二分类logistic回归建立油井热洗效果预测模型

Thermal well washing effect forecasting by using two-category logistic regression

热洗作为最简单、经济、有效的清蜡、解堵工艺,在我国各油田普遍采用,热洗效果和多因素有关:油井的产液量、动液面、泵挂深度、含水率、地质因素等等,如何对热洗效果进行预测,目前还没有学者在这方面进行理论研究,为此提出引用二分类logistic回归法建立油井热洗预测模型的思路,利用模型进行油井热洗效果的预测,将应用于医学的预测方法引入油田开采工艺中。在鄂尔多斯三叠系长8层的典型低渗稠油M油藏,利用油田已知的热洗数据推导出该油田的预测模型,对该油藏油井热洗效果进行预测,发现该理论能有效指导现场作业。

As the most simple,economic and effective wax,plugging removal technology,the thermal washing used in various fields of our country. The hot washing effect is related to many factors such as： liquid production,oil well dynamic liquid level,the pump setting depth,water content,the geological factors and so on. How to forecast the heat cleaning effect？ There is no theoretical research at present. This paper puts forward the thought of using two- category logistic regression to establish the prediction model of thermal well washing,i. e. the traditional forecasting methods applied in the medicine is used into the oil extraction process. M reservoir is a typical heavy,low permeability oilfield of Chang 8 layer of Erdos Triassic. The reservoir prediction model was derived by using the known thermal washing data of the field. Through the thermal washing effect forecast of the reservoir,it is found that the theory can effectively guide the field work.