测井盲反褶积及在储层构型界面识别中的应用

老油田早期测井曲线分辨率低、对薄层沉积界面识别能力差，难以满足储层建筑结构研究的需要，针对这一问题，将信号反褶积原理应用于测井曲线处理。通过将反褶积问题转化为期望输出与实际输出误差的极值问题构造方程组，利用测井曲线自相关间接求取线性方程组中的褶积因子自相关矩阵，求解方程组得到反褶积因子，从而实现测井曲线的盲反褶积处理。该处理消除了围岩的影响，增强了测井信号频谱中高频段能量，对测井曲线的低幅度细微变化产生“放大”效应，对曲线斜率的微小变化响应敏感，提高了测井曲线对沉积界面的识别能力。通过取芯井资料研究，证明测井曲线反褶积处理结果中高频成分具有沉积学意义，将该方法应用于秘鲁D油田非取芯井储层建筑结构研究，在D6井V层识别出5个五级界面，3个四级界面和23个三级界面。

WELL LOGGING BLIND DECONVOLUTION AND THE APPLICATION IN RESERVOIR TEXTURAL INTERFACE

The early well logging curve resolution is quite low in the old oil fields，the identification to sedimentary thin bed surface is poor，which can not meet the requirement of researching reservoir textural composition，aiming at the issue，the signal deconvolution principle is applied in well logging curve processing.By means of turning deconvolution over to the extreme value of the error between expected output and actual output，the formulas are constructed，by means of well logging curve autocorrelation，the autocorrelation matrix of convolution factor is obtained，then，the formulas are solved and the deconvolution factor obtained，the blind deconvolution processing of well logging curve is fulfilled.The processing eliminates the influence of adjacent rocks and enhance high frequency energy in well logging signal frequency，a “zoomed” effect is generated in low amplitude in logging curve，which improves the identification of well logging curve to sedimentary faces.Through core data analysis，it is verified that high frequency data in deconvolution processing are of the meaning of sedimentology，the method is used in non-coring well study in field D，Peru，in Bed V，well D6，5 five-class interfaces，3 four-class interfaces and 23 three-class interfaces are recognized.