Phase Identification of Single-Phase Customers and PV Panels via Smart Meter Data

With proliferation of single-phase rooftop photovoltaic (PV) panels, phase balancing in low voltage (LV) distribution feeders becomes the point of concern. In this way, identification of the hosting phase of connected single-phase customers and PV panels is a prerequisite. This paper proposes an optimization model for the phase identification problem. The objective is to minimize the summation of the absolute error between estimated and measured variables. Smart meters (SMs) data including active and reactive power absorptions/injections, nodal voltage magnitudes, and network configuration data form the input of the model. Potential errors in the input data are captured in the model while their adverse effects are tackled by incorporating data associated with numerous consecutive time intervals. The problem is thus inherently large-scale with prohibitively heavy computation. Accordingly, the proposed method takes Benders decomposition in use to make the model tractable. The final model lies in the mixed integer linear programming (MILP). For the sake of performance verification, three different feeders with 34, 457, and 906 nodes along with SM data are studied and discussed thoroughly.