Signal processing on graphs methodology for evaluating the load current variability in feeders with high integration of distributed generation

Abstract

The emergence of new elements as distributed generators in transmission and distribution networks is a challenge for supplying energy with quality, reliability and continuity. Although there are researches in the area, the impacts of distributed power generation (DG) in these systems are complex and involves many variables and due to this, power systems studies are essential to analyze the feeder behavior in this new scenario ensuring a good compliance with electricity quality levels. This thesis utilizes a graph-theory based model and proposes a novel method that associates concepts of power flow and graph signals to identify in distribution feeders with DG the cases in which the load current varies more for a steady-state analysis. The graphs were performed with Signal Processing on Graphs approach, where the nodes represent the feeder buses and the graph signal is a parameter calculated based on the power flow algebraic formulation. To fit and validate the methodology, the results are related with the current values obtained by solving a power flow problem, using MatLab/Simulink data. Results were presented for 13 and 34 bus electrical grid systems. The Spearman’s and Pearson’s rank-order correlation showed a good agreement between the results of the graph analysis and the Simulink data. The method proposes an alternative analytic way to identify the topological position of DG in a feeder that most impact in the current variation of the substation bus.