摘要
模式识别就是研究已知的模式 ,并对模式进行分类 ,利用计算机技术 ,将待识别模式进行归类 ,达到分类的目的。模式识别的过程可概括为 :首先 ,应用傅立叶频谱分析、回归分析及多点平滑法进行数据预处理 ;其次 ,采用旋回地层学理论 ,计算已知钻井地质数据的频谱图 ,利用米兰柯维奇旋回组分进行学习 ,提取特征 ,建立模式分类 ;最后 ,用已建立的模式类对其他钻井进行地层划分。运用模式识别方法 ,对大港油田 4 0多口钻井进行了地层划分 ,划分结果与原地质分层结果具有很好的相关性 ,而且精度更高。
Pattern recognition is to subsume undetermined patterns with computer technologies based on study and classification of the known patterns.Pattern recognition can be performed in three steps.First,preprocess data through Fourier analysis,regression analysis and multi-point smoothing;then calculate frequency spectrum of the known wells' geological data by using cyclical strategraphy and perform feature study and extraction and pattern classification with Milankovitch cyclic components;and finally classify strata in other wells by using the established patterns.Strategraphic division has been performed in 40 wells in Dagang oilfield with the pattern recognition method and the results are in good agreement with the original ones and have better accuracy.
出处
《石油与天然气地质》
EI
CAS
CSCD
北大核心
2004年第1期102-105,114,共5页
Oil & Gas Geology
基金
国家重点基础研究发展规划项目 (G19980 40 80 0 )
关键词
模式识别
地层划分
旋回地层学
pattern recognition
stratigraphic division
cyclical stratigraphy
