基于降维技术的烟叶质量可视分析方法  

Visual Analysis Method of Tobacco Quality Data Based on Dimension Reduction

作　　者：田东[1,2] 单桂华 迟学斌[1,2] 张艳玲 冯伟华[3] 王建伟 王爱国[3] 王锐[3] Tian Dong;Shan Guihua;Chi Xuebin;Zhang Yanling;Feng Weihua;Wang Jianwei;Wang Aiguo;Wang Rui(Computer Network Information Center,Chinese Academy of Sciences,Beijing 100190,China;University of Chinese Academy of Sciences,Beijing 100190,China;Zhengzhou Tobacco Research Institute,China National Tobacco Corporation,Zhengzhou 450001,China)

机构地区：[1]中国科学院计算机网络信息中心,北京100190 [2]中国科学院大学,北京100190 [3]中国烟草总公司郑州烟草研究院,郑州450001

出　　处：《系统仿真学报》2021年第9期2279-2288,共10页Journal of System Simulation

基　　金：中国烟草总公司烟草科研大数据重大专项(110201901029(SJ-08),110201901025(SJ-04))。

摘　　要：为了满足烟叶选材过程中跨区域匹配烟叶的需求,研发融合了降维和相关性分析方法的烟叶质量数据可视分析方法。通过基于香型区分类的降维算法、对比算法和针对烟叶质量数据的可视化交互方法,为研究人员提供了对烟叶质量数据进行探索式空间划分和相关性分析的可视化分析手段。全国烟叶质量数据分析案例和专家论证,证明该方法可以较好实现烟叶质量数据分析。In order to meet the requirements of tobacco leaf matching across regions in tobacco material selection,a visual analysis method of tobacco leaf quality data that incorporating dimension reduction and correlation analysis methods is developed.Through the dimension reduction algorithm,the comparison algorithm and the visual interaction method based on the classification of aroma area for tobacco leaf quality data,a visual analysis method for exploring space division and correlation analysis of tobacco leaf quality data is provided.National tobacco leaf quality data analysis cases and expert demonstrations show that the method can carry out the tobacco leaf quality data analysis well.

关 键 词：烟叶质量 降维 相关性 可视化 空间划分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]