Ayse Bugatekin
Conventional statistical process monitoring approaches primarily focus on changes in location, dispersion, or distributional characteristics. However, process deterioration may also emerge through variations in uncertainty and information structure. Motivated by this limitation, this study proposes an Entropy-EWMA monitoring framework for progressively Type-II censored lifetime data. Developed under the Exponentiated Weibull distribution, the framework incorporates a censoring-adjusted entropy estimator into an adaptive EWMA structure to account for information loss and monitor changes in process uncertainty. The statistical performance of the proposed approach is evaluated through Monte Carlo simulations under different entropy degradation levels, censoring structures, observed sample sizes, and auxiliary shift-estimation smoothing parameter values. Control limits are calibrated to achieve an in-control Average Run Length close to the nominal target of ARL0=370. The results show satisfactory in-control performance and substantially decreasing out-of-control Average Run Length as entropy degradation increases. Comparative simulations further demonstrate shorter run lengths for the Entropy-EWMA scheme under moderate and severe entropy degradation relative to the Classical EWMA. The practical applicability of the method is illustrated using Wind Turbine SCADA data under Uniform and Late Progressive Type-II censoring schemes. Neither method produces false alarms during the in-control phase, while both detect changes during the out-of-control phase. The Classical EWMA provides earlier and more persistent signals associated with location changes, whereas the Entropy-EWMA responds to changes in process uncertainty. These findings demonstrate that the two approaches capture different aspects of process behavior and can be interpreted as complementary monitoring tools.