Distributed Multi-Agent based Load Shedding in Power Distribution Systems

This paper proposes a service restoration algorithm in distribution systems with Distributed Generation (DG) and considering priority customers. When a fault occurs in Power Distribution Systems (PDS), it is essential for the utility companies to restore the service to their customers as soon as possible. To restore the service, faulty zones should be isolated first and then power could be restored to fault-free zones if enough power is available for restoration. If the provided power from DG sources and other supporting feeders is less than required power for restoration, some loads should be shed. A distributed Multi-Agent based Load Shedding (LS) technique has been used which can make efficient load shedding decisions to supply as many of higher priority customers as possible, while keeping the balance between the demand and generation. Agents have a load prediction capability and use the predicted values for restoration decision makings to prevent short-time service restorations. The restoration problem is modeled as a nonlinear optimization problem with the objective function of maximizing the power supplied while satisfying the power system and priority considerations. The simulation results show that the proposed algorithm is capable of making feasible LS decisions to minimize the non-supplied priority loads.