In this thesis, a new distributed stochastic load factor (LF) based energy management approach is proposed in the Stackelberg game framework. The game's leader is the distribution company (DISCO) that participates in the energy management game with two energy pricing mechanisms. In the first energy pricing mechanism, DISCO calculates the day-ahead energy pricing through maximizing its profit which is formulated as a stochastic conditional value at risk optimization problem. In the second one, a new pricing scheme is designed which discriminates individual customer's day-ahead energy price based on his/her contribution to the LF improvement. Customer's contribution is measured via the Shapley Value Method (SVM). Energy management Game's followers are price taker customers who participate in the distribution grid operation programs by holding a common constraint. The real-time uncertainty of renewable energies is tackled by customers' objective stochastic formulation in which day-ahead energy procurement and real-time scheduling of the energy storages and flexible loads are jointly decided. Depending on the number of game's equilibrium, the proper learning algorithms are designed for different structure of the customer's objective function. The convergence condition of the proposed energy management algorithms is investigated and proved. Numerical analysis confirms the effectiveness of the proposed stochastic energy management approach. Key Words: Smart grid, Shapley Value, K-means clustering, Stackelberg Game, Load Factor, Stochastic energy management
