The Economic Design of Zero-inflated Poisson Distributed Attributes Control Charts

• Main Authors: Jen-Hsiung Yu, 游仁雄
• Other Authors: Tong-Yuan Koo
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/34043868087528676737

碩士 === 雲林科技大學 === 工業工程與管理研究所碩士班 === 97 === The process capability increase rapidly in recent year, and it is rare to find a defect in samples. When a large number of non-defect samples appear, Poisson distribution will underestimate the dispersion of defects and this will cause the control limit to be narrower. Therefore, some alternatives have been developed to alleviate this problem, one of which is the control chart based on the zero-inflated Poisson distribution. This study develop an economic design of zero-inflated Poisson distributed attributes control chart and using Genetic Algorithm in searching the oprs (sampling interval and upper control limit) to make the expected cost per time unit be the lowest. There are some conclusions in this research： 1. The rate of influence (ROI) is an index of influence in expected total cost per unit time by those parameters and the results show the order of influence at the case if the process is continue during the search for assignable cause as follows: the mean time that the process is in control ( ) > expected fraction defective when the process is in control ( ) > expected fraction defective when the process is out of control ( ) > expected inspection and charting time per unit sampled ( ) > expected search time associated with a true alarm ( ) > fixed cost of sampling ( ) > expected search time associated with a false alarm ( )> penalty incurred per defective item ( ) > fixed cost of maintaining process control ( ) > expected repair time ( ) > the average of defect when the process is in control ( ) > the average of defect when the process is out of control ( ) > expected cost of downtime per unit time ( ), expected cost of repair per unit time ( ) > expected search cost associated with a false alarm ( ) > production rate per unit time ( ) > expected search cost associated with a true alarm ( ). 2. The order of influence at the case if the process is stop during the search for assignable cause as follows: the mean time that the process is in control ( ) > expected fraction defective when the process is in control ( ) > expected inspection and charting time per unit sampled ( ) > expected fraction defective when the process is out of control ( ) > expected search time associated with a false alarm ( ) > fixed cost of sampling ( ) > penalty incurred per defective item ( ) > fixed cost of maintaining process control ( ) > expected repair time ( ) > the average of defect when the process is in control ( ) > expected search time associated with a true alarm ( ) > the average of defect when the process is out of control ( ) > expected cost of downtime per unit time ( ) > expected cost of repair per unit time ( ) > expected search cost associated with a false alarm ( ) > production rate per unit time ( ) > expected search cost associated with a true alarm ( ).