基于变分模态分解的弹性参数核密度估计方法  

Kernel Density Estimation Method of Elastic Parameters Based on Variational Mode Decomposition

作　　者：朱鑫杰 张宏兵[1] 曾繁鑫 祝新益 ZHU Xin-jie;ZHANG Hong-bing;ZENG Fan-xin;ZHU Xin-yi(College of Earth Sciences and Engineering,Hohai University,Nanjing 211100,China)

机构地区：[1]河海大学地球科学与工程学院,南京211100

出　　处：《科学技术与工程》2024年第10期4005-4012,共8页Science Technology and Engineering

基　　金：国家自然科学基金(41374116)。

摘　　要：概率密度建模是地震随机模拟中至关重要的环节,而弹性参数高频成分的概率密度估计决定了高分辨率地震随机模拟结果的精度。针对常规方法中弹性参数高频成分提取精度不足、概率密度建模先验条件过度约束以及弹性参数的概率密度建模分层设计等问题,提出了一种基于变分模态分解(variational mode decomposition,VMD)的弹性参数核密度估计方法。该方法首先采用VMD对测井弹性参数数据进行模态分解,筛选出本征模态函数(intrinsic mode function,IMF)中的高频项叠加得到测井弹性参数的高频成分;然后使用核密度估计分层计算得到高频成分的概率密度模型,并通过该模型进行随机抽样生成随机高频成分叠加至井旁地震数据上以达到丰富地震弹性参数数据高频内容的目的。珠江口盆地34号井区的实验结果显示,VMD有效分离出了中心频率在70 Hz以上的测井弹性参数高频成分,分层设计的核密度估计方法凸显了高频成分的统计规律,叠加随机高频成分后地震弹性参数70 Hz以上的高频成分得到了明显补充。该方法为地震高分辨率随机模拟提供了新的思路。Probability density modelling is a crucial aspect of seismic stochastic simulation,and probability density estimation of the high frequency components of the elastic parameters determines the accuracy of high resolution seismic stochastic simulation results.To address inaccuracies in extracting high frequency components of elastic parameters,excessive constraints in a priori conditions for probability density modelling,and hierarchical design in conventional methods,an elastic parameter kernel density estimation method based on variational mode decomposition(VMD)were proposed.Firstly,the VMD method was employed to conduct modal decomposition on the logging elastic parameter data,followed by the selection of high frequency terms within the intrinsic mode function(IMF)in order to acquire the high frequency component of the logging elastic parameter.Then the probability density model of high frequency components was obtained by using kernel density estimation hierarchical computation,and the model was used to generate random high-frequency components by random sampling to be superimposed on the seismic data next to the wells in order to enrich the high-frequency content of the seismic elasticity parameter data.The experimental results in Well 34 of the Pearl River Mouth Basin show that VMD effectively separates the high frequency components of the logging elastic parameters with a central frequency above 70 Hz.The kernel density estimation method with layered design highlights the statistical law of the high frequency components.After superimposing the random high frequency components,the high frequency components of the seismic elastic parameters above 70 Hz are obviously supplemented.This method provides a new idea for high-resolution stochastic earthquake simulation.

关 键 词：高分辨率 地震数据 弹性参数 变分模态分解 核密度估计 

分 类 号：P631[天文地球—地质矿产勘探]