预测封堵影响产量比例的新方法NEW METHOD PREDICTING THE PRODUCTION RATIO INFLUENCED BY THE PLUGGING
董烈
DONG Lie
摘要(Abstract)：
大庆油田二类油层开始聚驱后,水聚驱共采一套层系,为避免影响聚合物驱效果,水驱目的层要进行封堵。预测水驱封堵影响产量成为老井产量预测的主要因素。目前预测封堵影响产量的方法主要有砂岩厚度比例法、多元线性回归法和数值模拟法。砂岩厚度比例法简单易操作但预测误差在10%以上,多元线性回归法由于样本数量少难以达到预测精度要求。因些提出了一种新的预测封堵影响产量的基于支持向量机的预测方法。该方法是利用17个区块的实际封堵数据,选取递减率等6个参数作为输入量,建立了支持向量机网络模型,针对3种核函数进行了优选并对预测结果进行了检验,最大误差为7.23%,最小仅为0.41%,证实了支持向量机方法的可行性。
After the polymer flooding of the second-class oil reservoirs in Daqing Oilfield,one set of series of strata is commonly produced by both water and polymer floodings.In order to avoid the effects of the polymer flooding,the target layers flooded by the water are needed to plug.The prediction of the influenced production by the water plugging becomes the main factor of the prediction of old well production.At present,the methods to forecast the influence show as follows:sandstone thickness ratio,multiple linear regression and numerical simulation.Sandstone thickness ratio method is simple and easy to operate,but the forecast error is more than 10%;because of the small number of samples,the forecast precision of the multiple linear regression is difficult to meet the requirements.Therefore a new predicting method is developed to forecast the plugging influence based on support vector machine.With the help of the actual plugged data in 17 blocks,choosing six parameters such as decline rate and so on as input parameters,the network model of the support vector machine is established to optimize three kinds of kernel function and verify the predicted results,the maximum error is 7.23%,the smallest is only 0.41%,the feasibility of support vector machine is soundly confirmed.
关键词(KeyWords)：
封堵;支持向量机;核函数;产能预测
plugging;support vector machine;kernel function;production prediction
基金项目(Foundation): 国家科技重大专项及示范工程“大庆长垣特高含水油田提高采收率示范工程”(2011ZX05052)资助
作者(Author):
董烈
DONG Lie
参考文献(References)：
- [1]罗宪波,赵春明,武海燕,等.海上油田多层合采层间干扰系数确定[J].大庆石油地质与开发,2012,31(5):102-104.
- [2]兰丽凤,刘仁强,付艳波,等.油井分层产量劈分新方法[J].大庆石油地质与开发,2012,31(3):59-62.
- [3]李成勇,柳金城,李晓茹.多层合采油藏油水井产量劈分原理及应用[J].大庆石油地质与开发,2010,29(1):55-59.
- [4]于海丽,程时清,张利军.根据油井含水率确定分层剩余油饱和度的图版法[J].大庆石油地质与开发,2011,30(2):105-109.
- [5]吴克柳,李相方,严必成.用产量数据动态跟踪地层压力的新方法[J].大庆石油地质与开发,2011,30(6):94-97.
- [6]侯玮,郭杰,贺成才.基于系统稳定性的复杂断块油藏水驱产量预测[J].大庆石油地质与开发,2011,30(4):70-75.
- [7]于代国,孙建孟,王焕增,等.测井识别岩性新方法——支持向量机方法[J].大庆石油地质与开发,2005,24(5):93-95.
- [8]赵传峰,姜汉桥,郭新华.支持向量机在小样本预测中的应用[J].油气田地面工程,2009,28(2):21-23.
- [9]王继强,韩大匡,金志勇,等.支持向量机在单井措施增油量预测中的应用[J].新疆石油地质,2008,29(1):109-112.
- [10]袁士宝,蒋海岩,鲍丙生,等.基于支持向量机的火烧油层效果预测[J].石油勘探与开发,2007,34(1):104-107.
- [11]Gunn S.Support vector machines for classification and regression[R].Technical report:University of Southampton!,1998.
- [12]Keerthj S,Shevade S,Bhattcharyya C.A fast iterative nearest point algorithm for support vector machine classifier design[J].IEEE Transactions on Neural Network,2000,11(1):124-136.