基于Bayes判别理论的地下水化学分类的分析方法

李彬; 史海滨; 李祯; 张建国

李彬,史海滨,李祯,张建国.基于Bayes判别理论的地下水化学分类的分析方法[J].干旱地区农业研究,2015,33(4):246~250

基于Bayes判别理论的地下水化学分类的分析方法

An analyzing method for chemical classifications of groundwater based on the Bayes Discriminant Theory

DOI：10.7606/j.issn.1000-7601.2015.04.37

中文关键词: 内蒙古河套灌区地下水化学类型 Bayes判别分类

英文关键词:Inner Mongolia Hetao Irrigation District chemical type of groundwater Bayes discriminant classification

基金项目:国家“十二五”科技支撑计划项目（2011BAD29B03）；内蒙古科技攻关与政府专项科研项目（nsn200767）

作者	单位
李彬	内蒙古农业大学水利与土木建筑工程学院，内蒙古呼和浩特 010018；内蒙古农牧业科学院资源环境与检测技术研究所，内蒙古呼和浩特 010031
史海滨	内蒙古农业大学水利与土木建筑工程学院，内蒙古呼和浩特 010018
李祯	内蒙古农业大学水利与土木建筑工程学院，内蒙古呼和浩特 010018
张建国	巴彦淖尔市水利科学研究所，内蒙古巴彦淖尔 015000

摘要点击次数: 1581

全文下载次数: 941

中文摘要:

将Bayes判别分析方法应用于地下水化学类型判别与分类中，建立了地下水化学类型综合评判的Bayes判别分析模型。模型选用Na⁺+K⁺、Ca²⁺、Mg²⁺、HCO^-₃、Cl^-、SO₄^2-等6个指标作为判别因子；将地下水化学类型分为3种，作为Bayes判别分析的3个正态总体；以内蒙古河套灌区地下水实测数据作为训练样本，建立Bayes线性判别函数；以Bayes线性判别函数计算待判样品的Bayes判别函数值，以最大值对应的总体作为样品所归属的总体；最后以刀切法对判别准则进行评价以检验模型的优良性。结果表明，Bayes判别分析模型误判率低，识别正确率达82.5%，输出结果正确率达86.6%。与传统分类方法相比，Bayes判别分析模型提供的判别分类结果，具有更明晰、易懂的水化学类型信息。

英文摘要:

Bayes discriminant analysis method was applied into the discrimination and classification of groundwater chemical types to set up a Bayes discriminant analysis model for synthetic evaluation of groundwater chemical types. Six indexes including Na⁺+K⁺、Ca²⁺、Mg²⁺、HCO^-₃、Cl^-、SO₄^2- were selected as the differentiation parameters in this model. Chemicals in the groundwater could be classified into three types that were chosen as the three normal populations for Bayes discriminant analysis. Groundwater in Hetao irrigation district was utilized as the trial sample for data collection to set up the linear discriminant function of Bayes. The Bayesian linear discriminant function was then employed to calculate and determine the Bayes discriminant function values of the samples. The population with the maximal value was thereby used as the population for entire samples. Finally, in order to test the validity of this model, the cutting ring method was applied to evaluate the discriminate criterion. Results of this research showed that the misjudgment rate of this Bayes discriminant analysis model was low, reaching as high as 82.5% in correct recognition rate and 86.6% in output accuracy. Compared with traditional method, this Bayes discriminant analysis model could provide classification outcome that was clear and informative information for chemicals in water.

查看全文查看/发表评论下载PDF阅读器