Comparison Semiparametric Bayesian Method with Classical Method for Estimating Systems Reliability using Simulation Procedure

Abstract

In this research, the semiparametric Bayesian method is compared with the classical method to estimate reliability function of three systems : k-out of-n system, series system, and parallel system. Each system consists of three components, the first one represents the composite parametric in which failure times distributed as exponential, whereas the second and the third components are nonparametric ones in which reliability estimations depend on Kernel method using two methods to estimate bandwidth parameter h method and Kaplan-Meier method. To indicate a better method for system reliability function estimation, it has been used simulation procedure for comparison and different sample sizes of size (14,30,60 and 100) using standard comparison Integral Mean Square Error (IMSE). For k-out of-n system, the results indicate that it is better to use Bayesian method for samples of size (30,60 and 100), and to use the classical method for samples of size (14), whereas for series system the best method to use is Bayesian method for samples of size (14,60 and 100) , and to use the classical method for sample of size (30). for parallel system, it is better to use Bayesian method for all sample sizes.