Research On Mechanism For Demand-side Management In Smart Grid

Smart Grid is the future direction of development for power grid all over the world. Compared to the traditional power system, Smart Grid builds an advanced metering infrastructure via Smart Meter that advanced technologies such as information, communication and computer network are integrated into the traditional power system. Optimal allocation of power resources could improve stability for grid while the introduction of renewable energy and distributed energy storage device could gain the incomes and energy saving. The current research focuses on security and stability on the supply-side while rational decision for the demand-side.In this dissertation, we consider the interaction between the action of users and also between the one of users and grid, meanwhile an effective price mechanism is proposed based on contradiction between supply and demand and game theory. With the promotion of distributed energy storage device in the smart grid, the role of users is changed from passive acceptor to active participant, and both generation and consumption are available for them which is the basic reason to formulate a flexible mechanism of price. Moreover, an effective mechanism not only improves profit for users but performs load shifting for load profile.Firstly, considering the distributed energy storage device, the load model of users is divided into basic load and strategy of distributed storage device, and then, a day ahead optimization mechanism is proposed by the relationship between price and load. On the supply-side, a centralized algorithm for load shifting could obtain a reference load, added into price to lead the action of users. Meanwhile, a non-cooperative game is built on the demand-side, in which we consider the convergence of best response. According to the relationship between best response and the strong monotone for the objective function, we reformulate the object and improve a new iterative algorithm. As a result, while the Nash Equilibrium is reached, the load profile performs load shifting.Secondly, as load of users performs random and private, we extend the load model which a new concept of load preference. Users could observe their own load preference as a priori and we assume that there is an interactive relation between load preference of users that is expressed via covariance matrix. A Bayesian game is formulated based on the cost function of users and the priori of load preference and partial information. Moreover, the result of best response for the Bayesian game performs load shifting while the cost of users is minimize.Finally, as the interactive is difficult to quantify, a more general model without the interactive relation of load preference such as covariance matrix is proposed. A mechanism for date exchange is built among users via communication function of smart meter. Users on the demand-side observe the load strategy of others and estimate the distribution of their load preference, and then, obtain the result of the dynamic Bayesian game by best response. Considering the fact that the relation of the object to be estimated and the observations is linear dependent, we propose sequential LMMSE (Linear Minimum Mean Square Error) to estimate the load preference of others. Above all, the proposed algorithm also performs load shifting while gaining the profit.