基于支持向量机的制造过程二阶多元质量诊断技术研究

Two-stage Multivariate Quality Diagnosis Model for Manufacturing Process Based on the SVM

针对目前制造过程中出现的多元过程的质量异常现象,文中整合了多元移动加权平均(MEWMA)控制图和支持向量机(SVM),以原始数据与其特征值分别作为分类器的输入向量,建立一种针对多元过程的质量诊断模型。该模型主要有两个阶段,第一阶段利用MEWMA控制图对输入样本进行监测,当控制图报警时进入第二阶段,对造成报警的样本及其特征值利用SVM进行分类诊断。最后以发动机缸体曲轴孔的加工为例验证该模型的有效性。

Aiming at abnormal problems of qual ity in multivar iate process of manufacturing, the mul t iple weighted moving average （MEWMA） control chart and the support vector machine （SVM） are integrated. The original data and statistical characteristic values are taken as the input vectors of the classifier, the two-stage multivariate quality diagnosis model for manufacturing process is set up. There are two phases. The first is to monitor samples by using MEWMA. Once alarming, the second is to diagnose samples and their characteristic values by using SVM. Finally the effectiveness of the model is verified using a practical case.