Research On Distributed Optimization Algorithm For Dynamic Economic Dispatch Of Energy Systems

Distributed optimization theory and application research have become an important development direction in control science and engineering,and have been widely used in communication networks,power systems,sensor networks and other information physical systems.The distributed method based on multi-agent theory can avoid high calculation,communication cost and privacy leakage caused by centralized algorithm.Aiming at the dynamic economic dispatch problem under the background of microgrid,this thesis studies the design and analysis of distributed optimization algorithm.The following points are the main motives of this thesis:(1)A dynamic economic dispatch model containing the cost of traditional energy,renewable energy and power generation pollutants is established,and a distributed optimization algorithm for undirected networks is designed.In the dynamic economic dispatch optimization problem,it is necessary to consider the supply and demand balance,ramp-rate constraints and generator capacity constraints to ensure that all kinds of constraints are guaranteed in different time windows.At the same time,considering the potential transmission loss in the transmission process,the equation constraint of supply-demand balance is transformed into a globally coupled inequality constraint,which further increases the difficulty to solve the problem.In the process of algorithm design,the limitation that the generator needs to meet the load demand in the initial state is removed,which makes the algorithm more universal.(2)Aiming at the social welfare maximization problem of smart grid,a dynamic economic dispatch model with transmission loss is established.By transforming the form of transmission loss,a class of social welfare maximization problem based on non-smooth dynamic programming is obtained.Under the assumption that the communication network is a directed equilibrium graph,a distributed optimization algorithm based on directed equilibrium graph network and a non-smooth adaptive completely distributed optimization algorithm are designed respectively.In order to make the designed algorithm more universal and meet the needs of real communication,a non-smooth distributed optimization algorithm based on directed unbalanced graph network is designed.For the distributed optimization algorithm,the convergence of the distributed optimization algorithm is proved by constructing different Lyapunov function.(3)A kind of uncertain dynamic economic dispatch model is established To deal with the impacts of uncertain factors caused by line aging,transmission loss and external environmental factors in smart grid.Using the duality principle in convex optimization theory,a robust resource optimization problem is obtained,and a robust distributed continuous time algorithm is proposed in which each agent depends only on its local objective function,local uncertain parameters,local constraint set and its neighbor information.By using differential inclusion theory,the convergence of robust distributed optimization algorithm is proved under more looser conditions,and furthermore verifies the effectiveness of the proposed algorithm.