摘要
钻井液的幂律流变模式的流变参数估计问题大多使用线性回归方法进行求解,然而,线性回归方法改变了测量误差的统计特征,使得所得到的流变参数估计不具有无偏性和方差最小等特点。针对幂律流变方程的特点,提出了一个非线性最小二乘估计的新算法,该算法不需要人工给定迭代初始值,迭代过程稳定收敛到最小点,不会陷入极小点陷阱,收敛速度很快。新算法是一个全局优化算法,所得到的流变参数估计具有拟合残差近似无偏性和方差几乎最小的优良统计特征。大量的实际钻井液算例表明,新方法具有比线性回归方法更小的拟合方差和相近的均值,拟合残差统计特性优于线性回归方法。该方法可以应用到钻井液幂律流变模式中流变参数的确定、钻井液流变模式优选、钻井液性能调整和评价等许多问题中。
Frequently the linear regression method used in the approximation of the solution to the power law model altered the statistic nature of the measurement errors,and did not give unbiased results and results with minimum variance.A new least square approximation algorithm was developed based on the characteristics of the power law model.This new algorithm,with a fast convergence and no requirement of defining an initial value,gave a converging minimum value and would not run into the 'minimum value' trap.The new algorithm was a global optimization one and yielded results that were nearly unbiased and variance that was minimal.Field practices showed that the new algorithm was better than the conventional linear-regression method,and can be used in the determination of the rheological parameters of the power law model,the optimization of the rheological model and the evaluation of the drilling fluid properties.
出处
《钻井液与完井液》
CAS
北大核心
2007年第B09期65-68,共4页
Drilling Fluid & Completion Fluid
基金
中国石油天然气股份公司油气勘探超前共性科技项目"辽河探区西部凹陷深化勘探理论与实践"(编号07-01C-01-04)的部分研究成果
关键词
钻井液
流变方程
幂律模式
线性回归
非线性回归
drilling fluids
rheology equation
power law model
linear regression
non-linear regression
