摘要
气井生产数据自动拟合是一个非线性参数估计问题 ,这类气藏工程问题中Hessian矩阵常常是严重病态的 ,其病态来自于待估计参数本身的不敏感性、观测数据不包含某个待估计参数的响应和参数之间有关相关性等。围绕克服Hessian矩阵的病态形成了多种方法 ,但是均没有挣脱求解正则化方程组的限制。奇异值分解方法不仅摆脱了求解正则方程的限制 ,而且具有克服病态能力强、运算中不放大误差的优点 ,成为目前求解病态方程组和线性最小二乘问题的最好方法。文章提出一种气井生产数据自动拟合分析方法 ,其应用实例表明该方法性能良好 ,是目前在计算机自动分析生产数据这一领域中最为先进的方法。
The automatically fitting of gas-well production data is to solve the problem of estimating nonlinear parameters and in solving such a gas reservoir engineering problem, the Hessian matrix is always in a serious ill condition. Its ill condition is caused by the estimation-waiting parameters' own insensitivity, the observational data's not including the response of one estimation-waiting parameter and the correlation among the relevant parameters, etc. Many methods have been formed around overcoming the iii condition of Hessian matrix, they all, however, dont't shake off the limit of solving regularized equation group. The singular value decomposition method, which not only shakes off the limit of solving regularized equation group but also is possessed of these advantages as strong ability of overcoming ill condition and not enlarging the error in calculation, has become the best one of solving the ill condition equation group and linear least square problems. An automatically fitting analysis method of gas-well production data is proposed in the paper. Its application example indicates that this method with good performance is the most advanced one in the domain of automatically analyzing production data by computer.
出处
《天然气工业》
EI
CAS
CSCD
北大核心
2002年第3期71-73,共3页
Natural Gas Industry
