Statistical inferences under step stress partially accelerated life testing based on multiple censoring approaches...

Area of Science:

• Reliability Engineering
• Statistical Inference

Background:

• Evaluating lifespan distributions for highly reliable products is costly and time-consuming.
• Accelerated life testing (ALT) offers a more efficient alternative for gathering failure data.
• Parametric inference methods are crucial for analyzing ALT data.

Purpose of the Study:

• To develop methods for parametric inference in step-stress partially accelerated life testing (PALT) using censored data.
• To apply the Tampered Random Variable (TRV) model with the Nadarajah-Haghighi (NH) distribution for lifespan analysis.
• To estimate model parameters and acceleration factors accurately.

Main Methods:

• Utilizing step-stress partially life testing with multiple censored data.
• Applying the Tampered Random Variable (TRV) model.
• Assuming a Nadarajah-Haghighi (NH) distribution for lifespan under normal conditions.
• Employing Maximum Likelihood Estimation (MLE) for parameter estimation.
• Constructing asymptotic confidence intervals using the observed Fisher information matrix.

Main Results:

• Demonstrated the methodology with real-world data from Boeing 720 jet aircraft air conditioning systems.
• Simulation studies confirmed the performance of the estimation procedures under various censoring strategies.
• Mean Squared Errors (MSE) and average confidence interval lengths decrease with increasing sample size.
• Parameter estimates improve as the censoring level decreases.

Conclusions:

• The proposed parametric inference methods are effective for step-stress PALT.
• Increasing sample size and reducing censoring improve the accuracy of parameter and acceleration factor estimations.
• The study provides a robust framework for analyzing the reliability of highly dependable products.