上海地区成年女性批量定制服装号型分类研究  

Investigation on Shanghai Adult Female Mass Customization Clothing Shape Classification

作　　者：齐雪良 袁惠芬[1] 王旭[1,2] 

机构地区：[1]"纺织面料"安徽省高校重点实验室安徽工程大学,安徽芜湖241000 [2]安徽工程大学纺织行业科技公共服务平台,安徽芜湖241000

出　　处：《武汉纺织大学学报》2016年第4期8-12,共5页Journal of Wuhan Textile University

基　　金："纺织面料"安徽省高校重点实验室开放基金(2015FZ001);安徽工程大学研究生创新项目(2015)

摘　　要：为确定批量定制服装合理板型数及提高合体度,以上海地区520名成年女性为研究对象,选择身高、胸围、腰围、领围为变量,提出了基于K-means聚类分析的非国标规格的号型分类方法。探讨了初始聚心的选择和最佳聚类数的确定,并以Calinski-Harabasz(CH)指标、变异系数和相对偏差为指标,对聚类效果进行评价。研究表明:以国标号型规格为初始聚心,号型被分为30类;相同CH指标时,以非国标规格号型为初始聚心,号型被分为20类,即板型数减少,且变异系数和相对偏差明显降低。In order to determine reasonable patterns number of mass customization and improve clothing fitness, this paper analyzed 520 Shanghai adult female body data by K-means cluster algorithm with four classified variables such as height, bust circumference, waist circumference and collar circumference.A new non-GB shape classification method based K-means cluster was proposed. Initial cluster centers and optimal clusters were discussed. Classification result in each cluster was evaluated in Calinski-Harabasz index, coefficient of variation and relative deviation. It showed that the number of clusters was 30 when the GB shape was as initial cluster centers. According to parity of Calinski-Harabasz principle, the number of clusters of non-GB standard shape was 20 and coefficient of variation and relative deviation was decreased obviously.

关 键 词：K-MEANS算法 批量定制 号型分类 聚类分析 

分 类 号：TS941-09[轻工技术与工程—服装设计与工程]