摘要
针对间歇过程采集的数据具有三维矩阵数据形式采样时间不等长,不能直接适用于传统故障检测方法的问题,提出一种基于统计模量分析的支持向量数据描述(SPA-SVDD)方法.利用统计模量分析(SPA)将三维矩阵形式转换为二维矩阵,剔除各批次采样时间不等长对检测的影响.统计模量分析可以提取过程数据的非线性、动态性、多模态等特性.用支持向量数据描述(SVDD)方法在由统计模量组成的训练集数据上进行建模,得到支持向量到球心的距离.对新的样本进行检测,对半导体工艺过程进行仿真实验,并对比M-SVDD、SPA-K NN方法验证方法的有效性和优越性.
In order to solve the problem that the data collected in batch process has different sampling time in the form of three-dimensional matrix data and can not be directly applied to traditional fault detection methods,a support vector data description(SPA-SVDD)method based on statistical modulus analysis is proposed.Firstly,the statistical modulus analysis(SPA)is used to transform the three-dimensional matrix form into two-dimensional matrix,and the influence of the unequal sampling time of each batch on the detection is eliminated.At the same time,statistical modulus analysis can extract the nonlinear,dynamic and multimodal characteristics of process data.Then support vector data description(SVDD)method is used to model the training set data composed of statistical modulus,and the distance from the support vector to the center of the ball is obtained.Finally,the new samples are tested.The simulation experiments of semiconductor process are carried out,and the effectiveness and superiority of M-SVDD and SPA-K NN methods are compared.
作者
谢彦红
薛志强
李元
XIE Yan-hong;XUE Zhi-qiang;LI Yuan(Shenyang University of Chemical Technology,Shenyang 110142,China)
出处
《沈阳化工大学学报》
CAS
2020年第2期158-164,共7页
Journal of Shenyang University of Chemical Technology
基金
国家自然科学基金面上资助项目(61673279,61490701)
辽宁省教育厅重点实验室基础研究项目(LZ2015059)
辽宁省自然科学基金项目(2015020164)
辽宁省教育厅一般项目(L2015432)。
关键词
间歇过程
统计模量分析
支持向量数据描述
半导体过程
故障检测
intermittent process
statistical patterns analysis
support vector data description
semiconductor process
fault detection
