偏最小二乘法在合采井单层产能贡献率配分中的应用

Application of PLS to the Calculation of Productivity Contribution in the Commingled Producing Wells

根据多层混采原油分层产能贡献色谱烃指纹相关理论,在实验室用分层原油配比实验过程模拟混采油井的生产过程。定量了指纹化合物绝对浓度(峰高比值)。结合偏最小二乘法方法,建立了合采井原油产量分配计算的多元回归理论模型。模型有效解决了噪声和共线性带来的其他数学模型不稳定问题,拟合效果好;并且应用该模型计算了三层配比实验的单层产能贡献率,相对误差平均值在8%左右。同时在辽河地区进行了实际应用。计算结果与其他数学算法相差不大,表明利用气相色谱指纹技术结合偏最小二乘法计算多层合采井的单层产能贡献是可行的。

According to the theory for chromatography fingerpinting concentration of hydrocarbon in dynamic monitor process of productivity contribution in single zone of multiple-zone production,oil allocation ratio test in laboratory was used to emulate the production process of commingled producing wells,then absoluted the relative concentrations of finger compounds,combined with partial least squares（PLS）,multiple regression model about the calculation of productivity contribution in the commingled producing wells was established.This model is an effective solution to the instability of other mathematical models caused by the noise and co_linear,and fitting well.Using the model to calculate the productivity contribution,the average relative error is about 8%.At the same time,the practical application was used in Liaohe region,the results do not much differ from other mathematical algorithms.It shows that,the gas chromatography technology combined with PLS is feasible in the calculation do productivity contribution.