利用遗传神经网络识别水淹层

Application of genetic neural network to waterflooded zone identification

胜坨油田水淹非常严重,有效的识别水淹层并定量地划分水淹级别,对油田开发非常重要。利用遗传神经网络实现了水淹层的自动识别。根据关键井的统计数据,首先建立标准的流体测井相数据库,合理地划分储层水淹级别,然后利用遗传神经网络对已知样本进行训练。遗传神经网络比简单的BP神经网络鲁棒性强,收敛速度快。用训练好的遗传神经网络对胜坨油田的水淹情况进行研究。

abstract It is very important for oilfield development to identify waterflooded zones effectively and classify its level quantitatively because of seriously water out in Shengtuo oilfield. Automatic identification of wateredout zones is realized by applying genetic neural network. According to the statistic data of the key wells, at first, a standard database of fluid logging facies is constructed, and wateredout level of reservoirs is classified rationally, then known samples are trained using genetic neural network. Genetic neural network is more robust and easily convergent than simple BP neural network. A trained genetic neural network is powerful in the study of the waterflooded zones in Shengtuo oilfield.