نوع مقاله : مقاله پژوهشی
1 گروه آموزشی آمار، دانشکده علوم ریاضی و رایانه، دانشگاه علامه طباطبائی
2 استاد، دانشگاه علامه طباطبایی، گروه آمار، تهران، ایران
عنوان مقاله [English]
One of the most important tools for statistical process control is the control chart. The construction of control diagrams is achieved by determining the design parameters of the sample size, sampling distance and coefficient of control limits. Statistical-economic design is the best way to determine these parameters by considering statistical properties and cost. Statistical design - Economics Control charts require a distribution for the process failure mechanism. The most popular distributions in the analysis of the failure mechanism of the exponential distribution process are gamma and Weibull. Recently, a new distribution called generalized exponential distribution has been added to conventional distributions in data life analysis and process failure mechanisms. If the failure mechanism data follows this distribution, the use of other distributions will give users misleading and inaccurate results. On the other hand, a fundamental assumption in most control charts is the independence of process-related observations; But in practice we have situations with correlated data. As a result, it is of particular importance to identify control charts that can be used to control such data.In this paper, statistical-economic design of control charts with correlated data under generalized shockwave and Weibull models and by providing a numerical example , We show their application. We have applied a sensitivity analysis to investigate the effect of changing the controlled properties of the first type error rate, power and correlation coefficient. The results show that this type of control charts have undesirable statistical properties and the limitation on the first type of error has a significant effect on determining the average cost per unit time and design parameters. Also, the correlation coefficient is indirectly related to the sample size and sampling distance and is directly related to the average cost per unit time.