Abstract: In this paper one of the pattern recognition methods,Fisher verdicts,is introduced fortreating the data of buried asphdalt coatings on pipes in the Daqing Oilfield.An extended lin-ear model and stepped verdicts are used to study the degradation of asphalt coatings. As a result,a simple expression with four variables:endurance(years);soil resistivity (Ω. m),saltcontent(%);and water content(%)is given as the criterion of coating degradation,Thecontributions of the above four variables on the classification of degradation degrees are alsoanalysed.In addition,the mathematical principles and the application methods of Fisher ver-djcts are introduced.

摘要： 本文介绍模式识别方法之一,Fisher判别法在处理大庆油田埋地石油沥青防护层老化状况现场调查数据中的应用;根据这类数据离散性大的特点,采用广义线性模型和分层判别法研究沥青老化规律,导出了以使用年份、土壤电阻率、含盐量、含水量为主要变量的判别沥青老化状况的数学表达式,讨论了各变量对分类的重要性和Fisher方法对调查数据的要求。文中对Fisher判别法的数学原理、应用技术也作了简要介绍。

关键词: 模式识别, Fisher判别, 数据处理, 石油, 沥青老化规律