Collaborative , Par Autonomous

Balancing electricity supply and critical for the stable performance of Grid. Demand Side Management (DSM) r consumers’ energy usage to off-peaks as muc avoid more electricity demand than available su times. Artificial intelligent planning algorit applied to enabling electric devices to reschedu to off-peaks. One such algorithm is Monte C (MCTS), which takes advantage of tree sear sampling on decision space in order to find an decision. In particular for DSM, MCTS has been use smart meters and also electrical devices. In b MCTS acts as a centralized consumption plan optimum approach for all devices in the sp computation thread limits these approaches in t and scalability. As applied in domains ou alternative, decentralized MCTS algorithm, MCTS (P-MCTS), allows every agent to run MCTS thread to have its own solution. In this paper, first we studied the feasibili MCTS in DSM to make demand planning m scalable. P-MCTS has been evaluated by runn scenarios one with 6 electrical vehicles (EVs) a 90 EVs which roughly results in 30% peak loa MCTS. To improve the results of P-MCTS, w decentralized collaborative approach, called MCTS (CP-MCTS), which exploits and exte enable each electrical device to actively aff process but also to improve the final decision knowledge obtained during the collaboration. comparison with P-MCTS, CP-MCTS obtaine including more peak load shifting and smoothe to 30% lower Peak to Average Ratio (PAR). Keywords— Monte Carlo Tree Search, Colla Side Management.