Bayes parametric estimation of insulation reliability under distorted voltage

Estimating the useful life of power system components is of considerable importance both in planning, to properly assess the costs, and in the management of such systems, to effectively schedule maintenance programs. The paper discusses the reliability of the insulation of power system components on the basis of well-known life models of such devices, with the aim of parametric statistical estimation of such models. With particular reference to insulated cables for traction systems under distorted regime, an experimentally-based and commonly adopted mathematical relationship between the lifetime of a given insulation affected by non-sinusoidal voltage and the levels of the harmonic voltages applied to such insulation is adopted, that is of the kind of the well-known Inverse Power Model (IPM). A Bayesian inference method for the estimation of the above model is illustrated, when the Gamma Exponential prior distribution - a new model here proposed - holds for the basic parameters of the above Inverse Power Model. The performance of these estimators are empirically analyzed through extensive numerical simulations under a wide range of parameter values. All the results show the feasibility and efficiency of such Bayes estimation, especially for very small sample sizes, as requested for the above applications.