过程神经网络在厚层细分水淹解释中的应用  

Application of Process Neural Network to Water-flooded Zone Recognition in Thick Zone Subdivision

作　　者：宋延杰[1] 杨艳[1] 杨青山[2] 马宏宇[2] 

机构地区：[1]大庆石油学院地球科学学院,黑龙江大庆163318 [2]大庆油田有限责任公司勘探开发研究院,黑龙江大庆163712

出　　处：《测井技术》2009年第4期340-344,共5页Well Logging Technology

摘　　要：针对油田注水开发中后期油层水淹状况复杂,储层性质发生改变,导致测井曲线的幅度和形态发生相应的变化,建立一种能描述测井曲线幅度和形态变化的模式识别技术,有利于储层水淹级别的准确判别。在分析水驱后储层性质变化规律及水淹层测井响应特征基础上,提取厚油层的曲线形态特征,弥补了由于厚层细分引起的曲线形态信息缺失。选取曲线形态特征参数、原始测井曲线以及成果曲线作为识别水淹层的特征参数,建立了基于过程神经网络的水淹层自动识别方法。应用7口取心检查井的176个样本建立水淹层模式库,进行网络训练,使用训练好的过程神经网络对大庆油田北1-55-检E66井等2口井进行水淹解释,结果表明,解释符合率为81.3%,该方法可提高水淹层测井评价的精度。Since long-term water-flooding changes reservoir property in middle or late period of oil field development,the amplitude and shape of logs will change correspondingly.A pattern recognition technology describing the amplitude and shape of logs is used to improve recognition accuracy for the grades of watered out zones.In accordance with analysis of reservoir property and character of log response for water-flooded zones,parameters describing shape of logs from thick zones are extracted to keep completeness of shape description in thick zone subdivision.Finally,log-shape parameters,log,and log interpretation result parameters are picked out,and a new pattern recognition technology is used to determine the grades of water-flooded zones automatically based on process neural network.Mode base for five grades of water-flooded zones is established with 176 samples from 7 coring wells to train network.The interpretation result for two wells in Daqing Oilfield indicates that the average interpretation accuracy is 81.3%,and this method can improve evaluation precision of grades of water-flooded zones.

关 键 词：测井解释 特高含水期 厚层细分 曲线形态特征 过程神经网络 水淹级别 

分 类 号：P631.84[天文地球—地质矿产勘探]