BP神经网络补偿算法在煤层气井产量预测中的应用

Application of BP Neural Network Compensation Algorithm in the Production Forecasting of CBM Wells

为了精细监测和了解排采过程煤储层参数的动态变化,本文提出了一种基于BP神经网络补偿算法,对未来一定时期的产气、产水量进行了预测。对大佛寺典型的煤层气水平井(DFS-C02井)进行实例分析,结果表明,未来30d的产水量、产气量的平均相对误差分别为0.79%(0.07~0.26%)和0.72%(0.01~2.4%),预测结果较准确。BP神经网络补偿算法为煤层气井的产量预测提供了一种新方法,同时为排采工作制度提供依据。

In order to carefully monitoring and understanding the dynamic change of the coal reservoir parameters during the mining process,this paper proposes a compensation algorithm based on BP neural network to forecast the future production of gas and water.Analysis of a typical CBM horizontal well(DFS- C02 Well) in Dafosi shows that,the average relative error of forecasted water and gas production is respectively 0.79%(0.07~0.26%) and 0.72%(0.01~2.4%) in the next 30 days,which are accurate predictions.BP neural network provides a new method for the production prediction of CBM wells,as well as provide the basis for development working system.