signal between the original data and the first IMF are noticed.Basd on these characteristics,the start and end positions of an abnormal signal can be determined.Then the abnormal signal within both positions is removed directly.The gap left is then interpolated by Spline function,which is constructed by the known data outside the abnormal interval.By superposing the all IMFs including the revised first IMF,a new data set with the abnormal signal removed is formed.Decomposing the new data set by EMD method again,a set of new IMFs is obtained,which is quite different than the one with abnormal signal.Our study shows that the abnormal signal removal is absolutely important in getting reasonable decomposing result and in displaying the temporal processes with different time scale.This is also a method for information extracting of abnormal signal. This method is suitable only for the short interval abnormal signal.Other methods are still needed for removing the abnormal phenomena with the interval longer than the time scale of background process. |
| |
| Keywords: | Empirical Mode Decomposition Method abnormal event intrinsic mode function high frequency signal |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
