EWMA图表
控制图
马尔可夫链
统计的
统计
惯性
马尔可夫过程
过程(计算)
计算机科学
瞬态(计算机编程)
统计过程控制
控制理论(社会学)
数学
控制(管理)
人工智能
物理
经典力学
操作系统
作者
Poune Ghasemian,Rassoul Noorossana
标识
DOI:10.1080/03610926.2023.2184190
摘要
Since exponentially weighted moving average (EWMA) control charts combine information from the current and past samples, they are relatively more effective in detecting small process shifts than Shewhart control charts. A prevailing problem in control charts is the property of not responding quickly to process shifts, that is, showing statistical control of the process while a shift has occurred in the process parameter(s). This property of control charts, which is referred to as “inertia”, is an important issue in control charting and has been a major concern to engineers and statisticians. This study surveys the inertial properties of EWMA control charts in detail. First, the distribution of EWMA chart statistic and signal resistance probability (SRP) are studied for many different cases when there are undetected transient shifts. All measures are calculated and evaluated via a Markov chain approach. Then, an optimal scheme is proposed to minimize the out-of-control average run length (ARL) for a specified shift, subject to SRP constraints. The results reveal that the inertia can be sometimes a serious problem in EWMA control schemes that should not be overlooked. It is shown that an appropriate scheme based on both ARL and SRP measures could be useful for processes that may have sustained and/or transient out-of-control states.
科研通智能强力驱动
Strongly Powered by AbleSci AI