|
Original research
ADAPTIVE RELIABILITY ESTIMATION UNDER PROGRESSIVE CENSORING FOR INDUSTRIAL DECISION SUPPORT AND PREDICTIVE MAINTENANCE APPLICATIONSPages 397-406
Abstract:
A risk-improved data-adaptive shrinkage estimator is proposed for the reliability function of the exponentiated Weibull distribution under progressive Type-II censoring, with emphasis on both finite-sample risk reduction and decision support in industrial reliability studies. The exponentiated Weibull maximum likelihood plug-in estimator is combined with a structured Weibull-target reliability estimator through a time-dependent shrinkage weight. The weight is determined by an observed-information risk rule that balances the estimated variability of the full-model estimator against the squared discrepancy between the full-model and target reliability curves. A Monte Carlo study over representative sample sizes and censoring proportions shows that the proposed estimator improves the full exponentiated Weibull reliability estimator in integrated mean squared error, with the largest gains appearing under heavier censoring. The proposed reliability curve is then interpreted as a practical input for threshold-based maintenance timing, inspection scheduling, warranty assessment, and risk-based operational planning. In this way, the method contributes not only a statistically coherent adaptive shrinkage mechanism, but also a more stable reliability tool for censored industrial life-testing data.
Keywords:
Adaptive shrinkage,
Exponentiated Weibull distribution,
Reliability function,
Progressive Type-II censoring,
Predictive maintenance,
Reliability-based decision support,
Weibull target,
Integrated mean squared error.
|
Omar Jabar Alial Al-Qaragholi, 
PDF 