基于WT与LSSVM的储层流动单元划分方法

摘要

某研究区储层特性较为复杂,为了依据测井数据准确求取储层参数,提出基于小波变换(WT)与最小二乘支持向量机(LSSVM)相结合的储层流动单元划分方法.选取21口关键井的岩心物性资料、测井资料,依据流动层带指数划分方法将取心井储层流动单元划分成Ⅰ、Ⅱ、Ⅲ类,建立流动单元的识别规则和划分标准.将WT与LSSVM相结合对取心井储层流动单元进行学习训练,使用WT对各测井曲线分别分解为高频和低频成分,利用C5.0决策树对不同频率成分的训练样本进行参数敏感性分析得到学习所用的训练样本集,利用LSSVM训练训练样本建立流动单元预测识别模型,使用该模型对取心或非取心段储层流动单元进行预测.实验表明,基于WT与LSSVM的储层流动单元划分模型具有较高的识别精度,为储层精细评价提供一种较有效的研究方法.%It is very difficult to describe the complex reservoir feature by using log data to calculate the reservoir parameters.In order to improve the accuracy of reservoir identification,the method of flow unit division has been proposed based on Wavelet Transform (WT) and Least Squares Support Vector Machine (LSSVM).On the basis of flow zone indictor (IFZI) method,combined with coring data with log interpretation data of twenty-one key coring wells,the reservoir is divided into three types of flow units.The identification rules and division standards of flow units are established by IFZ.Based on this,the combination of WT and LSSVM is used to learn and train the reservoir flow units of coring wells.WT is applied to decompose the logging signal into multiple levels of resolution.The training set is attained by analyzing the sensitivity of samples of multiple levels via C5.0 decision tree algorithm.A model for predicting the flow unit type is developed by LSSVM.The proposed model is used to learn and predict the flow unit types in the coring wells or non-coring wells.Experiment results show that the hybrid model has high recognition accuracy.Thus,the hybrid model based on WT and LSSVM provides an effective way for fine reservoir interpretation.