Optimal Reconfiguration of Distribution Network Using $\mu$ PMU Measurements: A Data-Driven Stochastic Robust Optimization

The proliferated penetration of renewable resources along with stochastic consumption pattern of electrical vehicles have arisen the prominence of real-time monitoring of the grid for the sake of obtaining the optimal topology of distribution network. This paper proposes a data-driven method based on the measurements of $\mu $ PMUs to figure out the hourly optimal configuration of distribution grid in a real-time manner. First, the node voltage and injected current phasors measurements captured by $\mu $ PMUs are processed via a linear state estimation to determine the net load at each node. Then, the real-time high resolution data of loads is turned into knowledge through a bi-level unsupervised information granulation technique. In the second stage, based on the uncertainty bounds obtained for each information granule, a stochastic robust optimization (SRO) is developed via second order conic programming method to find out the best network reconfiguration, while minimizing the corresponding objective cost function. The developed method is applied to IEEE 33-node distribution network and Brazilian 135-node test feeder.