Simulated Annealing Based Fuzzy Markov Game Energy Management for Smart Grids

For the energy management problems for demand response in electricity grid, a simulated annealing based fuzzy Markov game consumer energy management controller (CEMC) is proposed. Because of the uncertainties and highly time-varying, it is not easy to accurately obtain the complete information for the consumer behavior in electricity grid. In this case, the fuzzy Markov game offers a promising platform for robust control in the presence of the bounded external disturbances. The energy usage uncontrollable is considered as a disturber. The fuzzy inference and simulated annealing are introduced in order to facilitate generalization in large state space and balance exploration and exploitation in action selection in Q-learning individually. Simulation results show that the proposed controller can learn to take the best action to regulate consumer behavior with the features of lower end-user financial costs and higher consumer satisfaction.