论文部分内容阅读
本文通过拟合油藏的静、动态数据,解决了估算非均质多相油藏中孔隙度和渗透率分布的反演难题。方法包括建立目标函数的快速代理,函数的估值涉及执行一项费时的基于神经网络、DACE建模和自适应采样的数学模型(如油藏数值模拟模型),用自适应采样并考虑代理模型所提供的资料及误差期望值来探寻远景区。建议采用的方法是一个全局优化的方法,因此避免了潜在的局部最小值的敛合问题。例举的两个实例评价结果表明,该方法有助于优化包括涉及昂贵的油藏数值模拟计算的目标函数,并且在油藏描述和石油工程的其他方面均可应用。
By fitting the static and dynamic data of the reservoir, this paper solved the inversion problem of estimating the distribution of porosity and permeability in heterogeneous heterogeneous reservoirs. The method consists of establishing a fast proxy for the objective function whose evaluation involves performing a time-consuming mathematical model based on neural networks, DACE modeling and adaptive sampling (eg reservoir numerical simulation models), adaptive sampling and considering the proxy model The information provided and the expected value of error to explore the prospects. The proposed method is a globally optimized method, thus avoiding the convergence of a potential local minimum. The results of the two example evaluations show that this method can be used to optimize the objective function including numerical simulation involving expensive reservoirs and can be applied in other aspects of reservoir description and petroleum engineering.