Study On Multi-time Scale Economic Dispatch Method Of Microgrid Based On Distributed Optimization

The rapid development in world economy and over-exploitation of traditional energy resources pose a significant challenge in terms of energy demand,which makes the utilization of renewable energy as the best solution.Microgrid integrates the renewable energy distributed generation units and load into small controllable generation system,which significantly reduce the adverse impact on the main grid caused by the randomness and fluctuation of the renewable energy generation.The economic and environmental benefits of microgrid can be highly improved by optimal resources allocation,while demand response management provides alternative ways for economic dispatch of microgrid.However,traditional centralized optimization methods cannot match the distributed characteristics of the microgrid anymore.Therefore,it has been an urgent problem to develop new distributed optimization methods and study the economic dispatch methods based on distributed optimization.In order to realize the distributed optimization of micrgrid,this paper firstly studies the parallel and distributed optimization method with constraint decomposition,and then the distributed economic operation method with demand response and day-ahead and intra-day multi-time scale distributed economic dispatch method are studied.To improve the convergence speed of the current distributed optimization methods,a parallel and distributed optimization method based on decomposition of constraint is firstly proposed.Secondly,a bilayer optimization considering demand response is presented,in which the coordinated strategy of generation side and demand side is designed.Then the above model is solved by parallel and distributed optimization method.Thirdly,a day-ahead and intra-day multi-time scale economic dispatch method is proposed to decrease the impact of forecast error by intra-day rolling optimization,which is also solved by parallel and distributed optimization method.Finally,microgrid simulation platform is established with practical operation data and cases are designed to evaluate the performance of our proposed method.The main contents of this paper includes:(1)Parallel and distributed optimization method with constraint decomposition.Firstly,a multi-agent information exchange network is established,and the adjacency matrix and weighted matrix are derived,which are used to divide the network into local areas.Secondly,a decomposition method of objective function and equality constraint is presented to obtain local objective functions and local equality constraints corresponding to each agent.Each agent solves its local objective function and local equality constraints in parallel.After exchanging information with neighboring agents,constraint value is updated based on proposed method.Through multiple iterations,the optimal value can be gradually obtained.Then,the convergence of the proposed method is proved in theory,and the optimal value of the problem can be found in finite steps.Finally,four test functions are selected from standard test function library and three cases are designed to evaluate the performance of the proposed method.Results show that our method has a higher rate of convergence compared to traditional ADMM while having a relatively high solution precision,especially when the optimization scale grows.Besides,the solution precision of our method is not dependent on the topology of the communication network,and communication network with higher average out-degree has a higher rate of convergence.(2)Parallel and distributed economic operation method of microgrid with demand response.In order to reduce the operation cost,increase economic dispatch methods and improve prosumers’ benefits,this paper presents a parallel and distributed economic operation method.Firstly,a two-layer optimization model of microgrid is presented,in which the bottom layer is microgrid and top layer consists of agents,generation control center and load aggregator.Secondly,the minimum cost based generation side model and maximum profit based prosumer side model are established to achieve the overall optimization of both generation side and demand side.Then,the above models are solved by the proposed method to acquire the optimal operation plan.Finally,simulation platform of microgrid is established in MATLAB/Simulink and three cases are designed to evaluate the performance of our method.Results indicates that prosumers’ profits are improved with peak shaving and valley filling,while the generation costs are minimized by obtaining equal incremental costs of all DGs.Besides,the proposed method has strong robustness for communication line failures,which ensures the stable and economic operation of microgrid.(3)Day-ahead and intra-day multi-time scale distributed economic dispatch method.In terms of the impact of forecast error,a day-ahead and intra-day multi-time scale distributed economic dispatch method is proposed to minimize the impact by multi-time scale rolling optimization.Firstly,forecast model of renewable energy generation and load are established based on BP neural network to provide references,and day-ahead dispatch model composed of two sub-model including the minimum operation cost and maximum prosumer profits are established.For the power unbalance caused by forecast error,the minimum comprehensive adjustment costs are introduced to build intra-day rolling optimization model to minimize the impact on economic operation.Then plans for generation and purchasing or selling electricity from or to the main grid are obtained in day-ahead model,while the outputs of distributed energy storage system and adjustment for day-ahead plan are distributed optimized in intra-day model based on short-time forecast.Finally,simulation platform is established based on practical operation data and cases are designed to evaluate the performance of the proposed method.Results show that power fluctuation on the transmission line caused by forecast error and the intra-day comprehensive adjustment cost are reduced by intra-day rolling optimization.Moreover,the proper set of initial SOC of DESS can promote the economic benefit of our method.