Active Distribution System State Estimation: Comparison Between Weighted Least Squares and Extended Kalman Filter Algorithms

Abstract

Power distribution systems have a topology which is typically unknown to the distribution system operators. Remote Terminal Units and SCADAs monitor these networks primarily at the substation level. However, with the widespread integration of Distributed Generation units (DGs), the need for real-time control of Active Distribution Networks is urgent. While DGs can improve the performance of power systems through voltage support, price elasticity, and reduced emissions of greenhouse gases, they also present challenges such as voltage spikes and bidirectional power flows. The distribution systems' state needs to be known accurately with high refresh rates and low time latency to deal with these issues. Real-time state estimation (SE) that use of Phasor Measurement Units (PMU) data allows the prediction of the distribution systems' nodal voltages and phasor angles. This paper presents a performance analysis comparison between the Weighted Least Square (WLS) and the Extended Kalman Filter (EKF) algorithms on active distribution grids. The WLS is a static SE algorithm, while EKF is a recursive SE method. The paper first recounts the analytical formulation of both approaches and then quantifies the differences in their performance. The tests were carried out on a modified IEEE-33 bus test feeder that included an optimally placed DG. For the test feeder's nodes load profile, the PMU-data generated during the ADRES-CONCEPT project was used. MATLAB and OpenDSS software were used to run the experiments. The results show that if the process model is correct, the EKF approach performs better.