DEVELOPMENTS OF DESIGNING METHOED OF CONTROL CHART TO ENHANCE DETECTION POWER IN HIGH-YIELD PROCESSES

在本研究项目中，Ohta教授和Kusukawa副研究员在2002年至2004年期间主要接触了两个研究对象，并取得了研究成果。设计控制图，灵敏地检测小或中等向任何方向的变化，这些变化表明过程恶化和过程改进，在小尺寸子组的过程均值中，在高成品率过程中，过程不合格率或过程不良率，并实现上述质量特征的I类误差的几乎规定值。我们提出的控制图将非常有用，作为Shewhart类型图的替代控制图，如X^^-R图，p图和c图（参见参考文献1,2,3,5,6,7,9,11,12和13）。2. 上述课题1所开发的控制图的经济性设计。在各种工艺情况下，如工艺故障率随时间变化的动态工艺，使系统单位时间工艺周期的利润最大化或系统单位时间工艺周期的成本最小化。当需要使用上述研究课题1中我们提出的控制图经济地运行高收率过程时，我们提出的经济设计方法将非常有用。（见参考文献4、8和10）。作为未来的研究工作，可以在参考文献1和6的基础上，将参考文献13中的EWMA_<CCC-r>图与CCC-r图相结合，设计合成EWMA图。期望由时间序列数据构建的合成EWMA图比合成CS图对工艺不良率的监测更敏感。作为未来进一步的研究工作，在文献4、8和10的基础上，可以开发动态过程下综合EWMA图的经济设计。

In this research project, two study subjects have been mainly approached and the research findings have been obtained between 2002 and 2004 by Professor Ohta and Research Associate Kusukawa. 1. Design of control chart to detect sensitively small or moderate shifts toward any directions, which indicate the process deterioration and the process improvement, in the process mean obtained from small size of subgroups, the process fraction defectives or the process defect rate in high-yield processes and actualize almost a specified value of type I error of quality characteristics mentioned above. Our proposed control charts would be useful highly as alternative control charts of the Shewhart types charts such as X^^-R chart, p chart and c chart (See REFERENCES 1, 2, 3, 5, 6, 7, 9, 11, 12 and 13.). 2. Economic design of control chart developed in above study subject 1. to maximize the system profit per unit time of process cycle or to minimize system cost per unit time of process cycle under various process situations such as dynamic process where the process failure rate is varying over time. Our proposed-economic designing methods would be useful highly when it is required to run a high-yield process economically by using our proposed control charts in above study subject 1.. (See REFERENCES 4, 8 and 10.). As the future Research work, based on REFERENCES 1 and 6, the Synthetic EWMA chart that is integration of the CCC-r chart and the EWMA_<CCC-r> chart in REFERENCE 13 could be designed. It is expected that the Synthetic EWMA chart constructed from time-series data would be more sensitive to monitor the process defective rate than the Synthetic CS chart. As the further future Research work, based on REFERENCES 4, 8 and 10, an economic design of the Synthetic EWMA chart under dynamic process could be developed.