论文部分内容阅读
利用误差逆传播(Error Back Propagation)神经网络技术建立了适用于储层孔隙介质气体吸附的误差逆传播吸附模型(Er-ror Back Propagation Adsorption Model)。该模型同时考虑了储层孔隙介质本身的非均质性和吸附相结构的非均质性,并将这些非均质性信息分散记忆并贮存到模型的网络结构中,因而从理论上避免了传统吸附模型不能全面“包容”气体——储层孔隙介质吸附体系非均质特征的缺陷。最后,给出了EBPAM模型针对储层孔隙介质三元混合气和五元混合气吸附模拟的计算实例,用以证实EBPAM针对储层孔隙介质气体吸附的适用性。
The Er-ror Back Propagation Adsorption Model for gas adsorption in reservoir pore media was established by using Error Back Propagation neural network technology. The model takes into account both the heterogeneity of the pore media in the reservoir and the heterogeneity of the adsorbed phase structure, and disperses and stores these inhomogeneous information into the network structure of the model, thus theoretically avoiding The traditional adsorption model can not fully “contain” the defects of heterogeneous characteristics of the gas-reservoir pore media adsorption system. Finally, the calculation example of the EBPAM model for the simulation of the adsorption of ternary mixed gas and five-component mixed gas in porous media is given to verify the suitability of EBPAM for gas adsorption in porous media.