基于伽马射线技术的油井含水率神经网络预测  

ANN W atercut Prediction of Oil Wells Based on Gamma Ray Technology

作　　者：郑永建 张振朝 牛棚满 文鹏荣 曾桃 ZHENG Yongjian;ZHANG Zhenchao;NIU Pengman;WEN Pengrong;ZENG Tao(Zhanjiang Branch,CNOOC(China)Co.,Ltd,Zhanjiang 524057.China;Haimo Technologies Group Co.,Ltd.,Lanzhou 730010,China)

机构地区：[1]中海石油(中国)有限公司湛江分公司,广东湛江524057 [2]海默科技(集团)股份有限公司,甘肃兰州730010

出　　处：《自动化仪表》2021年第3期21-24,29,共5页Process Automation Instrumentation

基　　金：广东省海洋经济发展(海洋六大产业)专项基金资助项目(粤自然资合[2020]027)。

摘　　要：伽马射线含水分析仪已广泛应用于油井含水率测量。但由于探测器采集到的射线计数与理论值存在一定的误差,使得在实际应用中油井含水率的测量结果也与真实值存在一定的偏差。为了消除该误差提高含水精度,首次通过人工神经网络对含水仪的高、低能伽马计数率与实际化验的含水率数据进行分析与训练,将高、低能伽马计数作为网络的输人数据,含水率与含气率作为网络的输出数据对网络进行训练并使用训练的人工神经网络对含水率进行预测。测试结果表明,网络预测含水率与化验含水率的绝对误差约为±2%,含气率绝对误差小于±1%。基于神经网络结合高、低能伽马计数与含水率分析的方法将在油田提高测试含水率精度发挥重要的作用。The watercut meter based on the gamma-ray technology has been widely used in oil wells.However,as the gamma counting collcted by the detector gives a certain deviation with theoretical value,the measured oil well watercut in practical application presents a certain deviation compared with the true readings.To eliminate this error and improve the accuracy of watercut measurement,the measured watercut and sampling watercut are analyzed to build an,artificial neural network(ANN),for the first time.The built ANN is mainly used to predict watercut of the oil wells.The high energy counts and low energy counts of the detector are used as input data of the network,while the watercut and gas volume fraction,GVF,are used as output data of the network.The results show that the prediction data of the proposed ANN model has a good agreement with the oil field sampling data.The deviation between the predicted watercut of the network and the reference is about±2%,and the deviation of the GVF is less than±1%.The new method based on neural network combining high and low energy gamma counts and watercut analysis will play an important role in improving the accuracy of testing watercut in oil fields.

关 键 词：原油 含水率 含气率 含水分析仪 神经网络 多相流 伽马射线 测试精度 

分 类 号：TH-39[机械工程]