An Approach to Estimate the Location Parameter of Weibull Distributed Product Life Based on Simulated Big Data  
Author Liyang Xie


Co-Author(s) Chenggang Li; Ningxiang Wu


Abstract For the situations of only 3-5 life data or right-censored life data available, the difficulty to use three-parameter Weibull distribution to describe life random variable comes from the estimation of the location parameter. To estimate the Weibull location parameter in the situation of small size sample of right-censored life data, a great number of simulated life data are generated through Monte Carlo sampling, by which the relationship between the minimal observation in the life sample and the location parameter can be obtained for a specified Weibull probability density function. A baseline sample size, with that the location parameter can be appropriately assigned as the minimal observation, is estimated by means of the median rank estimator with an arbitrary confidence level. Based on the relationship between location parameter and the minimal observation in a sample of size n, an iteration procedure is designed to estimate the Weibull location parameter according to small size of sample life data. The three-parameter Weibull distribution with such estimated location parameter is evidently better than the conventional two-parameter Weibull distribution for bearing life description.


Keywords Product life, three-parameter Weibull distribution, small size of sample, censored life data, big data
    Article #:  RQD25-259
Proceedings of 25th ISSAT International Conference on Reliability & Quality in Design
August 1-3, 2019 - Las Vegas, NV, U.S.A.