Minimização de multas pagas por concessionárias de distribuição de energia elétrica através da otimização dos intervalos de manutenção
Keywords: Distribution Systems, Monte Carlo Method, Particle Swarm Optimization, Reliability Centered Maintenance, Surrogate Models, Probabilistic Methods
AbstractThis paper proposes a methodology for vegetation management in overhead distribution networks that reduces costs related to tree trimming and penalties due to transgression in targets for SAIDI (System Average Interruption Duration Index). Another objective of this paper is to compare Surrogate Models, based on Lognormal and Pearson distributions, with the Monte Carlo Method (MCM) to estimate the SAIDI probability distribution. In addition, penalties were estimated considering voltage restrictions and circuit loading limits during load transfers. The multi-criterial Particle Swarm Optimization algorithm was used to calculate the vegetation pruning intervals. The tests carried out on the RBTS bus 4 system showed that the distribution utilities costs have more reductions when the optimization process is based on the estimated penalty than on the expected value for SAIDI. In addition, the use of substitute models, to obtain the SAIDI probability distribution, presented low computational cost and good accuracy in relation to the MCM.