关键词: 人工神经网络, BP算法, 渗透率预测, 岩性控制, 储层物性, 涠西南凹陷

Abstract:

Permeability is one of the most important parameters in reservoir estimation.Compared with the calculated result by traditional experimental or statistical models,the BP neural net model can more accurately predict permeability because of its high nonlinear mapping ability and very strong abilities of self-adaptation and self-study.The present paper established a nonlinear model among reservoir property parameters,logging response and lithology by improving the conventional BP model,i.e.applying the quantitative lithology parameter to the BP model as a study sample.The permeability of the Liu-1 member of the Weizhou 11-7 oilfield in the Weixinan Depression,Beibuwan Basin was predicted by applying this method and the result was comparatively consistent well with the actually measured permeability,moreover,the precision of this method was much better than that applied by the conventional BP model without domination of lithology.Besides the application in the prediction of reservoir parameters,this method could be widely used in predicting reservoir microfacies and lithology as well.

Key words: artificial neural net, BP algorithm, permeability prediction, domination of lithology, reservoir properties, Weixinan Depression