论文部分内容阅读
油田产量预测工作一直是油田开发中的一项重要工作,许多传统的回归模型以及智能算法都已经在油田产量预测中有了应用.虽然神经网络以其较强的非线性拟合能力.而得到广泛应用,但是传统BP神经网络容易陷入局部最优值而影响预测结果.将利用遗传算法同时优化BP神经网络连接权值和阈值的算法应用到大庆油田BED试验区高含水阶段的油田产量预测,结果表明在面对高含水阶段更加复杂的地质条件和数据波动更强的情况下优化后的神经网络收敛速度更快而且预测精度更高.
Prediction of oilfield production has always been an important task in the development of oilfields, and many traditional regression models and intelligent algorithms have been applied in oilfield production prediction.Although neural networks have obtained their strong non-linear fitting ability, But the traditional BP neural network is easy to fall into the local optimal value and affect the prediction results.Based on the genetic algorithm to optimize BP neural network connection weights and thresholds simultaneously, this paper applies the prediction of oilfield production in the high water cut stage of BED in Daqing Oilfield, The results show that the optimized neural network converges faster and has higher prediction accuracy under the condition of more complicated geological conditions and more fluctuating data in the high water cut stage.