摘要
以影响压裂井产量的地层参数和压裂施工参数作为输入变量,压裂后的每米产量为输出变量,建立了压裂井产量预测的广义回归神经网络模型,并根据与目标油井地质参数的欧式距离的大小来选择学习样本。通过该模型预测不同砂量下油井压裂后的产量变化情况,并结合经济评价方法来确定以经济效益为目标的最优加砂量。实例分析表明,该方法是可行的。
The formation parameters and fracturing parameters influencing production rate of fracturing wells are used as input variables, the production rate per meter after fracturing is used as an output variable. A general regression neural network model is established for prediction production rate in fracturing wells. Learning sample is selected based on Euclid distance of geologic parameters in the target well. The model is used to predict production variation for different sand volumes after fracturing. Economy-orientated sand volume optimization is determined in combination with economic evaluation.
出处
《石油天然气学报》
CAS
CSCD
北大核心
2006年第3期88-90,共3页
Journal of Oil and Gas Technology
基金
中国石油天然气集团公司中青年创新基金项目(04E70102)
关键词
广义回归神经网络
压裂
压裂砂量
优化
经济评价
general regression neural network
fracture
sand volume used for fracturing
optimization
economic evaluation
