A combined self-organizing map neural network with analysis graphical approach for mixed-weibull parameter estimation

Abstract

The mixed-Weibull distribution is widely used to analyze the burn-in time. Kececioglu had presented its parameter estimation method with application of Weibull probability plot (WPP) such a graphic analysis method. However his method is not easy to estimate parameters when the data loses the failure mode information. A self-organizing map neural network (SOM) is used to cluster the classification of failure mode. We combined SOM with Kececioglu¿s method to estimate the parameters of mixed-Weibull distribution. Some simulation studies are given to present the accuracy of parameter estimation of our method under small sample size.