Real-time Pricing Strategies For Smart Grids Based On Demand Side Management

Demand side management(DSM)implies management activities that encourage power users to actively participate in power operation management,optimize their usage of power and improve the efficiency of power consumption in response to the fluctuations of electricity prices or economic incentives provided by utilities.The aging of traditional equipment,the increasing demand for energy and the rising uncertainty caused by a large number of storage devices and distributed energy(DE)integrated into the power market,make the balance of energy provision more complicated.It is impossible to keep the balance between supply and demand simply by increasing supply from the provider.Meanwhile,increasing power supply capacity only will lead to the idleness of a large number of power supply equipments and the frequent switch of power supply equipment will increase the cost caused by the damage to machines.DSM can avoid installing new generation and distribution infrastructures,make people better use the existing power,improve the efficiency of electricity consumption,so as to keep the balance of supply and demand and reduce the load at peak hours or during system contingencies while maximizing the social welfare.DSM relys on demand response,and price response is the core of demand response.Moreover,real-time pricing(RTP)is one of the most efficient and economic demand response programs for competitive electricity markets to address the problem of residential energy consumption,for it can reflect instantaneous power information of supply and demand and guide users to adjust their demand for electricity and participate in the operation and management of the power system effectively.The social welfare maximization method is often used to study RTP,which takes the social welfare maximization as the goal and converts the original problem into the dual sub-problem of supply side and demand side through the Lagrange duality decomposition,so as to achieve the maximization of social welfare and load shifting while protecting the privacy of individuals.However,when using this method,the optimization model must be a convex programming.Otherwise,RTP can not be obtained by Lagrange multipliers.In addition,the exsiting works have limitations such as focusing on one single time slot rather than multiple time slots,lack of research on the smart grid with multiple poviders,etc.In this dissertation,RTP is further explored based on optimization theory,game theory,etc.The models are built for two cases,i.e,one supplier and multiple users,and multiple suppliers and multiple users,where the uncertainty,the distributed energy integrated into the smart grid and the patterns the users choose are considered,and then algorithms are designed to obtain RTP through solving the corresponding optimization problems.The main contributions of this dissertation are as follows:1)The RTP problem for the smart grid with different kinds of users is studied through online power fluctuation data.First,it classifies users into different kinds according to their power usage patterns and uses different prices to guide users' consumption in the same period.Second,aimed at peak-cutting and valley-filling,a RTP scheme is formulated as an optimization problem whose goal is to minimize the peakvalley difference,and then,asynchronous perturbation random approximation method(SPAS)is proposed to solve the proposed optimization problem.In addition,the properties of the model are discussed and the convergence of the algorithm is proved.The new method is easy to be implemented,because it only needs online data about the maximum and minimum of power produced by the random perturbation to achieve the optimal solution with an iterative formula of price,instead of knowing each user's privacies.Simulations show that this proposed approach is helpful for shifting load,avoiding load synchronization and improving the benefits of both users and power suppliers.2)A RTP scheme is formulated based on bilevel programming to tackle the uncertainties for smart microgrids equipped with renewable energy sources,dispatchable resources and storage devices.By Karush-Kuhn-Tucker conditions and smoothing method,the primal bilevel programming is transformed into an equivalent single-level optimization problem with only smooth equality constraints,and then a rolling penalty function algorithm is designed to obtain the optimal solution.The convergence of the smoothing method is demonstrated.Simulations show that the proposed approach has good performance in cutting peak,balancing system energy distribution and improving benefits for both supply and demand.3)A noncooperative game considering both spatially and temporally coupled constraints is used to formulate RTP scheme for smart grid with DE and storage devices to keep balance between supply and demand.In the proposed model,satisfaction maximization and cost minimization are equally considered.Meanwhile,DE is assumed to be kept for the user's own use or to be sold to the smart grid.Dual decomposition is introduced to transform the coupled-constraint game into a decoupled one,then an online gradient projection method is designed to obtain the optimal solution of the subproblem,and then a distributed algorithm is proposed to obtain the best response,through which we can further obtain the NE of the decoupled game.With this approach,the privacies of users and the power provider can be protected and system scalability is guaranteed.4)Considering the effect of the random fluctuation of electricity consumption,a distributed genetic RTP scheme for smart grid with multiple utility companies and users is proposed based on expectation bilevel programming.The primal uncertain bilevel programing is transformed into an equivalent deterministic optimization problem,and then a distributed GA is designed to solve the bilevel programming,through which not only can we obtain the fittest solution by means of genetic operators such as selection,crossover and mutation,but also the privacy of utility companies and users can be protected.Besides,when the random variables obey the normal distributions with the mean of zero,the proposed algorithm also applies to the deterministic situation,thus the situation considered in this paper is of generality.Simulation results validate the proposed distributed genetic RTP can significantly reduce peak time loading and efficiently balance system energy distribution while maximizing benefits for both utility companies and users.5)Considering that in a multiseller–multibuyer smart grid,each user cannot choose all utility companies simultaneously in each time slot,a RTP scheme for this smart grid to tackle the demand response is formulated based on sparse bilevel programming.In the proposed model,the correlation of different time slots and delivery cost of each utility company caused by different geographical distances between users and suppliers are also taken into account.Considering that the sparse bilevel programming is nonconvex,a bilevel distributed genetic algorithm is designed to solve it.The situation considered in this paper is of generality.Moreover,with this bilevel distributed genetic algorithm,subproblems are solved by users or utility companies individually,which ensures privacies and system scalability.Simulation results validate the proposed distributed RTP can significantly reduce peak time loading and efficiently balance system energy distribution while maximizing the social welfare.