三维地质数据的神经网络处理及立体图形显示

NERVE NETWORK PROCESSING AND STEREOGRAPHIC DISPLAY OF 3 D GEOLOGIC DATA

在油气藏三维精细描述工作中，常需要对一些三维四变量（Ｘ，Ｙ，Ｈ，Ｚ）地质数据进行分析处理。传统的处理方法多为方位近点距离加权模型和趋势回归模型及克里金模型等。基于现代神经网络信息处理新技术，提出一种简易可行的ＢＰ神经网络三维估值模型，由此可根据研究区域内的已知资料对其未知井点处的储层参数值进行高分辨率估值；采用立体等值线图可视化技术来实现该模型估值的直观图形显示，充分揭示三维地质体某种特征参数的纵横向变化规律。通过对Ｓ气田马家沟组的地层压力等数据的分析处理，结果表明该法是一种有效的高分辨率估值技术，在油气藏描述中值得借鉴使用。

On meticulously describing oil gas reservoirs,it is necessary to analyse and process 3 D tetravarible(X,Y,H,Z)geologic data by traditional methods such as azimuth near point range weighting model,trend regression model,Kriging model etc.On the basis of modern nerve network datum processing techniques, an available BP nerve network 3 D estimation model,i.e.the high resolution appraisal of unknown well reservoir parameters can be done by the known data in research areas, is proposed out.The visual graphic display of the model estimated values can be accomplished by stereo isogram visibility technique and the variation laws of 3 D geological characteristic parameters in vertical and lateral directions can be fully delineated.It is proved that the model is available to high resolution appraisal of stratigraphic pressure data in Majiagou formation of S gas field.