摘要
质量控制图的模式识别是智能工序质量诊断分析系统的基础,产品的大规模定制趋势使得统计控制的样本量减少。在探讨以往的识别方法的基础上,研究了基于支持向量机(support vector machine,SVM)的质量控制图模式识别方法,该方法以控制图的12个时域特征作为分类的统计量,利用支持向量机作为分类器,对控制图的正常模式和各种失效模式进行识别。仿真实验表明,该方法在小样本条件下具有识别率稳健、识别速度快等优点,为实现大规模定制模式下工序质量在线诊断和事前控制提供了一种可行的途径。
Control chart patterns (CCPs) are widely used to identify the potential process problems in modem manufacturing industries. Recently, mass customization relate to the small sample of statistical process control. This paper discussed former recognition methods and presented an approach of small sample CCPs intelligent recognition based on support vector machine(SVM).The 12 time domain features were Statistics and SVM was used as classifier to recognize six types of CCPs.The proposed method has the advantages at robust recognition rate and high recognition speed under small sample. It may be applied in on-line process quality diagnosis and pre-control in customized manufacturing.
出处
《组合机床与自动化加工技术》
北大核心
2009年第8期1-4,9,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
国家自然科学基金(70871087)