基于最小二乘支持向量机的低渗透油藏调剖效果预测

Effect prediction of profile control based on least squares support vector machine in low permeability reservoir

调剖效果受油藏地质条件、开发状况和调剖工艺等诸多因素影响,调剖效果预测实质上是解决一个多因素共同作用的复杂非线性问题。常规的数值模预测对地质建模准确性要求高、历史拟合时间长而不能满足实际应用需求。本文提出利用最小二乘支持向量机法构建调剖效果与其主控因素之间的非线性映射模型,将未知井调剖效果预测转化为非线性函数的求解过程。以靖安油田五里湾区块24口调剖井资料为学习样本,建立了计算调剖后井组增油量、降水量的最小二乘支持向量机预测模型并进行了验证。结果表明,计算结果与实际值拟合精度较高、误差较小,该方法可用于矿场调剖效果预测与评价。

The efficiency of profile control is influenced by many factors such as geological conditions,development status and profile control technology,and effect prediction of profile control is considered as solving a complicated nonlinear problem which is influenced by many factors.It can not meet the actual application needs for conventional numerical simulation method with the demanding accuracy of geological modeling and the long time of history fitting.The least squares support vector machine（LSSVM） was used to construct the nonlinear mapping model between the profile control effect and its main controlling factors in this paper,and effect prediction of profile control for unknown well is transformed to the process of solving a nonlinear function.Based on the data of 24 profile control wells of Wuliwan block in Jing＇an oilfield,the LSSVM forecast model of increasing oil and decreasing water was established,and it was verified by inspection sample.The results show that the fitting accuracy between calculation results and actual values is relatively high and the error is small,the method can be used for prediction and evaluation of profile control effect in field.