This paper proposes two methods for combining optimization based redispatch and reinforcement learning for grid reconfiguration. A relaxed ACOPF formulation of the redispatch problem is introduced, as well as an iterative method for finding the highest feasible line capacity margin. The performance of the combined redispatch and grid reconfiguration methods is evaluated through a case study using the IEEE 14-bus system. It is found that the combined methods are able to significantly reduce redispatch costs while maintaining maintaining security levels comparable to OPF-based redispatch alone
QC 20261020