Comparing Different Estimators of Parameters and Reliability for Mixed Weibull by Simulation

Abstract

This paper concerned with estimating reliability function of one of the important failure models, which is used when the underlying population is non homogenous, this model is called mixed Weibull distribution, when there are only two sub – population with mixing proportions taking into account the case when items that fail can classified and can be attributed to appropriate sub – populations. Two methods were used to estimate the reliability function which are; (1) Maximum likelihood Method (ML). (2) Weighted Least Square Method (WLS). The simulation procedure using Monte Carlo Method, are used and several experiments are implemented to find the best estimators which have smallest mean square error. All results are explained in tables.