Novel distributed algorithms for power systems operations.

摘要

Power grid is probably the most complex engineering system in the world. The dominated centralized schemes for power system require complicated communication network to collect global operating condition and a powerful central controller to process the huge amount of data. In addition, centralized schemes lack flexibility and cannot easily adapt to intentional/unintentional network topology changes. Since distributed control schemes are more flexible, reliable, and cheaper to implement, they are promising choice for smart grid upgrade.;To address the needs of power systems and problems with existing solutions, the author proposes several fully distributed multi-agent system (MAS) based algorithms based on the most recent developments of control and optimization theory, which have been successfully applied to load management, optimal reactive power dispatch (ORPD), cooperation of multiple renewable generators (RGs), and optimal resource allocation.;For load management, a fully distributed consensus based global information discovery algorithm is proposed to discover the global net active power, loading level and load connection information. The information discovery algorithm can guarantee convergence for power systems of any size and topology and is robust against certain types of faults. For OPRD, a distributed MAS based Q-learning algorithm is utilized to coordinate the control settings of reactive control devices to minimize the active power loss while satisfying certain operation constraints. The proposed algorithm is model free and can adapt to changes of system operating conditions automatically. For the coordination of multiple RGs in a microgrid, a distributed subgradient-based control scheme is proposed to synchronize the utilization levels of RGs based on local frequency measurement and available generation prediction. The proposed control scheme can be applied to the power systems with different types of generators under both excessive and insufficient renewable generations. For optimal resource allocation, a fully distributed pricing strategy is proposed to maximize the social welfare, while respecting system constraints.;The effectiveness of the all proposed algorithms has been demonstrated through simulations. The corresponding work has produced five papers in the IEEE transactions on Power Systems, Smart Grid and Systems, Man, and Cybernetics.