摘要
图版拟合是现代试井分析的主要方法,由于图版上典型曲线分辨率的限制,使得典型曲线图版拟合分析不可避免地存在误差。除此之外,复杂试井问题的图版分析常常同时需要多幅图版,因而限制了典型曲线图版拟合分析方法的应用。利用前馈神经网络的函数逼近性质,建立了神经网络典型曲线图版,并给出了相应的解释方法。所建立的神经网络图版包含了全部理论曲线的信息,克服了传统典型曲线图版的不足,使得任意的实测曲线都能在神经网络典型曲线图版上直接得到精确拟合。同时,该图版大大简化了传统图版的拟合分析过程,使图版拟合分析更易实现。神经网络图版的研制成功,使实现图版拟合分析的智能化成为可能;另外,利用神经网络的函数映射能力,建立的神经网络图版可以包含更多信息,从而可以解释出更多的参数值。图3参4(邓远忠摘)
Type curve match analysis is the main method of modern well test interpretation, but the error of this method is unavoidable because of the limitation of the type curves in a sample plot. Besides this, it usually needs a few of type curve sample plots to deal with the complex well test problem, so the using of this method is actually limited. In this study, by means of the property of artificial neural network (ANN) that any function can be approached by a three layer ANN, a ANN type curve sample plot has been established and the analysis method is also delivered. This ANN type curve sample plot includes all curves message, and the error of the traditional method is eliminated. Any well test data can get an accurate match in this ANN type curve sample plot. The analysis process is also simplified. The establishment of ANN type curve sample plot makes the intelligence realization of type curve match analysis possible. On the other hand, we can make the ANN type curves including more message, so as to obtain more parameters through once analysis.
出处
《石油勘探与开发》
SCIE
EI
CAS
CSCD
北大核心
2000年第1期64-66,共3页
Petroleum Exploration and Development
关键词
试井解释
图版拟合
神经网络
采油井
Well test interpretation, Typical curve, Typical curve graph, Fitting, Nerve network, Function, Parameter
