Competing Risks in Accelerated Life Testing: A Study on Step-Stress Models with Tampered Random Variables

Abstract

This study introduces a novel approach to accelerated life test experiments by examining competing risk factors using the Tampered Random Variable (TRV) model. This approach remains extensively unexplored in current research. The methodology is implemented for a simple step-stress life test (SSLT) model and accounts for various causes of failure. The Power Chris–Jerry (PCJ) distribution is utilized to model the lifetimes of units under different stress levels, incorporating unique shape parameters while maintaining a fixed-scale parameter. This study employs the TRV model to integrate constant tampering coefficients for each failure cause within step-stress data analysis. Maximum-likelihood estimates for model parameters and tampering coefficients are derived from SSLT data, and some confidence intervals are presented based on the Type-II censoring scheme. Furthermore, Bayesian estimation is applied to the parameters, supported by appropriate prior distributions. The robustness of the proposed method is validated through comprehensive simulations and real-world applications in different scientific domains.