Prioritized EWMA control chart for time-sensitive process

Numerous challenges are faced by manufacturing industries in recent years, and process variation becomes the major source of poor quality in manufacturing control. A control chart in statistical process control (SPC) is a powerful tool to achieve stability and improvement in process capability by reducing the variability. A manufacturer normally has to deal with time-series observations for monitoring the processes. New studies recommend using machine learning techniques due to the ability of these methods to automatically detect data patterns and to exploit such data patterns for future prediction and process improvement that usually evaluated through control charts. This research outlines the exponentially weighted moving average(EWMA)control charts that are applied on time series data of a dairy distribution process. Different data are simulated from the AR, MA, and ARMA processes. MATLAB -based simulations of EWMA control charts for AR(1), AR(2), MA(1), MA(2), ARMA(1,1), and ARMA(2,2) are performed for each process at various sample size and replication, The average run length (ARL) is a significant measure to assess the performance of the control chart. In this work, an ARL-based EWMA chart is discussed for monitoring the process variance for AR, MA, and ARMA processes. The efficiency of these charts is compared in terms of ARLs. The EWMA for ARMA(2,2) chart is more efficient than other discussed charts in terms of ARLs. A real example is given illustrating the proposed chart in the industry. The work shows that the EWMA control chart highlights several data points exceeding the upper control limit for AR(1), MA(1), and ARMA(1,1) processes, which indicates that the process is out of control at these points, while shows that the process is in control for AR(2), MA(2), and ARMA(2,2). Such analysis ensures a stable quality and shows that each production process requires to be maintained within a predefined time limit. Moreover, certain industries need such a capable system to detect the quality at an early stage before it over shifted. The results of applying the AR, MA, and ARMAshow that the developed model can succeed to approximate time series data patterns, and as the order of these models has increased the ability to fit observations become more accurate for the cases studied in the control chart. The significant insight of the presented model in this work is to focus on the benefits of using EWMA on different types of time series data. This action will enhance the quality of the products, by offering an effective solution that will lower the time consumed during the management of the transportation time of the product’s processes.

Text IPM 2021 6 UPMIR.pdf Download (1MB)

Actions (login required)

View Item