Reliability Analysis of Limited Failure One‐Shot Devices With Covariate‐Dependent Failure Times

ABSTRACT

In this work, we develop a covariate‐dependent limited failure population (LFP) model for one‐shot device testing, allowing explanatory variables to influence the defect probability as well as the lifetime distribution of susceptible units. Unlike existing approaches that incorporate covariates only in the susceptibility mechanism, the proposed framework combines a logistic regression model for the defect probability with a general parametric specification for the failure‐time distribution, enabling separate inference on factors affecting defect occurrence and failure progression. Parameter estimation is carried out by means of an expectation–maximization (EM) algorithm, and covariate effects are assessed using Wald tests. Structural identifiability of the proposed model is also established. Simulation studies and a real‐data application demonstrate the accuracy of the estimators and the practical usefulness of the proposed methodology in reliability analysis.