海上油田数字化、智能化是发展的必然趋势,油气井多相流量软测量是智能油田建设的重要一环。为进一步研究数据驱动软测量模型对海上油田多相流计量的适用性和应用效果,基于某海上油田160余口井约16 000个历史生产数据,优选电潜泵功率、电潜泵前后差压、油嘴直径、油压等生产测量参数作为模型输入辅助变量,建立了基于极端梯度增强树(Extreme Gradient Boosting,XGBoost)算法的数据驱动软测量模型。训练后的模型对油田2 400个生产样本进行测试计量,结果表明模型对产油量、产气量和产水量的计量平均绝对百分比误差均≤6%。结论认为,基于XGBoost算法的数据驱动软测量模型对海上油田多相流计量的适用性和应用效果良好,可作为油田单井计量的解决方案之一,对智能油田建设具有借鉴意义。

 The digitalization and intelligent development of offshore oilfield is an inevitable trend. The soft sensing of multiphase flow measurement in oil and gas wells plays a crucial role in the development of intelligent oilfield. To study the applicability and impact on the application of data-driven soft sensing model of multiphase flow measurement in offshore oilfields, this paper bases its research on approximately 16 000 historical production data from more than 160 wells. Six production measurement parameters were selected as model input auxiliary variables, including electric submersible pump power, differential pressure upstream and downstream of the electric submersible pump, nozzle diameter and oil pressure, to develop a soft sensing model driven by the Extreme Gradient Boosting (XGBoost)algorithm. The trained model was tested on 2 400 oil field production samples and the results show that the average absolute error for measuring oil production, gas production and water production was within 6%. It is concluded that the data-driven soft sensing model based on neural network demonstrates good applicability and application effect for multiphase flow measurement in offshore oilfield and can serve as one of the solutions for singlewell measurement in oilfield. This study results offer significant reference for the development of intelligent oilfields.