Estudio comparativo de métodos de detección e identificación de fallas (FDI) de sensado sobre el control local de micro redes eléctricas híbridas AC/DC

Abstract

The present research develops a comparative study of three different methods that are applied for the detection and identification of faults (FDI). The faults are sensing in hybrid electrical microgrids (MG) in AC / DC. The study addresses the use of methods based on: Kalman Filter, Artificial Neural Networks and Fuzzy Logic, all applied to local MR controllers. For the comparison and validation of the performance of the proposed methods and the three researched methods, three failure conditions were proposed: operation without failure, abrupt failure or loss of sensing and an incipient additive failure. For the three methods studied and developed, it can be concluded that the Kalman Filter is faster in its execution and decision-making, however, it can be observed that the method based on Fuzzy Logic presented a lower average residual for the 3 analyzed cases. All simulations were developed in Matlab / Simulink, software that is specialized in simulating dynamic control models. Finally, an algorithm based on the minimum error was proposed in order to permit the automatic selection of one of the FDI models studied.