作　　者： ; ; ; ; ;

机构地区： 长江大学石油工程学院湖北省油气钻采工程重点实验室

出　　处： 《中外能源》 2011年第11期51-54,共4页

摘　　要： 针对低渗透油藏易受水锁损害的特点,介绍水锁损害机理,提出水锁损害实验室评价方法。同时,针对水锁损害实验评价既需要代表性的储层岩心,又耗费大量人力物力的缺点,介绍了BP神经网络原理和灰色系统理论,把两者引入水锁损害研究中来,将灰色GM(0,N)预测法和BP神经网络法有机结合,建立一种新的预测模型——灰色-神经网络预测模型,并分析其可行性,用计算机C语言程序实现了上述过程。以塔里木油田轮古7井区15块有代表性的岩心室内水锁损害评价结果为学习样本,另外5块岩心为预测样本,建立了灰色-神经网络水锁损害预测模型。预测结果表明,模型预测结果与实验室实测结果吻合程度较好。并通过与回归分析法、灰色GM(0,N)预测法和神经网络法这三种预测方法进行比较,发现灰色-神经网络水锁损害预测模型效果明显优于其他几种方法。 For low permeability oil reservoirs in the development process is vulnerable to be damaged by water locking,mechanism of water locking damage is introduced,and laboratory evaluation methods of water locking damage are proposed.Because representative reservoir cores are needed and there is a shortage of spending a lot of manpower and resource in the process of experiment evaluation of water locking damage lock,BP neural network theory and gray system theory are introduced.Both of them are also introduced into the research of water locking damage ,and a new forecasting model Gray-neural network prediction model is established through the combination of the gray GM （0,N） prediction method and BP neural network.And its feasibility is analyzed and C language computer program may achieve the above process.In this paper,the gray-neural network prediction model of water locking damage is established taking the laboratory assessment results of water locking damage in 15 representative cores of Lungu 7 well field in Tarim Oilfield as the study sample and another five core samples as the predicted samples.Through repeated training,the results obtained show that the model predictions and laboratory test results match each other.And through comparison of three prediction methods of regression analysis ,the gray GM （0, N） and neural network prediction method, the effect of the gray-neural network prediction model of water locking damage is better than other methods.

关 键 词： 水锁损害 界面张力 渗透率损害率 灰色 神经网络预测模型 评价

领　　域： [石油与天然气工程]