Advanced Statistical Process Monitoring using Simulation-Based Algorithms

In recent years, significant advancements have been made in the field of statistical process monitoring. Traditional control charts typically rely on the normality assumption, which is often not satisfied in practical applications. As data complexity increases, many control charts introduced in contemporary literature can only be effectively analyzed through simulation methods due to their analytical or numerical intractability. When control charts are calibrated using simulation-based methods, they can be used in a wider variety of contexts compared to traditional methodologies. This transition towards simulation-based methodologies introduces several practical challenges that warrant careful analysis. The ongoing development of more efficient algorithms and solutions is essential from an applied perspective. This thesis contributes to the field of statistical process monitoring by presenting novel computational methodologies and software solutions. These are designed to enhance the applicability of control charts in scenarios where simulation techniques are preferred over conventional numerical methods. The thesis is organized as follows: the first chapter comprises a software solution that provides general tools for practical statistical process monitoring; the second chapter introduces a novel approach for monitoring mixed data, which is increasingly relevant in practice; third and fourth chapters present two algorithmic contributions focused on enhancing the efficiency of control limit calibration and hyperparameter tuning for control charts, respectively.