摘要
本文介绍一种基于概率神经网络(PNN)的拟声波曲线构建方法。首先论述了基本PNN的数学模型,在此基础上,设计了适合于多源测井信息融合的多输入、单输出PNN网络拓扑结构,并利用PNN的插值功能以拟合误差最小为准则推导出新模型的输出;最后利用该模型对实际测井资料进行处理,能够快速、自适应构建拟声波曲线。通过对处理结果的分析,验证了该方法的合理性和有效性。
An approach to pseudo-acoustic log curve construction based on probabilistic neural networks(PNN) is presented in the paper.Firstly,the basic mathematical model of PNN is discussed.Then,a multi-input and single-output PNN network topology which is fit multi-source logging information fusion is designed on the basis of the model.After that,the output of the new model with the minimum fit error criterion is derived using PNN interpolation function.Finally,the actual logging data is processed by the model,and a fast pseudo-acoustic curve construction is adaptively obtained.The processing results show the rationality and effectiveness of the method.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
2012年第5期803-807,844+681-682,共5页
Oil Geophysical Prospecting
基金
国家自然科学基金项目(40874066
40839905
41274127)资助
关键词
概率神经网络
模型设计
测井信息融合
拟声波构建
probabilistic neural network,model design,logging information fusion,pseudo-acoustic log construction
