近红外光谱技术结合支持向量机法在保水剂鉴别工作中的应用  

Study on identification of super absorbent polymer by near infrared reflectance spectroscopy and support vector machine model

作　　者：岳征文[1] 王红柳[2] 成格尔[3] 赵铭军[4] 王百田[1] 

机构地区：[1]北京林业大学水土保持学院,北京100083 [2]北京林业大学草地资源与生态实验室,北京100083 [3]内蒙古农业大学生态环境学院,呼和浩特010019 [4]中国石化管道储运公司徐州输油处,徐州221008

出　　处：《干旱区资源与环境》2012年第4期172-175,共4页Journal of Arid Land Resources and Environment

基　　金：国家"十一五"科技支撑计划(2006BAD03A0301)困难立地工程造林新材料;新产品及应用技术;林业行业公益项目(200704031)资助

摘　　要：基于近红外光谱技术结合应用主成分回归法(PCR)、偏最小二乘法(PLS)、BP神经网络法、支持向量机(SVM)等四种方法对来自不同国家和地区的24种聚丙烯-聚丙烯酰胺型保水剂进行了品种的鉴别研究。结果表明:近红外光谱技术结合SVM法可以有效的进行保水剂分类鉴别工作。当光谱范围选择5000cm-1～9000cm-1,经验参数c=8,g=0.0313时,PCA-SVM模型的预测准确率可以达到100%。研究证明此种方法可以应用于保水剂品种的鉴别。This studies identified the varieties of 24 kinds of Polypropylene - polyacrylamide from different countries and regions by near infrared spectroscopy （NIS） technology companied with four kinds of methods, princi- pal component recession （PCR）, partial least squares （PLS）, BP neural network, support vector machine. Theresults showed that method of NIS technology combined with SVM was efficient work on identifying different SAP. Prediction accuracy of PCA , SVM model was got to 100% when spectroscopy scope between 5000cm^-1 with 9000cm^-1, and experience parameters c = 8 , g = 0. 0313. The conclusion is that PCA - SVM model can effi- ciendy apply to identify type of SAP, which is the best prediction model in four methods.

关 键 词：近红外光谱法 支持向量机 保水剂 高级吸水树脂 

分 类 号：S157.9[农业科学—土壤学]