Real-time Identification of Load and Upstream Network Models in Distribution System Operation

Abstract

In distribution system management, if the system models are not accurately identified, the anticipated cost reduction is not realized, and decisions might violate constraints. Conventional state estimation (SE) evaluates nodal active/reactive demands, but cannot estimate the other parameters of load models, e.g., constant impedance shares, and the parameters of upstream system model, i.e., Thevenin voltage and impedance. This paper shows such parameters profoundly affect the optimality/feasibility of the operation strategies, and proposes an algorithm to estimate these parameters. To this end, two distinct sets of measurements are required. The first set entails the current values of measurements. The second set must be achieved by the intentional changes of control variables in the downstream system, referred to as control actions, e.g., increasing the power production of distributed generators. Instead of costly random changes, using an additional optimization, the proposed identification algorithm finds the minimal subset of control actions that adequately change the measurements. By limiting the control actions to those that reduce the system cost or mitigate the constraint violations, unnecessary control actions are avoided in the identification step. As the identification problem is recast as a computationally cheap optimization, it can be integrated into the real-time operation of distribution systems.