基于支持向量回归机的油田系统预测方法

Prediction of oilfield systems based on support vector regression

为克服传统预测方法的局限性,提出一种基于支持向量回归机算法的油田试井系统建模和预测方法.应用原-对偶算法,可解决支持向量回归机中的二次规划问题,而把该支持向量回归机应用于试井压力预测,结果表明:该方法具有较高的建模和预测精度,建模和有验证预测的平均相对误差在1%以内.

A new way of modeling and prediction in oil fields based on the SVR is proposed in this paper.And the primal-dual algorithm is used to solve the dual programming in SVR algorithm.Support Vector regression is adopted to predict the pressure of test wells in oil fields.The application example shows that it is highly precise and the above new algorithm is applied to modeling and testing prediction for test well system.The average relative errors of modeling and prediction are within 1%.