基于BP神经网络的断层封闭性评价

Evaluation of fault sealing based on BP neural network

在前人关于断层封闭性研究基础上,总结当前评价断层封闭性的一些方法的不足,提出利用BP神经网络来评价断层封闭性的方法,分析影响断层封闭性的各种因素,把影响断层封闭性的因素分为定性和定量两类,并主要从定量上去考察断层的封闭性,优选出了三个因子来定量表征断层的封闭性,即:泥岩地层流体压力、断面正应力、断层泥比率(SGR),并以南海西部X油田实际断层资料为例,验证了该方法的适用性,并且认为该方法具有速度快、误差小、人为因素少且不用考虑输入神经元与输出神经元是否具有明确的关系等优点,是一种比较有效的断层封闭性评价方法。

Based on previous researches on fault sealing and deficiencies of some current evaluation methods for fault sealing,BP neural network was proposed to evaluate the fault sealing.According to analysis of all influencing factors on the fault sealing,the factors were divided into two qualitative and quantitative types.The fault sealing was mainly evaluated from quantitative study.Three factors were selected as quantitative characterization of fault sealing,which are fluid pressure in shale formation,normal stress of fault,and shale gouge ratio.The method applicability was verified by actual fault data of X Oilfield in the western South China Sea.The method has advantages of quick velocity,small error,little artificial factor and no consideration of the relationship of input neurons and output neurons,which is considered as a more effective evaluation method of fault sealing.