基于神经网络的复杂储层流体分级识别

Fluid hierarchical identification of complex reservoir based on neural network method

YD油田具有束缚水饱和度高、地层水矿化度高、黏土矿物含量高、油气水分布规律复杂,以及无统一油气水界面的特征,储层中不同流体的测井响应特征区别不明显,采用常规测井图版无法准确识别油层、气层、油气层,以及低电阻率油层。文中通过选择相关性强的测井参数,应用神经网络建立分级解释模型,实现了对复杂储层中不同流体的自动化、准确识别。研究结果表明,基于神经网络的储层流体分级识别技术,成功识别了油层、气层、油气层,以及低电阻率油层,解决了复杂储层的流体识别问题,并成功应用于YD油田开发。

With the characteristics of high irreducible water saturation,high-salinity formation water,high clay content,complex distribution regularity of oil,gas and water,no uniform oil/gas/water contact,and no obvious logging response characteristics in the complex reservoir in YD oilfield,conventional logging identification methods are impossible to accurately distinguish oil layer,gas layer,oil and gas layer,and low resistivity oil layer.Selecting highly-correlated well logging parameters,and applying neutral network method to hierarchical identification of complex reservoir,the reservoirs fluids can be identified automatically and accurately.The study shows that fluid identification technology of complex reservoir based on neural network successfully identified the layers of oil,gas,oil and gas,and low resistance oil,and solved fluid identification problem of complex reservoir and it was applied to YD oilfield development successfully.