Approximate power flow solutions-based forecasting-aided state estimation for power distribution networks

This paper presents an approximate power flow model-based forecasting-aided state estimation estimator for power distribution networks subject to naive forecasting methods and nonlinear filtering processes. To this end, this estimator designs a voltage perturbation vector around the priori-determined nominal value as the dynamic state variable, which enables more detailed depictions of voltage changes. Then, a state transition model incorporating nodal power variation is derived from the approximate power injection model. The constant state transition matrix working on power variations only consists of nodal impedance, which reduces the extensive parameter tuning effort when facing different estimation tasks. Furthermore, an approximate branch power flow observation equation is proposed to improve the filtering efficiency. The observation matrix with branch admittance information presents the linear filtering relationship between power flow measurements and forecasted states, omitting the complex iterative updates of the Jacobian matrix for nonlinear measurements. Finally, the overall estimated voltage state at each time sample is entirely obtained by combining the filtered voltage perturbation vector with the priori-determined nominal value. Numerical simulation comparisons on a symmetric balanced 56-node distribution system verify the performance of the proposed estimator in terms of accuracy and robustness under normal and abnormal conditions.