基于微粒群算法的神经网络储层物性参数预测
|
摘要
建立了一个具有自适应、复杂非线性储层预测模型,在计算方法上,由于多层前向型神经网络BP算法存在易陷入局部最优的缺点,而微粒群算法具有较强鲁棒性和全局收敛的优点。结合二者长处,利用基于微粒群算法的神经网络计算方法,对神经网络结构进行了改进。利用四川洛带地区气田的测井资料,用所设计的算法对储层的物性参数(孔隙度、渗透率)进行预测,并对其预测精度与用常规基于BP算法和基于LMBP算法得到的预测结果进行了比较分析,发现地质效果明显,有效地克服了基于BP算法和基于LMBP算法的缺点。
A Predictive reservoir model with the self-adoption and complicated nonlinear property is set up. Because Multi-layer Feed Forward Neural Networks BP Algorithm exists weakness of getting bogged down in the local optima, stronger robustness and global convergence of PSO Algorithm, this research makes use of the particle swarm optimization (PSO) to improve the neural network, then, on the basis of the Luodai gas field in Sichuan Province, by the computation methods of PSO of the neural network, the reservoir characters(such as porosity, permeability) is forecasted, also the precision is tested and is compared with traditional computation methods of BP and LMBP, by which a obvious geography efficiency superior to traditional explanation methods is obtained, the shortcoming base from BP and LMBP algorithm are effectively overcome.
A Predictive reservoir model with the self-adoption and complicated nonlinear property is set up. Because Multi-layer Feed Forward Neural Networks BP Algorithm exists weakness of getting bogged down in the local optima, stronger robustness and global convergence of PSO Algorithm, this research makes use of the particle swarm optimization (PSO) to improve the neural network, then, on the basis of the Luodai gas field in Sichuan Province, by the computation methods of PSO of the neural network, the reservoir characters(such as porosity, permeability) is forecasted, also the precision is tested and is compared with traditional computation methods of BP and LMBP, by which a obvious geography efficiency superior to traditional explanation methods is obtained, the shortcoming base from BP and LMBP algorithm are effectively overcome.
引文
[1]吴大奎,李亚林.地震、测井资料联合反演的神经网络算法研究[J].天然气工业,2004,24(3):55-57.
[2]陶淑娴,肖慈,杨斌,等.神经网络在测井解释中的应用[J].石油物探,1995,34(3):90-108.
[3]刘争平,何永富.人工神经网络在测井解释中的应用[J].地球物理学报,1995(38):323-330.
[4]黄述旺,窦齐丰,彭仕宓,等.BP神经网络在储层物性参数预测中的应用———以梁家楼油田沙三中为例[J].西北大学学报,2002,32(3):271-274.
[5]杨丽娜,解国军.油气资源丰度预测的人工神经网络方法———以济阳拗陷为例[J].石油天然气学报.2007,29(1):55-58.
[6]杨斌.非线性测井储层特性智能反演方法及应用研究[D].四川成都:成都理工大学,1999.
[7]王文娟,刘诚.储层物性参数和储层特性的预测模型[J].工程数学学报,2005,22(8):25-28.
[8]Kennedy J,Eberhart R C.A discrete binary version of the particle swarm algorithm[C].Proceedings of the IEEE congress on systems,man,and cybernetics.piscataway,NJ:IEEE Service Center,1997,4104-4109.
[9]曾建潮,介婧,崔志华.微粒群算法[M].北京:科学出版社,2004.
[2]陶淑娴,肖慈,杨斌,等.神经网络在测井解释中的应用[J].石油物探,1995,34(3):90-108.
[3]刘争平,何永富.人工神经网络在测井解释中的应用[J].地球物理学报,1995(38):323-330.
[4]黄述旺,窦齐丰,彭仕宓,等.BP神经网络在储层物性参数预测中的应用———以梁家楼油田沙三中为例[J].西北大学学报,2002,32(3):271-274.
[5]杨丽娜,解国军.油气资源丰度预测的人工神经网络方法———以济阳拗陷为例[J].石油天然气学报.2007,29(1):55-58.
[6]杨斌.非线性测井储层特性智能反演方法及应用研究[D].四川成都:成都理工大学,1999.
[7]王文娟,刘诚.储层物性参数和储层特性的预测模型[J].工程数学学报,2005,22(8):25-28.
[8]Kennedy J,Eberhart R C.A discrete binary version of the particle swarm algorithm[C].Proceedings of the IEEE congress on systems,man,and cybernetics.piscataway,NJ:IEEE Service Center,1997,4104-4109.
[9]曾建潮,介婧,崔志华.微粒群算法[M].北京:科学出版社,2004.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心