人工神经网络在油层对比中的应用

Application of neural network in rsservoir correlation

本文根据油层对比的特点，模拟人工对比的思路，建立了利用神经网络技术进行油层对比的总体流程：①预处理；②用神经网络法进行油层对比；③结论检测；④绘制油层对比图。其中，结论检测为选择项。该方法能充分利用各种“显性”和“隐性”知识，通过学习建立潜在的对比规则实现油层对比。文中实例表明，所获得的油层对比剖面其结果合理、可信。这为地质家提供了一种新的油层对比方法。

According to reservoir correlation characteristic and modeled thinking of man'sreservoir correlation,we put forward neural network reservoir correlation method, which includes the steps: 1)preprocessing, 2) reservoir correlation using neural network, 3)conclusion prediction and 4) plotting the reservoir correlation result, here conclusion prediction being an option.The method can determine a potential rule for reservoir correlation by making full use of explicit and implicit knowledges and by self-learning. It is illustrated that the reservoir correlation in seismic sections is reasonable and reliable. This method is a new one for geologists' reservoir correlation.