基于模糊神经网络的储集层含水饱和度预测

Predicting Reservoir Water Saturation Based on Fuzzy Neural Networks

通过改进补偿模糊神经网络,使模糊系统较强的知识表达能力与神经网络强大的自学习能力优势互补。然后,提出了一种动态调整学习步长的机制,能够避免较大震荡现象的出现;同时加快了迭代速度。最后,将该方法应用到油田测井解释中预测储集层含水饱和度,结果较为满意;与常规神经网络相比,迭代速度和误差精度都有大大的提高;实践证明该方法值得进一步推广运用。

A kind of improved Compensation Fuzzy Neural Networks can impersonate the advantages of fuzzy system with ability to be prone to express knowledge and the advantages of neural networks with fairly strong self-adaptive ability. Then, a mechanism that can dynamically adjust the learning step is presented. So the sway phenomenon can be minimized and the learning step can be quickly speeded. Finally, the system is applied to predicting reservoir water saturation in logging interpretation of oil field. The result of experiment is satisfying. Compared to conventional neural networks, the convergence speed and the error precision are improved a lot. Practice has proved that the method is worth further extending.