Model Selection Methods for Reliability Assessment Based on Interval-Censored Field Failure Samples  
Author Tzong-Ru Tsai


Co-Author(s) Sih-Hua Wu; Yan Shen


Abstract Incomplete field failure data from automated production are often applied for evaluating the system reliability. But the evaluation could be impacted by the uncertainty of the product’s lifetime distribution, which is usually predetermined but may be misspecified. In this paper, we assume that the system lifetime distribution follows a location-scale family instead of a certain distribution. Then a model selection mechanism is proposed to assign the most likely candidate distribution from a pool of the location-scale distributions. Based on interval-censored field failure data sets, we study how to select the most suitable candidate model as the best system lifetime distribution. Also, the maximum likelihood estimates (MLE) of parameters of the candidate distribution are estimated by using the Newton-Raphson method. The MLE of the δth quartile can be as the quality measure for assessing the system’s quality. To illustrate the application of the proposed method, an example of high-speed motor with interval-censored lifetime data is given, and extended Monte Carlo simulations are carried out. The simulation results show that the proposed method is efficient for model identification and can provide reliable reliability assessment.


Keywords Akaike information criterion, field failure data, location-scale family, maximum likelihood estimation
    Article #:  RQD25-205
Proceedings of 25th ISSAT International Conference on Reliability & Quality in Design
August 1-3, 2019 - Las Vegas, NV, U.S.A.