优化局部均值分解在趋势信息提取中的应用

Application of optimized local mean decomposition in trend information extraction

针对局部均值分解(LMD)求取乘积函数分量(PF)时内层循环的迭代停止条件采用硬阈值准则不具备自适应性这一问题,该文提出一种基于自适应迭代停止准则的局部均值分解算法,利用迭代停止准则函数判定内层循环过程,该函数对包络估计函数从整体偏差和局部异常两方面综合考虑,利用连续相邻三次的函数值判定循环迭代的停止条件,以此计算纯调频信号,最终求取对应的乘积函数。仿真实验结果表明,与传统局部均值分解和经验模态分解(EMD)相比,该文方法的分解质量更优,具有更好的挖掘隐含信息能力。利用基于自适应迭代停止准则的局部均值分解对卫星测高获取的全球平均海平面(GMSL)变化数据进行分解和分析,结果表明提取出的GMSL长期趋势性变化信息是有效的。

Aiming at the problem of using hard threshold criterion for the iteration stopping condition of the inner loop when calculating the product function components of local mean decomposition,a local mean decomposition algorithm based on adaptive iteration stopping criterion was proposed,and the inner iterative loop process was determined by the iteration stopping criterion function.The function comprehensively considered the global deviation and local abnormality of the envelope estimation function,and used the function values of three consecutive adjacent times to determine the stop condition of the loop iteration,so as to calculate the pure FM signal,and finally obtained the corresponding product function.The results of the simulation experiments showed that,compared with the traditional local mean decomposition and empirical mode decomposition,the method proposed in this paper had better decomposition quality and better ability to mine hidden information.Using the method of this paper to decompose and analyze the global mean sea level change data obtained by satellite altimetry,and the results showed that the extracted long-term trend change information of GMSL was real and effective.