Scanning Cluster of Manufacturing Shift with Spatial Structure

• Main Authors: Chen-Yu Lin, 林陳裕
• Other Authors: Chen-ju Lin
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/28975931886547409929

碩士 === 元智大學 === 工業工程與管理學系 === 98 === Statistical process control (SPC) is widely applied to monitor manufacturing processes. With the progress of production yield, Shewhart type control charts fail immediately detect small shifts, Cumulative Sum (CUSUM) and Exponentially Weighted Moving Average (EWMA) control charts are developed for higher detection powers. The development of SPC, however, mostly focuses on temporal data. Those techniques are not appropriate for analyzing the manufacturing data embedded with spatial structure. If the monitoring process treats manufacturing records merely as multivariate data, the process control procedure misses important information of spatial relationship and patterns among the data. Thus, this paper proposes a spatial EWMA (SEWMA) test procedure and Standardized SEWMA test procedure to identify process shifts. The goal is to analyze whether there is an outbreak cluster in the investigated area. The SEWMA procedure weights likelihood ratios considering their relative distance on the plane, and the Standardized SEWMA procedure Standardized the weight of SEWMA statistics. The simulation results show that the SEWMA test procedure and Standardized SEWMA test procedure have higher detection power than the classical multivariate control chart of Hotelling’s T2 chart.