Machine learning-assisted neutrosophic MDS t-control chart for monitoring exponential processes with application

Statistical Process Control (SPC) techniques assume that process observations are precise and determinate. However, industrial and reliability application observations are affected by measurement errors, incomplete information, or subjective assessments, resulting in uncertain and interval-valued data. This study proposes a machine learning-assisted neutrosophic Multiple Dependent State (MDS) t-control chart for monitoring exponentially distributed quality characteristics under uncertainty. Three indeterminacy frameworks are considered: known indeterminacy, unknown random indeterminacy modeled by a Beta distribution estimated from interval observations, and correlated indeterminacy between the process observations and the indeterminacy levels. Double control limits are developed using two control coefficients, and the neutrosophic expected value, variance, and Average Run Length (ARL) are derived under each framework. A Random Forest classifier is integrated as a secondary decision-making mechanism. Monte Carlo simulation demonstrates that the proposed hybrid chart detects small and moderate process shifts faster than classical and existing neutrosophic MDS charts. Performance is evaluated using ARL, Extra Quadratic Loss (EQL), Relative Average Run Length (RARL), Performance Comparison Index (PCI), and Relative Mean Index (RMI). A temperature monitoring dataset with interval observations illustrates the practical applicability of the model. The finding shows that an adaptive, flexible, and robust SPC framework for monitoring uncertainty processes.