摘要
利用测井资料预测储层孔隙度对于储层评价具有重要意义。针对现有孔隙度预测模型不能深度挖掘测井数据和孔隙度之间潜在关系的问题,论文提出一种利用麻雀搜索算法(SSA)优化双向门控循环神经网络(BiGRU)的储层孔隙度预测模型(SSA-BiGRU),以测井数据为输入,通过BiGRU深度挖掘测井曲线与孔隙度之间的非线性和时序特征;采用麻雀搜索算法对BiGRU神经网络模型中各层神经元个数、批处理大小、学习率等参数进行智能迭代优化,得到最优参数值,克服依靠经验选取或手动调参而导致预测精度低的问题。实验结果表明,相较于BP、LSTM、BiGRU等孔隙度预测模型,SSA-BiGRU预测模型有效提高了预测精度。
It is very important for reservoir evaluation to predict reservoir porosity by logging data.Aiming at the problem that the existing porosity prediction model cannot deeply explore the potential relationship between logging data and porosity,this paper proposes a reservoir porosity prediction model based on the sparrows search algorithm(SSA)to optimize bidirectional gated recurrent neural network(BiGRU),which takes logging data as input.BiGRU is used to explore the nonlinearity and time series characteristics between logging curves and porosity.The sparrow search algorithm is used to optimize the parameters of BiGRU neural network model,such as the number of neurons in each layer,batch size and learning rate,and get the optimal parameter values.The problem of low prediction accuracy caused by empirical selection or manual parameter adjustment is overcome.The experimental results show that the SSA-BiGRU prediction model effectively improves the prediction accuracy compared with BP,LSTM,BiGRU and other porosity prediction models.
作者
高雅田
王冉
吴润桐
GAO Yatian;WANG Ran;WU Runtong(School of Computer&Information Technology,Northeast Petroleum University,Daqing 163318;Daqing Oilfield Oil Production Engineering Research Institute,Daqing 163318)
出处
《计算机与数字工程》
2025年第6期1613-1618,共6页
Computer & Digital Engineering
关键词
孔隙度预测
双向门控循环单元神经网络
麻雀搜素算法
预测精度
porosity prediction
bidirectional gated recurrent unit neural network
sparrow search algorithm
prediction accuracy
