多参数油气预测系统

Multiparameter hydrocarbon prediction system

本文针对目前国内外多参数油气预测研究的现状，提出并建立了一套结合地质、测井资料以及地震资料进行多参数油气预测系统，该系统包括三个部分，即多参数提取、特征参数压缩和油气预测。在多参数提取中，除吸收和利用了目前比较实用的常规的参数外，还对关联维参数的求取作了改进；首次利用突变论和小波分析进行特征参数提取。特征参数压缩是在多参数提取的基础上，利用K－L变换进行特征空间维数的压缩。油气预测是将BP网络和SOM网络联合起来，组成一个预测系统进行多参数油气预测。BP网络主要适用于多井地区，而SOM网络主要应用于少井地区的多参数油气预测，从而使这套油气预测系统能在不同勘探程度的地区应用。

In view of the research situation of worldwide multiparameter hydrocarbon prediction,we put forward the multiparameter hydrocarbon prediction system that analyses geological, logging and seismic data comprehensively. The system works in three steps: multiparameter extraction, characteristic parameter reduction and hydrocarbon prediction. In multiparameter extraction, the usual parameters are still used,the taking of correlation dimension parameter is improved,and abrupt variation theory and wavelet analysis are applied to characteristic parameter extraction for the first time. The characteristic parameter reduction is to reduce characteristic dimensions by making K-L transform after multipararneter extraction. In hydrocarbon prediction,BP network is combined with SOM network to form the multiparameter hydrocarbon prediction system. BP network is applicable to some multi-borehole areas,and SOM network to sparse-borehole areas. With the combination of the two networks,the multiparameter hydrocarbon prediction system can be used desirably in various areas where exploration degrees are different.