A Modified Branch-Current Based Algorithm for Fast Low Voltage Distribution Grid State Estimation using Smart Meter Data
We present a modified version of the branch-current based weighted-least-squares (WLS) state estimation for application in three-phase low-voltage power distribution grids with a single transformer connection to the superordinate medium-voltage level. In our approach, voltage magnitude measurements are utilized to estimate the slack node voltage in a backward-sweep procedure separated from the WLS loop. As a result of excluding the voltage measurements from the WLS loop, the measurement functions are linear in the state variables as long as only power and power flow measurements are considered besides voltage measurements. Therefore, the computational complexity is significantly reduced compared to the straightforward way of including the nonlinear voltage magnitude measurement equations in the WLS loop. The proposed method is numerically evaluated in time-series simulations using various low-voltage benchmark grids. The results in terms of accuracy and speed are compared to the nonlinear node voltage based WLS approach as well as to a linear sensitivity-based method. It is shown that our algorithm yields an accuracy similar to the nonlinear WLS approach while requiring significantly less computation time.