accelerated life testing

To address the scarcity of failure data and susceptibility to outliers in step-stress accelerated life testing (SSALT) for highly reliable products, this paper proposes a robust parameter estimation method based on the minimum density power divergence estimator (MDPDE), the first application of MDPDE to exponential mixture distribution modeling under step-stress conditions. The method preserves asymptotic efficiency while significantly enhancing robustness against outlying failure observations; its asymptotic distribution is rigorously derived. Monte Carlo simulations and empirical analysis demonstrate that, compared with conventional maximum likelihood estimation (MLE), the proposed estimator reduces bias by over 30% in the presence of outliers, with an efficiency loss of less than 5%. Thus, it achieves an optimal trade-off between high robustness and high statistical efficiency.

Quality engineers lack systematic degradation modeling methodologies, hindering the accuracy and practical implementation of reliability assessment. Method: This study establishes an industrially oriented degradation analysis framework that unifies diverse degradation data sources—including repeated measurements and accelerated destructive testing—and integrates path models (e.g., general path models) with stochastic process models (e.g., Wiener processes), augmented by Bayesian and likelihood-based statistical inference techniques. A standardized modeling workflow and lifetime prediction toolkit are implemented in R/Python. Contribution/Results: The framework bridges the gap between theoretical degradation modeling and engineering practice, significantly improving the accuracy and reproducibility of reliability predictions for complex systems. It delivers an actionable guideline and open-source software support for industry-standardized deployment, enabling robust, traceable, and scalable reliability engineering.

This study addresses the challenge of scarce failure data for highly reliable products under normal operating conditions, where step-stress accelerated life testing is commonly employed. Recognizing that conventional maximum likelihood estimation (MLE) is sensitive to outliers in such tests—leading to unreliable inference—this work extends minimum density power divergence estimation (MDPDE) to step-stress models with mixture distributions. Under the assumption of Weibull-distributed lifetimes, the paper establishes the asymptotic theory for parameter estimation. The proposed method maintains high estimation efficiency while substantially enhancing robustness against contaminated data. Comprehensive simulations and a real-data analysis demonstrate that the MDPDE-based approach significantly outperforms MLE in the presence of outliers, achieving both strong robustness and high statistical efficiency.

This paper addresses the challenge of optimal design for multi-stress-level constant-stress accelerated life testing (ALT). Methodologically, it proposes a simulation-based global optimization framework that integrates differential evolution (DE) with Monte Carlo simulation, optimizing both stress-level selection and test unit allocation jointly to minimize the root-mean-square error (RMSE) of model extrapolation. The study reveals two key insights: (i) an intrinsic matching relationship between the optimal number of stress levels and model parameters, and (ii) an inverse-proportional relationship between test unit allocation ratios and corresponding stress levels. These findings provide a scalable, surrogate-based optimization pathway for high-dimensional, complex ALT designs. Experimental results demonstrate substantial improvements in lifetime prediction accuracy and testing efficiency—particularly for large-scale, multi-point ALT scenarios where analytical solutions are intractable.

The PHM (Prognostics and Health Management) community has long suffered from a lack of systematically evaluated, freely accessible degradation datasets. Method: This work establishes the first multi-dimensional unified evaluation framework for PHM datasets, incorporating critical dimensions—data provenance, equipment types, sensor configurations, failure modes, and annotation completeness—integrated with structured metadata analysis, cross-dataset comparative assessment, task-specific PHM mapping, and physics-of-failure-informed semantic annotation. Contribution/Results: We systematically curate and analyze 32 high-quality public datasets, identifying 11 recurrent deficiencies. Based on this analysis, we provide task-oriented data selection guidelines and benchmarking recommendations. This study fills a critical gap in systematic surveys of PHM public data resources, explicitly delineates the applicability boundaries and modeling limitations of existing datasets, and has been widely cited and adopted within the PHM research community.

This study addresses optimal design for simple step-stress accelerated life tests involving two independent competing failure modes. Within a Bayesian framework, it integrates the cumulative exposure model with a log-linear stress–life relationship and employs a pre-posterior variance minimization criterion to achieve exact small-sample optimization without relying on large-sample approximations. The work innovatively extends the quantile reparameterization approach—previously limited to single-failure-mode settings—to the competing risks context, enabling priors to be directly elicited from engineering knowledge. Posterior inference is conducted via Stan’s No-U-Turn Sampler, and Monte Carlo search over a candidate design grid identifies the optimal test plan. Validation using real data from solar lighting devices demonstrates that the optimal low-stress level consistently aligns closely with normal use conditions, yielding robust results.

This study addresses the challenges posed by interval-censored data and the sensitivity of conventional maximum likelihood estimation to outliers in cyclic accelerated life testing (CyALT). Under the assumption that product lifetimes follow a log-normal distribution, the authors propose a robust inference method based on the weighted minimum density power divergence estimator (WMDPDE). The proposed approach achieves both high statistical efficiency and strong resistance to contamination. Notably, this work establishes the asymptotic theory and influence function of WMDPDE within the CyALT framework for the first time. Simulation studies demonstrate that WMDPDE substantially outperforms traditional methods in the presence of outliers while retaining high efficiency under clean data conditions. The practical applicability and robustness of the method are further confirmed through analysis of real-world air conditioner reliability data.

This study addresses a key limitation in traditional step-stress accelerated life testing (ALT) models, which assume product homogeneity and thus fail to capture heterogeneous aging behaviors that may emerge under high stress. The authors propose a hazard-rate-based heterogeneous step-stress ALT model that retains the homogeneity assumption at the initial stress level but introduces a finite mixture model at the second stress level to represent m latent subpopulations with distinct failure mechanisms. Parameter estimation under Type-II censoring is handled via the EM algorithm. Notably, this approach is the first to embed a finite mixture model within a hazard-rate framework, establishing an interpretable paradigm for heterogeneity modeling. The model rigorously reduces to the existing exponential heterogeneous model when the Weibull shape parameter equals one, confirming its theoretical generalizability. Simulations demonstrate that neglecting heterogeneity induces substantial bias in lifetime predictions across all quantiles, particularly at early failure quantiles.

This study addresses the challenge that conventional static testing fails to account for device-to-device variability and the dynamic interplay among multiple degradation mechanisms—such as bias temperature instability (BTI), electromigration (EM), and time-dependent dielectric breakdown (TDDB)—in semiconductor reliability assessment. To overcome this limitation, the work formulates reliability qualification as a partially observable sequential decision-making problem and introduces an adaptive testing framework that integrates Monte Carlo tree search with simulated annealing (MCTS-SA) and an extended Kalman filter (EKF). This approach enables closed-loop optimization of stress conditions through real-time belief-state estimation, dynamically balancing accurate degradation characterization against the risk of catastrophic failure while maintaining required safety margins. Experimental results demonstrate that after 5,000 iterations, the characterization success rate improves from 20% to 54%, yielding a cumulative gain of 39%; at termination of the optimal test sequence, the EM and TDDB damage fractions reach 0.564 and 0.537, respectively.