储层敏感性预测技术研究——以新疆油田准东地区为例  

Technical Study on Predicting Reservoir Sensitivity in Zhundong Area of Xinjiang Oilfield

作　　者：王松[1] 胡三清[1] 赵伟民[1] 魏霞[1] 宋文广[2] 王祖文[3] 黄秋伟[3] 

机构地区：[1]长江大学化学与环境工程学院,湖北荆州434023 [2]长江大学计算机科学学院,湖北荆州434023 [3]新疆石油管理局试油处,新疆克拉玛依834000

出　　处：《石油天然气学报》2011年第10期121-124,169,共4页Journal of Oil and Gas Technology

摘　　要：储层敏感性情况在保护储层的过程中占有很重要的地位,储层敏感性快速预测技术可以节省大量的人力、物力和时间。因此,采用人工神经网络建立模型来对储层敏感性进行预测。为了提高神经网络的应用效果,通过对4种改进的BP算法进行优选,最终确立采用Levenberg-Marquardt算法建立模型,采用MATLAB和VC++.NET混合编程,充分利用MATLAB优秀的数学建模能力和VC++.NET先进的面向对象的编程工具生成敏感性快速预测软件,大大缩短了软件的开发周期,并且提高了软件性能。使用该敏感性快速预测软件对新疆油田储层的敏感性进行预测,预测的结果与试验结果的符合度达到90%以上,说明该软件达到了实际生产中的需要。Reservoir sensitivity was very important for protecting reservoirs,rapid prediction of reservoir sensitivity could save times and manpower and material resources.Therefore,artificial neural network was used to establish a model to predict reservoir sensitivity.In order to improve the application effect of neural network,four kinds of improved BP algorithms were optimized,and finally model was established with the optimized Levenberg-Marquardt algorithm,Matlab and VC++ mixing programming.NET were used to make full use of Matlab excellent mathematical modeling ability and VC++ asp.net advanced object-oriented programming tools to generate quick sensitivity forecasting software,shorten its development period and improve the performance of software.The software is used for sensitivity forecasting in Xinjiang Oilfield,the results are compared with that of actual prediction and the coincident rate is above 90%,it indicates that the software is satisfied with actual production.

关 键 词：储层保护 敏感性 预测 BP神经网络 预测效果 

分 类 号：TE258[石油与天然气工程—油气井工程]