基于分频智能反演的储层预测研究——以大庆黑鱼泡南部探区为例

分频反演是近年发展起来的一种先进的地震反演技术。大庆黑鱼泡南部探区存在鼻状构造，是油气富集有利部位。本文基于分频智能反演的储层预测研究，根据地震资料频谱确定分频参数：低频在8 Hz左右，主频35 Hz左右，高频60 Hz左右。根据地震频谱计算得到具有一定带宽的低、中、高频属性。运用支持向量机算法对目标曲线进行分频智能学习训练，学习后曲线相关性达到0.90，满足储层预测的要求。最终，本文将重构曲线与三维地震数据的非线性映射关系引入到二维地震进行分频智能反演，得到二维区阻抗体，对砂岩的预测性较好，为后续的地质建模提供了良好的基础输入数据。

Research on reservoir prediction based on frequency-divided intelligent inversion: A case study from southern exploration area of Heiyupao in Daqing

Frequency-divided inversion is an advanced seismic inversion technique developed in recent years. The nose structure exists in the southern exploration area of Heyupao in Daqing, which is a favorable location for oil and gas enrichment. In this paper, reservoir prediction research is based on frequency-divided intelligent inversion, and frequency division parameters are determined according to the spectrum of seismic data:low frequency is about 8 Hz, dominant frequency is about 35 Hz, and high frequency is about 60 Hz. From seismic spectrum calculation, low, medium and high frequency properties with certain bandwidth are obtained. Further, the support vector machine algorithm was used for frequency-divided intelligent learning and training of the target curve, and the correlation of the curve reached 0.90 after learning, which met the requirements of reservoir prediction. Finally, in this paper, it was introduced that the nonlinear mapping relationship between the reconstructed curve and the 3D seismic data into the 2D seismic frequency-divided intelligent inversion, and the 2D regional impedance body was obtained. The prediction of sandstone was good, which provided a solid basic input data for the subsequent geological modeling.