采用神经网络建模对海下有机材料加固的油井产量预测研究

Research on oil production prediction of offshore oil wells based on long and short term memory neural network

海下油井油产量预测在开发调整和优化中继续发挥着越来越重要的作用;但海下油井的加固会进一步影响油产量预测。研究建立注气效应的长短期记忆(LSTM)神经网络模型,预测已有机材料环氧树脂乳液加固渤海碳酸盐岩储层的生产性能;将该模型的计算结果与传统储层数值模拟(RNS)在相同条件下进行了对比。结果表明,LSTM所需的CPU负载仅为4%,LSTM方法的总CPU时间和综合计算功耗分别仅占RNS的10.43%和36.46%。LSTM方法在计算方面具有显著优势,为人工智能在油气开发中的应用提供了新的方向。

Oil production prediction of subsea oil wells continues to play an increasingly important role in develop⁃ment adjustment and optimization.However,the reinforcement of offshore oil wells will further affect the oil produc⁃tion forecast.The long short memory(LSTM)neural network model of gas injection effect is established to predict the production performance of Bohai carbonate reservoir reinforced with organic epoxy resin lotion.The calculation results of this model were compared with the traditional reservoir numerical simulation(RNS)under the same con⁃ditions.The results showedthat the CPU load required by LSTM was only 4%.The total CPU time and comprehen⁃sive computing power consumption of LSTM method only accounted for 10.43%and 36.46%of RNS respectively.LSTM method had significant advantages in calculation.It provides a new direction for the application of artificial intelligence in oil and gas development.