利用支持向量机优化压裂加砂规模研究

压裂效果受多种因素的影响，与工艺参数、储层条件等影响因素之间是一种非线性映射关系。针对压裂设计过程中确定加砂规模难的问题，利用前期压裂井数据作为样本，考虑储集层有效厚度、含油饱和度、孔隙度、渗透率、加砂规模、砂比对压后效果的影响，通过支持向量机的学习训练，建立影响因素与压裂效果之间的预测模型，预测不同加砂规模下的压后效果，从而优选出最佳的压裂加砂规模。对压裂井进行实例计算并通过现场实施，表明该方法能有效克服小样本难题并且收敛速度快，优化结果与实际压裂效果吻合度高，对压裂加砂规模设计具有指导意义。

THE STUDY ON OPTIMIZING THE PROPPANT QUANTITY IN FRACTURING BY SVM

Fracturing effect is affected by many factors.There is the nonlinear relationship between the effect and reservoir parameters and reservoir conditions.In view of the difficulty of determining the sanding scale in designing fracturing process，the data of pre-fracturing well are used as sample data，the influence of net thickness，oil saturation，porosity，permeability，sanding scale and sand ratio on fracture effect should be considered，samples are learned by SVM，the forecasting model between influencing factors and fracture effect is established in the purpose of predicting the fracture effect under the different sanding scale，thereby the sanding scale is optimized.It is showed by calculation of case study in fracturing wells and the field implement that SVM has a fast convergence speed and can overcome effectively the problem of small samples，and optimal results effectively coincide with the actual fracturing effect，so it has a certain guiding sense to designed the sanding scale in fracturing.