Distributed Energy Management And Optimization In Smart Grid

With the integration of smart infrastructure,advanced management system and intelligent protection system in the traditional power grid,the smart grid composed of multi-agent system emerges at the moment.Smart grid belongs to a power network integrated with information security,automation and computer control and other advanced electronic technologies and make it possible to manage the energy system intelligently,efficiently,safely and economically.Nowadays,under the background of carbon neutrality and peak,the development of smart grid will be broader as the control and optimization of smart grid is an important approach to support the carbon neutrality and peak,mitigate the warmhouse effect and overcome global warming.As a key component of smart grid,the energy management guarantees it operating in an efficient,stable and reliable way.Therefore,it is always a popular research topic.With the continuous penetration of distributed generation and large scaling of power network,the centralized dispatch is not suitable for the traditional power grid anymore.Instead,the distributed energy management in smart grid has flexible scalability and robustness to single point of failure.It maximizes the utilization of energy consumption by coordinating each controllable and communicating node.It also provides great opportunities for innovation of energy internet.After many years of extensive research,the distributed energy management in smart grid has made great progress.However,the existing distributed energy management researches have three deficiencies:i)Penalty function and project method are generally used in the research of energy management for the interval constraint of power.The former results in a gap which is caused by the penalty parameter between the new model and the original model.While the latter is a special case of proximal operator,thus it is necessary to study the composite optimization problem with nonsmooth part in a general way;ii)Many distributed energy management studies ignore the situation that the nonsmooth part in objectives is complex in practical.Besides,the subgradient-based algorithm is usually designed instead of taking advantages of problem structure.Meanwhile,it ignores coupled inequality constraints apart from the power supply-demand balance constraint and power internal constraint;iii)The study of dynamic economic dispatch of renewable power generation and energy storage system lacks simple and efficient algorithms.Based on the latest research progress at home and abroad,this paper studies from static composite energy management to the dynamic economic diapatch problem with wind power and energy storage,and the corresponding algorithm is proposed.The main work and contribution of this paper can be summarized as follows:1)For traditional energy management problem,the problem is transformed into a more general nonsmooth composite optimization problem model considering transmission losses,which is analyzed from a new perspective of composite optimization.The energy management is solved by the designed distributed proximal primal-dual algorithm,according to the structural characteristics of the objectives in the optimization problem and proximal operator.The proposed algorithm is verified by theoretical and simulation analysis.It shows that the convergence speed is faster than the traditional nonsmooth optimization algorithms.2)For complex economic dispatch problem with general coupled inequality constraints in which more complex nonsmooth part is considered.We introduce three operators splitting technique,proportional integral strategy and dual decompositions to design a proximal operator based primal dual algorithm.By convergence analysis,the proposed distributed optimization algorithm reaches at a sublinear convergence rate and is verified in an economic dispatch problem.3)For dynamic economic dispatch problem with wind power and energy storage system is considered.We design a distributed dynamic economic dispatch algorithm based on alternating direction method of multipliers.Under this framework,coupled dynamic economic dispatch problem is decoupled into static economic dispatch model.The effectiveness of the proposed algorithm and peak load shaving are demonstrated in simulation.In the end,the thesis is concluded and some future works are prospected.