Service restoration is important in distribution networks following an outage. During the restoration process, the system operating conditions will fluctuate, including variation of the load demand and the output from distributed generators (DGs). These variations are hard to be predicted and the load demands are roughly estimated because of absence of real-time measurements, which can significantly affect the restoration strategy. In this paper, we report a robust restoration decision-making model based on information gap decision theory, which takes into account the uncertainty in the load and output of the DGs.
For a given bounded uncertain set of parameters, the solutions can ensure feasibility and that an objective does not fall below a given threshold. We describe the implementation of a robust optimization algorithm based on a mixed integer quadratic constraint programming restoration model, the objective of which is to restore maximal outage loads. Numerical tests on a modified Pacific Gas and Electric Company (PG&E) 69-node distribution network are discussed to demonstrate the performance of the model.