Control diagrams are graphical communication media that can be used as tools in the process of controlling statistics statistically to assist in monitoring and improving a quality process by helping to separate specific cause and general cause variables. Basically the control diagram plots the area of acceptance of a hypothesis test. This study aims to detect disruptive causes in the control diagram in time series cases and to justify the statistical model in accordance with the procedure. A process is said to be controlled if the observational data from that process behaves like random iid variables (independent identically distributed), and if it is not controlled, it is out of control. The study uses 25 days’ time series observational data in PT. Guccitex Cimahi. In this time frame analysis, research that is influenced by specific causes can be treated as outliers of the time series model and can be detected by outlier detection methods. The results in this study explain that a traditional control diagram is very dependent on the assumption of iid (independent identically distributed). Application in real life faces observations that are not iid (independent identically distributed), which show the effects of trends, seasonality, and other influences. The control diagram is in principle the same, using time series analysis to overcome these effects by plotting the area of acceptance of a hypothesis test, but if the process is not controlled, the control diagram cannot provide any guidance, so it is necessary to do a data analysis using analysis time series.

Quality control must be carried out very tightly for the production of single wing multivariate dispersion aircraft manufacturing. This research was conducted at PT. Dirgantara Indonesia in producing components of Airbus aircraft wings centered in Germany and France. The method used includes three stages, namely determining the single observation MEWMS diagram, determining the algorithm for determining L, and implementing MEWMS. The data used are L simulation data for p = 2 and p = 3, ARL = 370, and various values, key characteristic data from 31 wing component components of the aircraft fly with. The results showed that the more variables, the L value as the determinant of the width of the MEWMS control limit was smaller, and the greater the value of ω, the value of L as the width of the MEWMS control border width is wider.