基于BP神经网络的三维地质参数预测及立体图显示

基于现代神经网络数理统计新技术，提出一种简易可行的ＢＰ神经网络三维估值模型，以此可根据研究区域的已知资料对其未知点处的储层参数值进行高分辨率的预测；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．最后成功地处理了Ｓ盆地中部气田马家沟组地层压力数据，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推广使用．

Prediction and Display of 3-D Geologic Parameters Based on BP Nerve Network

On the basis of the modern nerve network mathematical statistics ,a simple and feasible mod-el is established for estimating the values of 3-D geologic parameters. By means of the model ,the values of the reservoir parameters at a point can be predicted according to known information of reservoir. And the estimated values can be displayed with 3-D isopleth map. At last,the formation pressure data of Majiagou group of a gas field in the middle of S Basin is processed with the method. The result shows that themethod has high resolution ,it is a effectivc value estimating method and is worth being spreading.