结合粗糙集与支持向量回归进行油藏物性参数预测

Reservoir parameters predicting by combination of rough set and support vector regression

为了更准确的预测油藏物性3个重要参数:孔隙度、渗透率、饱和度,提出了结合粗糙集属性约简和支持向量机回归的方法。首先用粗糙集理论对测井数据样本属性进行约简,从而选出决策属性,构成新的样本数据。然后用支持向量回归理论对数据样本进行训练,建立支持向量回归模型,并且对测试样本进行预测。实验结果表明,该方法获得了较好的拟合结果,并且减少了支持向量机在训练中的计算复杂度,提高了物性参数预测的准确率。执行该方法可为油藏开发提供决策依据。

To exactly predict three important parameters that are porosity, permeability and saturation, a method based on attribute reduction of rough set and support vector machine regression is presented. Firstly, the rough set theory is used to reduce the attributes of sampling data and to select the decision-making attributes constituting a new simply data. Secondly, the theory of support vector regression （SVR） is used for training data and the predicting model is established. Then, the test data will be predicted. The experimental results show that the method can get a better fitting result and reduce the computational complexity of SVR in training data and improve the accuracy of reservoir physical parameters. Implementation of the method can provide the foundation of decision making for reservoir development.