一种新的BEMD筛分停止准则  

Novel BEMD Criterion for Stopping Sifting Process

作　　者：葛光涛[1,2] 桑恩方[1] 刘卓夫 朱蓓蓓[3] 

机构地区：[1]哈尔滨工程大学水声工程学院,哈尔滨150001 [2]浙江工商大学信息与电子工程学院,杭州310018 [3]大连海事大学机电与材料工程学院,大连116026

出　　处：《数据采集与处理》2010年第2期195-200,共6页Journal of Data Acquisition and Processing

基　　金：国家自然科学基金(10504004)资助项目;浙江工商大学校级科研项目基金(1120KU110043)资助项目;浙江工商大学引进人才科研启动基金(1120XJ2010004)资助项目

摘　　要：为优化二维经验模态分解(Bidimensional empirical mode decomposition,BEMD)算法,提出了根据空间位置信息的稳定性制定BEMD筛分停止准则的思想;明确了一维经验模态分解和二维经验模态分解在筛分停止需求上的本质区别。均值曲面的物理状态预示了待筛分曲面极值点数量和分布变化规律,基于此制定出一种BEMD筛分停止准则。实验表明,该准则较传统的停止准则有明显优势。To optimize the bidimensional empirical mode decomposition （BEMD） algorithm, the BEMD criterion is established according to the stability of the position information. The essential difference of requirements between the one-dimensional EMD and the two-dimensional EMD is explicated to stop the sifting process. The physics state of the local mean surfaces indicates the diversification rule of the local surface extremum point number and the distribution. Therefore, the BEMD criterion is constituted based on the rule. Experiments prove that the novel criterion has obvious advantages compared with the current criterion.

关 键 词：二维经验模态分解 筛分停止准则 极值点 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]