基于多元线性回归与BP神经网络的缝口强力预测  

Prediction of seam strength based on multiple linear regression and BP neural network

作　　者：陈佳珍 丁笑君[1,2,3] 邹奉元[1,2,3] 杜磊[1,2,3] CHEN Jiazhen;DING Xiaojun;ZOU Fengyuan;DU Lei(School of Fashion Design&Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China;National Virtual Simulation Experimental Teaching Center of Clothing Design,Zhejiang Sci-Tech University,Hangzhou 310018,China;National Experimental Teaching Center of Clothing,Zhejiang Sci-Tech University,Hangzhou 310018,China)

机构地区：[1]浙江理工大学服装学院,杭州310018 [2]浙江理工大学服装设计国家级虚拟仿真实验教学中心,杭州310018 [3]浙江理工大学服装国家级实验教学示范中心,杭州310018

出　　处：《浙江理工大学学报（自然科学版）》2020年第6期749-756,共8页Journal of Zhejiang Sci-Tech University(Natural Sciences)

基　　金：服装国家级实验教学示范中心及服装设计国家级虚拟仿真实验教学中心实验教学项目(ZX 2019006)。

摘　　要：缝口强力的影响因素众多且相互间的关系复杂,有效地预测缝口强力有利于服装的品控。采用经选定的涤棉混纺面料为实验对象,通过控制变量法进行缝口强力5因素3水平的全面实验。选定缝纫因素,包括缝型、线迹类型、缝边宽度、机针号数和线迹密度。用SPSS分析单因素对缝口强力的影响作为预测依据;基于多元线性回归和BP人工神经网络,用Matlab编程建立缝口强力预测模型,并比较两种预测方法的准确性;最后用较优的预测模型搭建针对服装企业的缝纫工艺参数推荐框架。结果表明:多元线性回归与BP神经网络模型预测误差均值分别为8.579%和2.642%,说明BP神经网络的整体预测精度更高,建议采用BP神经网络预测模型来进行缝纫工艺参数推荐。There are many factors affecting seam strength, and the relationship among the factors is complex. Effective prediction of seam strength is beneficial to clothing quality control. The polyester-cotton blended fabric was used as the experimental object. The comprehensive experiment of five factors and three levels of seam strength was carried out by the method of controlling variables. The sewing factors included seam type, stitch type, seam width, number of stitch needle and stitch density. The influence of single factor on seam strength was analyzed by SPSS as the basis for prediction. Based on multiple linear regression and BP artificial neural network, the prediction model of seam strength was established by Matlab programming, and the prediction accuracy of the two methods was compared. Finally, a better prediction model was used to build the recommendation framework of sewing process parameters for garment enterprises. The results showed that the average prediction errors of multiple linear regression and BP neural network model were 8.579% and 2.642% respectively, indicating that the overall prediction accuracy of BP neural network is higher. Hence, the BP neural network prediction model is suggested to recommend sewing process parameters.

关 键 词：缝纫参数 缝口强力 多元线性回归 BP人工神经网络 预测模型 缝纫参数推荐 

分 类 号：TS941.63[轻工技术与工程—服装设计与工程]