Topology and Parameter Identification in Electrical Distribution Systems using Spatial Priors

This manuscript presents novel methods that allow the consideration of spatial priors derived from Geographic Information Systems (GIS) for System Identification (SI), i.e., Topology Identification (TI) and Parameter Identification (PI), in electrical distribution systems. The proposed methods are designed to allow flexibility in the assumed measurement devices and the integration of micro-Phasor Measurement Units (µPMU) and Non-Phasor Measurement Units (NPMU), based on power flow approximations and GIS data associated with the location of measurement devices to deduct topological priors and cable and line parameter ranges based on spatial relationships between measurements. Based on a Mixed-Integer Quadratic Programming (MIQP) optimization problem, the proposed SI approach can handle measurement errors and noise. The presented method is demonstrated on a benchmark 17-node Low Voltage (LV) grid for three different scenarios, analyzing errors with respect to topology and parameters, as well as the computational effort. It is shown that by using spatial priors, the proposed SI method performs better than existing techniques.