基于鲸鱼算法优化AM-BiLSTM模型的储层产气量预测

Gas production prediction using AM-BiLSTM model optimized by Whale Optimization Algorithm

产气量是评估天然气井生产能力和开发工艺效果的重要指标。准确的预测产气量是保证高效生产的关键。为了准确预测储层产气量,本文提出了一种基于鲸鱼优化算法(whale optimization algorithm,WOA)、注意机制(attention mechanism,AM)和双向长短时记忆(bi-directional long short-term memory,BiLSTM)相结合的日产气量预测模型。首先,以Volve油田的排采数据为研究对象,分析了产气量与这些排采参数之间的相关性,并利用LightGBM算法进行重要性排序;然后使用有较强非线性处理能力的基于注意力机制的双向长短时记忆神经网络(bi-directional long short-term memory based on attention mechanism,AM-BiLSTM)构建日产气量预测模型;最后通过鲸鱼算法对AM-BiLSTM模型中的相关参数进行优化,并将参数优化后的模型(WOA-AMBiLSTM)应用于Volve油田的A井。实验结果表明,WOA-AMBiLSTM模型的综合预测性能优于传统的反馈神经网络模型(the back-propagation neural network model,BP)和其他提出的深度学习模型(LSTM、BiLSTM和AM-BiLSTM)。WOA-AM-BiLSTM模型预测曲线与实测测井曲线更加接近,具有更好的预测表现,为储层产能预测提供了一种新思路。

Gas production is an essential metric to assess the gas production capacity and engineering development performance.Accurate gas production prediction is key to ensuring effcient production.To accurately predict gas production,a daily gas production prediction model developed by combining the whale optimization algorithm(WOA)and bi-directional long short-term memory using the attention mechanism(AM-BiLSTM)is proposed.First,the correlation between the daily gas production and discharge parameters is analyzed based on the discharge data of the Volve oilfeld,and the importance is ranked by the LightGBM algorithm.Then,a daily gas production prediction model is created using the AM-BiLSTM algorithm.Finally,WOA is applied to optimize the hyperparameters of the network structure of the AMBiLSTM model.The experimental results show that the comprehensive performance of the WOA–AM-BiLSTM model is better than that of the back-propagation neural network model(BP)and deep learning models(LSTM,BiLSTM,and AM-BiLSTM).The prediction curve obtained using the WOA–AM-BiLSTM model is closer to the measured Well logging curve and has better prediction accuracy.Therefore,the proposed method provides an effective method for predicting gas production and designating a reasonable drainage system.