基于实数编码的遗传神经网络水淹层综合判别

Automation distinguishing oil from water bearing reservoirs based on real coding genetic neural network

为解决油层水淹后 ,使用一般测井解释方法很难识别水淹层的问题 ,提出了一种基于实数编码的遗传 BP ( Back Propagation)神经网络进行水淹等级综合判别的方法。在此基础上建立的神经网络经反复训练验证 ,性能稳定、学习收敛速度快 ,同时有很强的记忆能力、推广能力和适应性。对大庆长垣 7口井的资料处理表明 ,准确率达到 87%以上 ,取得了很好的效果。

To solve difficult problem of that to identify the flooded formation with the conventional interpretation methods, a new method of automation distinguishing oil bearing reservoirs from water bearing reservoirs based on real coding genetic and BP algorithm and neural network technology is given. With many times train, the result proved: the network keeps properties of being stable; fast learning rate converged; awfully memorable, generalized and adaptive abilities. Testing on 7 wells of Changyuan of Daqing oil field, the accuracy reach 87%,the result is nice.