Post-Storm Vehicle Routing for Distribution Grid Restoration: An OLUCT Based Learning Approach

Natural disasters such as storms usually bring significant damages to distribution grids. Many distribution grids are not equipped with sensors that can pinpoint the location of actual faults. This further complicated the outage restoration process in distribution grids. This paper focuses on the optimal routing of repair vehicles to restore outages in the grid as fast as possible after a storm. First, the vehicle routing problem is formulated as a sequential stochastic optimization problem. In the optimization model, the belief state of the power grid is updated according to the phone calls from customers and the information collected by the repair vehicles. Second, an Open Loop Upper Confidence Bound for Trees (OLUCT) algorithm based stochastic lookahead policy is utilized to achieve the realtime dispatching of the repair vehicles. Simulation results show that the proposed approach can effectively navigate the vehicle to repair all outages.