CAJ | 학술논문

基于野外采样和室内分析相结合的方法，采用电感耦合等离子体质谱法(ICP-MS)对羊草沟煤矿研究区表层土壤样品中的 Cd、Cr、Zn、Pb和Cu含量进行测定，应用非线性支持向量机模型中的分类支持向量机，选用sigmoid核函数，利用MATLAB编写程序，进行土壤环境质量评价，并利用模糊综合评判法对评价结果进行验证.在此基础上，运用对应分析方法对样品和变量进行了关联分析，进一步了解重金属污染特征.评价结果表明，研究区土壤环境质量多为Ⅰ类，与模糊综合评判法的相同率达到91.67%，将支持向量机用于土壤环境质量评价是可行的.相比于传统的评价方法，支持向量机采用结构风险最小化原则，将复杂的非线性问题转化为线性问题，成功的解决了多分类、高维运算等问题.
기우야외채양화실내분석상결합적방법，채용전감우합등리자체질보법(ICP-MS)대양초구매광연구구표층토양양품중적 Cd、Cr、Zn、Pb화Cu함량진행측정，응용비선성지지향량궤모형중적분류지지향량궤，선용sigmoid핵함수，이용MATLAB편사정서，진행토양배경질량평개，병이용모호종합평판법대평개결과진행험증.재차기출상，운용대응분석방법대양품화변량진행료관련분석，진일보료해중금속오염특정.평개결과표명，연구구토양배경질량다위Ⅰ류，여모호종합평판법적상동솔체도91.67%，장지지향량궤용우토양배경질량평개시가행적.상비우전통적평개방법，지지향량궤채용결구풍험최소화원칙，장복잡적비선성문제전화위선성문제，성공적해결료다분류、고유운산등문제.
This paper presented a study on the soil environment quality analysis with support vector machine method (SVM) at the Yang-cao-gou coal mine area (Jilin province, China). Incorporating field investigation and laboratory analysis, copper and lead in soil samples were measured by using inductively coupled plasma mass spectrometry (ICP-MS), the pollution characteristics of five soil heavy metals, Cd、Cr、Zn、Pb and Cu were analyzed. The nonlinear SVM classification model was employed to evaluate soil environmental quality by sigmoid kernel function programmed with MATLAB codes, and the validation process for the evaluation results was performed with fuzzy comprehensive evaluation method. Meanwhile corresponding analysis was applied to investigate the main pollution factor in each soil partition of the study area, considering the variable load size and the relationship between variables and sampling point partition. The results showed that soil environmental quality ranked almost in theⅠclass in the study area, SVM method obtained almost the same results compared with fuzzy comprehensive evaluation method with a similarity rate of 91.67%, demonstrating that the method (SVM) method is appropriate for soil environmental quality assessment. Compared with conventional assessment methods, SVM adopted the structural risk minimization principle, which resolved the problem of multi-classification, high dimensional algorithm through nonlinear to linear transfer.