基于支持矢量机的聚驱井压裂选层方法

聚驱井压裂效果评价是压裂优化设计中的重要环节.提出了筛选压裂效果影响因素的主成分分析方法,通过主成分分析可以筛选出关键影响因素组合,减少参数间多重相关性,降低问题复杂性.根据统计学习理论建立了聚驱井压裂效果评价的统计学习模型,与模糊数学和人工神经网络等模型相比,该模型具有更好的对已知数据的综合能力和对未知数据的推广能力.使用支持矢量机算法进行数值计算,最终得到压裂设计的优化方案.

The Method of the Selection of Fractured Interval of Polymer-flooding Well Based on the Support Vector Machines

The effect evaluation of polymer-flooding well fracture is the important step in fracture optimization design. The principal component analysis method that screens influencing elements of fracture effect is put forward in the article. The key influencing element combination can be screens out by the principal component analysis method, the multiple correlations among parameters can be decreased and the problem complexity can be reduced. The statistic learning model of the fracture effect evaluation of the polymer-flooding wells is established according to the statistic learning theory. Comparing with the fuzzy math model and the neural net model and so on, the model has the better summarizing ability to known data and the polularizing ability to unknown data. The final optimization program of fracture design can be received by using the support vector algorithm calculates data.