| Summary: | 碩士 === 輔仁大學 === 應用統計學研究所 === 86 === Due to the development of technology and the riches of material civilization, the needs of consumers are tremendously increasing. To satisfy these needs, industries have adapted to multiple stream processes to increase the production of processes. The production of multiple stream processes is massive, and consequently, the monitoring or detecting the assignable causes of multiple stream processes becomes a very important issue for industry processes. To address this issue, this thesis discusses three means of sampling schemes and four statistical control charts for monitoring multiple stream processes. These sampling schemes are stratification, mixing, and simple sampling; the four statistical control charts include Shewhart, cumulative sum (Cusum), exponentially weighted moving average (EWMA), and cumulative score (Cuscore) control charts. This thesis examines the pros and cons of different combinations of three sampling schemes and four control charts for monitoring multiple stream processes. The two indices, average performance measures (APM) and average time to signal (ATS), are employed to evaluate the gain of these combinations. The research findings show that the simple sampling scheme is, in general, the best sampling scheme among the three sampling schemes. However, the disadvantages of using a simple sampling scheme are the needs of maintaining more statistical control charts and more samples. The research findings also indicate that Cusum control charts have the greatest capability of detecting assignable causes of multiple stream processes.
|
|---|
