摘要
分析了现有控制图识别器在实际应用中存在的缺陷,并提出了一种基于支持向量机(SVM)的新方法.为了克服HAH多分类SVM(HAH-SVM)的缺陷,提高识别速度和准确率,设计了一种有针对性的SVM多分类器进行模式识别.仿真实验结果表明,该方法相对现有的BP和HAH-SVM方法能得到更高的识别率和识别速度,适合于工序的实时在线控制.
This paper analyzes the limitations of current control chart recognizers in practical applications, and presents a new method based on support vector machine (SVM). In order to overcome the shortcomings of Half- Against-Half SVM (HAH-SVM) and improve the recognition speed and accuracy, a special multi-class SVM- recognizer is designed for pattern identification. Simulation and experimental results show that, compared with BP (backpropagation) and HAH-SVM methods, the presented method can obtain a faster recognition speed and a higher recognition accuracy, and can be applied to an online real time control process.
出处
《信息与控制》
CSCD
北大核心
2007年第2期187-191,198,共6页
Information and Control
基金
国家自然科学基金资助项目(70272032)
关键词
多分类支持向量机
统计质量控制
控制图
质量诊断
multi-class support vector machine
statistical quality control
control chart
quality diagnosis