Statistical Analysis of Progressive Interval Type-I Partially Accelerated Test and Its Inspection of Optimum Periods

Abstract

This study explores a general framework for reliability analysis named progressive interval Type-I censoring within a multi-stage step-stress partially accelerated life testing setting. A more flexible alternative to the traditional exponential model, known as the length-biased exponential (LBE) distribution, is employed to model failure times. Due to its symmetrical feature, it has extensive applications in real-world domains such as survival analysis, actuarial science, reliability, and mathematical finance. The maximum likelihood approach of estimation is utilized to estimate the model parameters, along with bootstrapping techniques to assess estimation efficiency. Confidence intervals for the LBE parameters are also derived based on asymptotic variances. To optimize the inspection period, two competing optimality criteria—variance minimization (Varoptimality) and determinant maximization (D-optimality)—are investigated. A comprehensive Monte Carlo simulation study is conducted to evaluate the performance of different estimation strategies, demonstrating the superiority of the proposed methodology. Steel is particularly valued for its toughness, wear resistance, and hardness, all of which can be significantly modified through heat treatment and annealing processes. So, a real-world application using hardened steel failure data validates the practical relevance of the developed inferential framework. The findings offer valuable insights for statisticians and reliability engineers in designing efficient life-testing experiments under constrained resources.OPEN ACCESS Received: 21/03/2025 Accepted: 07/05/2025 Published: 22/09/2025

Document