储层粒度神经网络预测模型研究

国内外多年的研究表明，储层粒度特征值（d50）、非均质系数（d40=d90）是防砂设计的基础。常规获取粒度分布 范围的方法主要有激光粒度测试法（LDA）与筛析法（SA），两种方法均需要通过岩芯粒度测试来获取数据，而在制定 开发井的完井防砂措施时往往没有实际开采层位的岩芯，只能参照探井粒度数据进行设计，从而导致较大的误差。针 对该问题，从测井的角度出发，开展了储层粒度与多种测井曲线的响应关系的研究，采用神经网络技术，建立了探井伽 马、密度测井项与实测粒度特征值三者样本库，训练出满足工程需要的学习网络，进而结合开发井测井资料，获得了整 个粒度纵向分布剖面，为防砂分层设计提供准确的基础数据支撑。目前，该方法在中国海上多个油田的分层防砂优化 设计中获得了成功应用，预测误差可控制在10% 以内。

AN ANALYSIS OF THE PROBLEM OF LEFT-BEHIND OLD PEOPLE IN THE COUNTRYSIDE

Based on a month’s survey in No.four villager group of Xiaogan,Hubei,we found that the problem of left-behind old people in rural areas is particularly serious among the "three left-behind" problems.Old people left behind in the countryside lack stable economic security,and hence their inadequate ability to resist risks.The shortage of family care and the insufficient communication between the old and the young have greatly contributed to the old people’s sense of loneliness.And the most prominent problem in recent years is abnormal death of the old people left behind in the countryside.According to the problems,the author suggests taking the construction of association of the elderly as the key point and breakthrough,so that we can explore a "low cost,high welfare" mode of practice,which really provides a sense of security,fun and worthiness for the left-behind old people.