基于支持向量机的早期聚合物驱动态预测研究

海上稠油油田由于受到开采时间的限制,一般在含水率达到40%之前进行聚合物驱,以提高油藏采油速度和开发效果,开发动态与陆上油田高含水期聚合物驱存在较大差异。文中应用统计方法,分析海上稠油油田早期聚合物驱的生产动态,从数据序列分析的角度研究其动态规律,认识不同动态参数之间的内在联系;基于支持向量机方法,建立时间与含水率、采出程度等动态参数之间的变化关系,并根据实际动态资料及时修正,进而建立能够反映早期注聚合物规律的预测方法。利用建立的预测模型,对渤海油田典型聚合物驱油藏进行实例计算分析。结果表明,建立的模型能够实现时间序列的多步预测,且预测结果满足精度要求,为早期聚合物驱开发指标预测和动态适时调控提供了一种有效的方法。

Development index prediction of early polymer flooding based on support vector machine

Due to the limits of development time in offshore heavy oilfield, in general, polymer flooding is used to improve the production rate and development effect before the water cut reaches 40%. However, there are differences in production performance comoared with polymer flooding of onshore thin oil oilfield at high water cut stage. Statistical method was applied to analyze theproduction performance of early polymer flooding in offshore heavy oil oiffield, and dynamic laws and development trends of early polymer flooding were studied from the perspective of data sequence analysis to reveal the internal relations among different dynamic parameters. The change relations among water cut, recovery percentage and production time were constructed based on support vector machine, corrected in time according to actual dynamic data. At last a forecast method to reflect development law of early polymer flooding was established. A case of typical polymer flooding reservoir was analyzed in Bohai oilfields through the use of established forecast model. Results show that the model can fulfill the muhistep prediction of time series, and the prediction results can meet the accuracy requirement, which provides an effective method to predict the development index of early polymer flooding and adjust the reservoir performance in time.