基于核主成分-小波能谱分析的复杂储层油水界面预测

Oil/water contact prediction of complex reservoir using kernel principal component analysis and wavelet power spectrum analysis

针对复杂储层油水界面确定困难的问题,研究基于核主成分-小波能谱分析的确定方法,分析油水界面特征并确定最优小波基函数。将深感应电阻率、中子、密度、声波和自然伽马等常规曲线作为输入信号,应用核主成分方法提取第一核主成分信号进行小波变换得到近似和细节信号后分别计算不同频带的能谱,基于相邻两级能谱差异最大原理确定单井的油水界面。在单井分析的基础上确定整个油藏的油水界面。研究结果表明:Coiflet3基函数在研究工区中应用效果最好;将第一核主成分进行6级分解,并根据第5级和第6级能谱最大差法确定的油水界面与生产测试资料十分吻合;油水界面高度由西向东呈先增大后减小的趋势,与储层岩性、物性的平面变化特征一致。

To solve the problem of oil-water contact determination for complex reservoir, a new method based on kernel principal component and wavelet power spectrum analysis was carried out. The optimal wavelet base function was analyzed and chosen. The deep induction resistivity, neutron, density, acoustic and natural gamma data were used as the input signals, the kernel principal component（KPCA） method was applied to extract the first kernel principal component signal, and then wavelet transformation was exacted to get the approximate and detail signals. Different frequency band power spectrums were calculated based on the decomposed approximate and detail signals respectively. The oil-water contact of single well was determined based on the principle of the indication of maximum difference between adjacent power spectrum. The oil-water contact distribution of the whole reservoir was determined based on the analysis of single well. The results show that Coiflet3 base function is best applied in the study area; Applying 6 levels of decompositionto the first principal component is the best fit for the determination of the oil-water contact according to the biggest difference method of grade 5 and grade 6 power spectrum value. The result is in good agreement with that of the production and test. The oil-water contact first increases and then decreases from west to east and is consistent with the planar variation characteristics of reservoir lithology and physical properties.