Abstract

Control charts, a fundamental tool in Statistical Process Control (SPC), have emerged as a highly effective method for detecting anomalies and have seen widespread application in the industrial sector in recent years. Smart Manufacturing (SM) has become increasingly important in defining the ultimate goal of factory digitization, driven by advancements in technologies such as Artificial Intelligence (AI). Traditional SPC control charts increasingly struggle to cope with real-world responsibilities such as creating, detecting patterns, and deciphering processes. To counter these issues, Machine Learning (ML) algorithms have proved to be extremely useful and powerful analytical tools that can be combined with SPC control charts. This research will discuss the application of ML techniques for creating a multivariate AEWMA control chart. The proposed control chart utilizes the value of the smoothing constant, determined through a trained machine learning method, to effectively monitor changes in the variance-covariance matrix of a multivariate process. Monte Carlo Simulation is used to generate the run length profile, ARL, and SDRL. Finally, a case study of a control chart with machine learning capability is outlined.