RBF神经网络在储层表征中的应用研究

研究利用RBF神经网络技术进行石油储层表征中有关储层参数的计算与岩性的识别;建立了储层参数(渗透率)预测模型与岩性识别模型,并利用该两个模型对未知样本进行预测,预测结果与实际测量结果相比具有较好的一致性,其渗透率预测精度与收敛速度较BP神经网络模型有了很大的提高;应用表明,RBF神经网络在储层表征问题中有着广阔的应用前景。

BRF neural networks for petroleum reservoir characterization

This paper studies the parameter calculation relating to petroleum reservoir characterization and lithologic identification based on RBF neural networks.Two models for reservoir parameters(permeability) prediction and lithologic identification have been built and they are applied to predict the unknown samples.The prediction result for reservoir permeability has higher consistency with the practical cases.The prediction and identification precision have been greatly improved compared to the traditional BP neural networks.The results show that the RBF neural network is very promising for the application of petroleum reservoir characterization.