System Identification in Distribution Grid Without Phase Angle using Expectation Maximization

Abstract

There has been a large increase in distributed energy resources (DERs) on the distribution grid (DG) and it is expected to continue. To support this increase active grid management and optimisation is required, for which accurate information of system state variables is required. This study presents a method for simultaneously estimating the line parameters and system state variables; active and reactive power at nodes. In addition, the study estimates the voltage magnitude accurately and requires no information about phase angle. The proposed approach uses a novel combination of Linearized Distflow and expectation maximisation (EM). The proposed approach is demonstrated on the IEEE 37 node test feeder and achieves accurate estimates of line parameters and state variables. Furthermore, the results show that the estimated line parameters give better voltage estimates using Linearised Distflow than the true parameters from the AC model which we argue is due to the estimated parameters making up for the approximation error.