摘要
由于织物悬垂性能评价指标的多维性、数据聚类边界的模糊性,以及测量误差的不可避免,使得数据集通常会含有噪声点,而常用的FCM聚类算法无法消除噪声点对聚类中心的影响。为解决这一问题,提出了采用FPCM算法对悬垂性的测量值进行聚类分析,发现并剔除噪声点,从而更加客观地评价织物悬垂性,并通过实测数据验证了算法的准确性及有效性。
Aiming at the problems that dataset contains noisy points, drape evaluation indexes of fabric is multiple dimensions, data clustering boundaries is fuzzy and the unavoidable measurement error is unavoidable, and commonly used FCM clustering algorithm can＇t eliminate the serious effect of noisy points on clustering centers, Fuzzy Possibilistic C-Means Clustering algorithm is proposed to analyze and cluster the measured values, discover and eliminate noisy points, accordingly further evaluate fabric drape objectively. Simulation experiments conducted by FPCM algorithm valida- ted its correctness and effectiveness.
出处
《纺织科技进展》
CAS
2008年第6期43-45,共3页
Progress in Textile Science & Technology
基金
北京市教育委员会科技发展计划项目(KM200710012002)
关键词
悬垂性
模糊聚类
噪声点
评价
FPCM
drape
fuzzy cluster
noisy points
evaluation
Fuzzy Possibilistic C-Means Clustering （FPCM）
