Markov Chain Monte Carlo Bayesian Analysis of the Nonlinear Characteristic of a Three-Phase Alternator

We have applied the methods of the Bayesian probability theory as an alternative to the Potier's triangle construction to rigorously analyze the nonlinear characteristics of a saturated alternator. This analysis comprises the choice of the prior probabilities, the setting up of the models, the calculation of multidimensional integrals with MCMC sampling methods as implemented in the winBUGS software and the discussion of the results. Our objectives have been to clearly illustrate the main advantages of the method: first, its ability to take into account all the cogent information previously available about a given problem; second, the parameter estimation feature and, third, the possibility of performing a true model comparison.