摘要
针对传统层次分析法的缺陷,提出结合因子分析、聚类分析的改进层次分析方法,应用到特体样板识别中。首先,通过因子分析确定各级指标的影响因素及其权重系数。然后,采用K-means聚类分析将特体样板聚为臀部、胸部和腹部等3类典型样板。最后,运用层次分析法确定各指标的权重,构建了包括6个二级指标、13个三级指标、3个四级指标的特体样板递阶层次结构模型。随机选取3个样本进行实证研究,以因子分析中提取的5个主成分为聚类指标,以K-means方法进行聚类,采用特体样板递阶层次结构模型识别被测样本与臀部、胸部、腹部样板的隶属度。结果表明,该方法可有效地表征特体样板的变异程度并识别所属的样板类别。
Aiming at the shortcomings of conventional analytic hierarchy process, an improved analytic hierarchy process based on combination factor analysis and cluster analysis was proposed and applied to special model recognition. Firstly, factor analysis was applied to determine the influencing factors and weight coefficients of each level of indicators. Then, K-means cluster analysis was applied to collect the special samples into three typical models such as hip, chest and abdomen. Finally, the analytic hierarchy process was applied to determine the weight of each indicator, and a special model-level hierarchical structure model including six secondary indicators, 13 third-level indicators and three four-level indicators was constructed. Three samples were randomly selected for empirical research. The five principal components extracted from the factor analysis were used as clustering indicators, and the K-means method was used for clustering. The specific sample-level hierarchical structure model was applied to identify the sample to be tested and hip and chest. The results show that the method can effectively characterize the variation of the special model and identify the model category.
作者
周捷
李健
马秋瑞
黄晓杰
ZHOU Jie;LI Jian;MA Qiurui;HUANG Xiaojie(School of Apparel and Art Design,Xi′an Polytechnic University,Xi′an,Shaanxi 710048,China;College of Information & Business,Zhongyuan University of Technology,Zhengzhou,Henan 450007,China)
出处
《纺织学报》
EI
CAS
CSCD
北大核心
2019年第5期124-130,共7页
Journal of Textile Research
基金
陕西省科技厅国际科技合作计划项目(2018KW-056)
陕西高等教育教改研究项目(17BZ037)
关键词
特体样板
因子分析
聚类分析
改进层次分析法
special template
factor analysis
cluster analysis
improved analytic hierarchy process