BP神经网络在预测石南31油田产量变化中的应用

Application of BP Neural Network Technique for Forecasting Productivity in ShiNan31 Oilfield

本文提出了利用BP神经网络快速预测油田单井产量的方法。石南31油田在投产初期即投入注水开发,并在开发过程中实施过酸化压裂、补孔等措施,使得产能影响因素较多,且各种因素交互重叠,为用常规方法进行产量动态预测造成很大的困难。为了准确实时地调整开发技术政策,本文选用了BP神经网络,利用油田生产动态数据,应用MATLAB7.0建立了石南31油田产量预测模型,提高了预测的精度。研究结果表明,这是一种可行的方法。

Methods for quickly predicting the reservoir productivity were proposed by using BP neural network.Waterflooding program has been taken in the initial production of ShiNan31 Oilfield,where some measures have been taken,such as acid fracturing,re-perforating during the development.So there are many productivity influence factors,and it was insufficient to predict the productivity with conventional methods.For accurately and timely adjusting the production strategy,BP neural networks were selected and dynamic data were used to establish a productivity predicting model for improving the predicting accuracy in ShiNan31 Oilfield.It is proved that our method is a significant improvement for predicting the reservoir productivity with multiple stimulation methods.