|
Abstract
|
Accelerated life tests (ALTs) are key in reliability analyses, providing lifetime estimates of highly reliable products. The step-stress design increases the stress at predefined times, while maintaining constant stress between successive changes. This approach accelerates the occurrence of failures, effectively reducing experimental duration and cost. While many studies assume a specific form for the lifetime distribution, in certain applications a general form satisfying certain properties, such as the proportional hazards (PH) requirement, should be preferred. This work examines particular forms of baseline hazards, namely, linear and quadratic forms. Moreover, certain experiments may face practical constraints making continuous monitoring infeasible. Instead, devices under test are inspected at predetermined intervals, and resulting data are then grouped as counts of failures. Recent works have shown an appealing trade-off between the efficiency and robustness of divergence-based estimators under intervalcensored data. This paper introduces the step-stress ALT model under PHs and presents a robust family of minimum density power divergence estimators (MDPDEs) for estimating device reliability. The MDPDEs of related lifetime characteristics and their asymptotic distributions are derived, providing approximate confidence intervals Empirical evaluations via Monte Carlo simulations demonstrate the estimators robustness and efficiency, supplemented by an illustrative example demonstrating the model and method's utility.
|