基于遗传神经网络的煤层气含量预测

摘要

The prediction of coalbed Gas content is a key issue in coalbed gas exploration .As the traditional prediction technique of linear regression can not reveal the intrinsic relationship between impact factors and coalbed gas content , a new nonlinear intelligent algorithm is put forward based on the genetic -BP neural network .In the algorithm ,the neural network is used to learn and the genetic algorithm is used to optimize the link weights and threshold value of neural network . This method can effectively avoid falling into local minimum and it has the feature of global optimization .The research shows that the prediction accuracy of new intelligent algorithm is relatively higher than that of the BP neural network .Therefore ,this new method can provide a reliable basis for the prediction of coalbed gas content in the future .%指出了煤层气含量预测是煤层气勘探开发的一个关键问题，传统的线性回归预测方法已不能反映各影响因子和煤层气含量之间的内在关系，据此提出了基于遗传－BP神经网络的非线性新型智能算法。其中，神经网络用来学习，遗传算法用来优化神经网络的连接权值及阈值，该方法能有效避免陷入局部极小值，具有全局寻优的特点。研究表明：构建的新型智能算法相对标准的BP网络预测精度要高，该方法为今后煤层气含量的预测提供了可靠依据。