改进OSELM的服用织物折皱智能分类  

Improving the Intelligent Classification of Crease Patterns in OSELM Fabric Folding

作　　者：童振辉 TONG Zhenhui(Light Industry Department,Quanzhou Vocational College of Economics and Business,Quanzhou,Fujian 362000,China)

机构地区：[1]泉州经贸职业技术学院轻工系,福建泉州362000

出　　处：《福建技术师范学院学报》2024年第5期46-53,共8页JOURNAL OF FUJIAN POLYTECHNIC NORMAL UNIVERSITY

基　　金：福建省职业技术教育中心福建省服装行业人才需求与职业院校专业设置研究项目(ZKY202102).

摘　　要：为提高服用织物折皱分类的精度,基于在线序列极限学习机OSELM原理,提出一种DE-SCA-OSELM的服用织物折皱分类方法.首先,以OSELM作为基础分类模型;然后采用差分进化算法DE中的侦查峰算子初始化正余弦优化算法SCA中的较差个体,以增加种群多样性,提升SCA算法全局寻优能力;最后采用DE-SCA算法对OSELM分类模型参数进行优化,以实现服用织物折皱精准分类.仿真结果表明,在参数最优情况下,织物折皱分类精度可达94.17%,高于其他传统折皱分类算法,说明DE-SCA-OSELM分类性能好,可为后续的折皱等级分类奠定基础.In order to improve the accuracy of garment fabric crease classification,a DE-SCA-OSELM garment fabric crease classification method is proposed based on the principle of online sequential limit learning machine.Firstly,OSELM was used as the basic classification model.Then,the detection peak operator in the differential evolution algorithm DE was employed to initialize the poor individuals in the sine and cosine optimization algorithm SCA,so as to increase the population diversity and improve the global optimization ability of the SCA algorithm.Finally,the DE-SCA algorithm was utilized to optimize the parameters of the OSELM classification model to achieve accurate crease classification of garment fabric.The simulation results show that under the condition of optimal parameters,the fabric crease classification accuracy can reach 94.17%,which is higher than other traditional crease classification algorithms,indicating that DE-SCA-OSELM enjoys good classification performance and can lay a foundation for subsequent crease classification.

关 键 词：服用织物 折皱分类 OSELM 智能分类 

分 类 号：TP392[自动化与计算机技术—计算机应用技术] TS941.1[自动化与计算机技术—计算机科学与技术]