基于等距特征映射的纺纱生产过程决策参数特征提取
Feature extraction of spinning-production process decision parameters based on Isomap
胡胜 1王品鉴 1赵小惠 1张刚1
作者信息
- 1. 西安工程大学机电工程学院,陕西西安 710048
- 折叠
摘要
纺纱生产中涉及的过程参数繁多,彼此相互关联且存在冗余,针对过程参数间相关性带来的质量指标难以精确预测与控制等问题,文中提出一种基于等距特征映射(Isometric Mapping,Isomap)的纺纱生产过程决策参数特征提取方法.首先,通过灰色关联分析计算过程参数与纱线质量指标间的灰色关联度,筛选出对产品质量指标影响较大的过程参数作为决策参数;然后,采用Isomap算法对决策参数进行特征提取降维,获得彼此独立的低维决策参数特征空间,将其输入到粒子群优化向量机模型(PSO-SVM)中验证所提出算法的特征提取效果;最后,通过算例进行验证,结果显示提出的方法可采用相较于原数据更少的特征空间维度获取更佳的预测效果.
Abstract
The spinning production involves many process parameters,which are interrelated and redundant.Since it is diffi-cult to accurately predict and control the quality indicators caused by the correlation between the process parameters,in this arti-cle a feature-extraction method is proposed based on Isometric Mapping(Isomap)for the spinning-production process decision pa-rameters.Firstly,the grey correlation analysis is used to rank the grey correlation between the process parameters and the yarn quality indicators;efforts are made to select the process parameters that have a significant impact on the produc t quality indicators as the decision parameters.Furthermore,these decision parameters are used for feature extraction and dimensionality reduction by means of Isomap;efforts are made to obtain the independent low-dimensional decision-parameter feature spaces and input them in-to the Particle Swarm Optimization Vector Machine Model(PSO-SVM),so as to verify this method's feature-extraction perform-ance.Finally,the numerical examples are used to verify the results;it is shown that this method has better prediction perform-ance by utilizing fewer feature space dimensions compared to the original data.
关键词
纺纱生产/决策参数/特征提取/等距特征映射/相关性分析Key words
spinning production/decision parameter/feature extraction/Isometric Mapping/correlation analysis引用本文复制引用
基金项目
国家自然科学基金资助项目(72001166)
陕西省自然科学基础研究计划项目(2022JQ-721)
出版年
2024
NETL
NSTL
万方数据