摘要
人工神经网络是近期发展最快的人工智能领域研究成果之一.本文在介绍BP神经网络的有关原理的基础上,提出一种基于BP神经网络模型的波阻抗反演方法,该方法克服了常规基于模型的波阻抗反演方法严重依赖于初始模型的选择和易陷入局部最优等局限性.利用该方法对实际地震剖面进行了波阻抗参数反演处理,结果表明人工神经网络方法在波阻抗反演中的应用是可行的并且是有效的.
Artificial neural network is one of study fruits in artificial intelligence field which is being developed very quickly at present time. Based on introducing the theory of BP network, this paper has put forward a impedance inversion method using BP network model to solve some problems such as the initial model sensitivity, frequent local traps for the conventional impedance inversion based on model. The result of real data indicates the method is effective and getting the satisfying high precision inversion results.
出处
《地球物理学进展》
CSCD
北大核心
2005年第1期34-37,共4页
Progress in Geophysics
基金
中国科学院知识创新重大项目(KZCX1 SW 18 6) 资助.
关键词
神经网络
波阻抗反演
地震资料
测井
neural network, impedance inversion, seismic data, logging