【摘要】 利用传统的数学统计方法计算采收率具有较大的困难.在对已知油藏最终采收率分析的基础上,提出了一种基于支持向量机的油藏建模方法,将影响采收率的主要因素作为输入信息,建立诸因素与采收率之间的关系,得到采收率预测模型.利用该模型计算出的采收率与实际最终采收率基本吻合,最大拟和误差为0.142%,最大泛化误差为2.744%.仿真结果表明,该模型能较好地建立各因素与最终采收率的复杂非线性关系,反映油藏的本质,具有较高的预测精度.更多还原

【Abstract】 It is difficult to calculate the recovery ratio by traditional statistics. A novel method based on support vector machine is proposed, whose fundamental is analysis on the oil recovery ratio. Some factors which affect are mainly regarded as inputs in order to establish the relation between recovery ratio and factors. Thus the prediction model of recovery ratio in oil field is established. The experiment results given have proved the recovery ratios computed by support vector machine tally with the actual ones. The most fitting error is 0.142%, and the biggest generalized error is 2.744%. The simulation shows that the new model reflects the relation between factors and final recovery ratio, which embodies the essence of the oil field. The results show the new algorithm used in the oil is effective.更多还原