Tirza Routtenberg

The problem of bus imbalance identification in a three-phase power network using a phasor measurement unit (PMU) is considered. We propose new algorithms to identify and localize imbalances occurring at any location in the power grid, based only on single-phase PMU data. First, we develop a technique using the minimum description length (MDL) criterion that is carried out at the control center. The centralized MDL methodology is time consuming and has computational complexity that grows exponentially with the network size. Therefore, we next develop a projected orthogonal matching pursuit (POMP) algorithm, which is a low-complexity method for bus imbalance identification in large-scale power networks implemented at the control center. POMP is a computationally efficient compressive sensing technique that exploits the sparse structure of the voltage measurements. The proposed methods are validated through three case studies: A two-port \pi-model, an IEEE-14 bus system, and an IEEE 118-bus system. Simulations show that our networked algorithms, MDL and POMP, obtain improved performance over local bus-level identification techniques. The performance of POMP is close to that of centralized MDL, with the advantage of being applicable to large-scale networks.