GENERALIZED RENEWAL PROCESS PARAMETERS ESTIMATION VIA MAXIMUM LIKELIHOOD AND GENETIC ALGORITHM

Although “as good as new” and “as bad as old” assumptions are usually adopted to describe the system state after a maintenance, they occasionally do not have a practical use. Generalized Renewal Process (GRP) is a point process capable of modeling all classes of repair, including the imperfect repair, which is a more realistic assumption. In this paper, we propose an approach to find the estimators of the Weibull-GRP parameters, by maximizing the likelihood function using Genetic Algorithm (GA). We compare the proposed solution with results in the literature, obtained through Monte Carlo Simulation (MCS). Results show that our methodology reach a smaller error through MSE criterion and GRP present a better fitting to the real data, when compared with the Renewal Process (RP) and the Non-Homogeneous Poisson Process (NHPP).