Posterior Distributions for Bayes Assessment of Photovoltaic Inverter Reliability and Availability

The paper deals with the quantitative assessment or prediction of photovoltaic inverter system performance as measured by its reliability and availability, with the aim of developing an efficient and robust evaluation the above performance, in the presence of uncertainty due to a scarce knowledge of the basic parameters affecting such evaluation (e.g., the hazard rate and repair rate). The problem is dealt with by means of a Bayesian method of inference, in which the parameters to be estimated are considered as random variables. The paper illustrates – by means of numerous numerical simulations - the efficiency of a new Bayesian procedure of estimation. An extensive numerical analysis is also performed, showing that the efficiency of estimation method and its robustness with respect to chosen prior distributions.