Studies on the Monitoring of Linear Profiles

• Main Authors: Huei Pin Lin, 林惠萍
• Other Authors: P. C. Wang
• Format: Others
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/71792988513294009226

碩士 === 長庚大學 === 工商管理學系 === 99 === Control chart is the most common statistical tool of statistical process control (SPC). Previous studies are based on the product with a single characteristic or multiple characteristics. The objective of this research is to study control chart on the product with a linear function, called the linear profile. It could be used to check the quality of product, and to monitor the process used for producing a product. When abnormal products occur, they can be detected immediately and engineers can respond accordingly. To monitor a process, we must choose an appropriate control chart. There are three such charts available：Kim et al. (2003) proposed to control intercept, slope and variance of profile separately, the combined control chart referred to as KMW. Zou et al. (2007) explored the performance of combined statistics and used the multivariate exponentially weighted moving average scheme referred to as MEWMA. Zhang et al. (2009) used the likelihood ratio test to control the profile referred to as ELR. Usually, the average run length (ARL) to detect changes is used for comparison of these three methods. We add the standard deviation of run length (RL) for comparison. At last, this research is based on the simulation to analyze the average and standard deviation of run length in variance situations, the results are reported.