Learning Topology of Meshed Microgrids Using Voltage Magnitude

Topology identification plays an important role in situational awareness, state estimation, and cyber security in microgrids. We show that existing algorithms for topology detection can be enhanced to reconstruct power grid topology solely from voltage measurements taken at different buses instead of both voltage and phase measurements. It is shown that, in power networks with dominantly resistive lines, the active power flow depends only on the voltage magnitudes at the buses and not on the phase differences. Using the independence of active power form phase, we establish that the inverse of covariance matrices estimated from voltage magnitude samples alone is enough for complete reconstruction of power network topology. The ability to reconstruct topology solely from voltage magnitude samples reduces the reliance on costly phasor measurement units (PMUs) where cheaper RMS voltmeters suffice. Three different algorithms are developed to perform topology reconstruction and change detection in power grids. To demonstrate the efficacy of the algorithms, results based on MATPOWER simulations as well as OPAL-RT based real-time simulation of a modified IEEE 33 bus network is presented.