Research On Distributed Economic Dispatch Problem Of Smart Grid Based On Consensus Algorithm

In the past decades,smart grid technology has developed rapidly,but there are also many problems and challenges in the development process of smart grid.Among them,the economic dispatch problem of smart grid is to find the optimal power output of multiple power generators at the lowest possible cost in the short term under consideration of various constraints to meet the overall load of the system,it is essentially an optimization problem.It is of great significance to study the economic dispatch of smart grids to ensure the economical and stable operation of smart grids.However,with the continuous expansion of smart grid scale and the increasingly complex network structure,the economic dispatch problem in smart grid has become more complex and challenging.Combined with convex optimization theory,penalty function method and multi-agent system consensus algorithm,this paper studies the distributed economic dispatch problem of smart grid,and proposes an effective optimization algorithm.The main work and research results focus on the following two aspects:(1)This paper studies the economic dispatch problem of smart grid that is more suitable for reality,aiming at the optimization problem of smart grid economic dispatch under multiple constraints such as capacity constraints,ramp-rate constraints,and supply-demand balance constraints for generators,renewable energy and energy storage devices,the original optimization problem is transformed into an optimization problem under unconstrained conditions by using the internal penalty function method and Lagrange dual theory.And based on the unconstrained optimization problem combined with the consensus algorithm,a distributed second-order consensus algorithm is designed,and the algorithm is proved to be convergent in theory.Finally,the feasibility of the algorithm is verified by IEEE six-bus system simulation,and compared with the traditional consensus algorithm,it has a faster convergence speed and convergence accuracy.(2)Since the penalty factor tends to zero,the augmented objective function will be illconditioned and the optimal solution of the algorithm will deviate from the real optimal solution,that is,the total power generation cost of the smart grid has not been optimized.Aiming at this problem,this paper introduces a smooth exact penalty function.When the penalty factor satisfies certain conditions,the penalty problem constructed based on the smooth exact penalty function is equivalent to the original smart grid economic dispatch optimization problem.The original economic dispatch problem is reconstructed through the smooth exact penalty function,and the distributed second-order consensus algorithm is modified and adjusted.Since the selection of the initial point of the algorithm is relatively strict,that is,the initial point must meet the supply and demand balance constraints,and the change of the initial point will affect the convergence speed and convergence accuracy of the algorithm,which greatly reduces its practicability.Therefore,an initialization process is added to the algorithm,thereby avoiding the selection of the initial point and the adjustment of the algorithm parameters.Finally,the feasibility of the algorithm is verified by simulation experiments.