PSO-SVR模型在油井产能预测中的应用

Application of PSO-SVR Model in Oil Well Productivity Prediction

针对支持向量回归机（Support Vector Regression,SVR）用于油井产能预测中模型参数具有不确定的问题,采用粒子群算法（Particle Swarm Optimization,PSO）对支持向量回归机参数进行优化,减少模型参数的不确定性,产能数据测试表明,PSO能快速,准确地优化SVR参数,二者的结合有效地进行产能预测,并取得较为理想的效果。

In order to solve the uncertainties of model parameters in support vector regression（SVR） for oil well productivity prediction, Particle Swarm Optimization（PSO） is used to optimize support vector regression parameters to reduce the uncertainty of model parameters. The test of production data shows that the PSO can quickly and accurately optimize the SVR parameters. The combination of the two can effectively predict the production capacity and achieve a satisfactory result.