Integrating ICA and SVM for identifying mixture control chart patterns in a multivariate process

• Main Authors: Chao-Liang Chang, 張兆良
• Other Authors: Yuehjen E. Shao
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/26147923777968786684

碩士 === 輔仁大學 === 應用統計學研究所 === 99 === Mixture control chart patterns (CCPs) mixed by more types of basic CCPs together usually exist in the real manufacture process. However, most existing studies are considered to recognize the single abnormal CCPs. This study utilizes independent component analysis (ICA) and support vector machine (SVM) for recognizing mixture CCPs recognition in multivariate process. The proposed scheme, firstly, uses ICA to the monitoring process data containing mixture patterns for generating independent components (ICs). The undetectable basic patterns of the mixture patterns can be revealed in the estimated ICs. The ICs are then used as the input variables of the SVM for building CCP recognition model. Experimental results reveal that the proposed scheme is promising for recognizing mixture control chart patterns in a multivariate process.