Uma adaptação do estimador de estado trifásico para sistemas de distribuição de energia elétrica baseado no método da soma de potências

Abstract

State estimation is an essential mathematical tool for the real-time monitoring of power transmission and distribution systems. Unlike the classical load flow methodologies, state estimation can deal with uncertainties associated with measurements. Its application started in the study of transmission systems, so it is consolidated in this type of configuration. However, in distribution networks, the use of state estimators is less frequent due to the typical characteristics of radiality, imbalance and limited number of measurements, which increase the complexity of the problem. In order to contribute to the state of the art of state estimation in electric distribution networks, this work presents an adaptation of methodologies for generation of pseudomeasurements and state estimation present in the work of Monteiro (2010). For this, the equation of adjustment of loads was adapted in order to consider the losses in all the sections that composes the distribution network. In addition, was proposed a technique that uses the normal probability distribution to determine the most probable value of power factor that all loads present. In the estimation technique, modifications were made in relation to the number of reduction sections and the choice of state variables and measured quantities. Two test cases were used to validate the technique, one with actual measurements and another with simulated measurements with a load flow program. In the real system, the solution of the proposed methodology was compared with Monteiro's (2010) work. In this case, it was verified that the methodology developed in this work provides a more accurate solution for the estimation of current and voltage modules, since the maximum error for these two variables was 4,27% and 4,83 %, respectively, in the analyzed situation. However, for the same case, the reactive load flow estimated in one measurement point presented an error of about 31%. For the simulated system, which used measures from a load flow program, however, a maximum error of 4,21% was obtained for reactive power estimation. Therefore, the developed methodology proved to be feasible in the determination of the state of the network, contributing to smaller estimation errors than those obtained in Monteiro (2010), with respect to voltage and current values