Local and Remote Estimations Using Fitted Polynomials on Distribution Systems

This work describes a technique to define a single parametric equation that estimates remote conditions within a distribution network, in an online setting, without dedicated telemetry. In this novel approach, a departure from conventional state estimation is explored to facilitate a fully decentralized operation. Derived based on fundamental treatment of ac power flow, the electrical behavior of a section of network is defined to a tractable constraint space using regression analysis. In this non-iterative technique, a measurement set consisting of the local voltage magnitude, active and reactive power injections at a single node are the input to pre-computed polynomials. Remote current flow, active power and reactive power flow as well as remote and local voltages and sensitivities are estimated in a direct calculation from period to period. Accurate estimations are also found in the presence of transducer errors. To assess the applicability of this technique at differing voltage levels a range of reactance to resistance ratios are considered.