Research On Virtual Power Plant Scheduling Strategy Based On Scenario

Under the pressure of energy shortage and environmental pollution, distributed generation has been paid more and more attention and has gained rapid development in the word and its proportion in power production has increased year by year. However due to the small installed capacity and wide distribution, it has brought new problems to the security, reliability and economic operation of network, especially the distributed clean energy. Virtual power plant is proposed to break the contradiction between the distributed generators and power grid and provide an effective way for the efficient and reasonable utilization of clean energy. Therefore, this paper takes the virtual power plant scheduling strategy as the research content, analyzes and discusses the scheduling method of generation units in the plant which provides theoretical basis for the scientific use of distributed energy.Firstly, the frame structure and characteristics of the virtual power plant are introduced, and the characteristics of the common distributed units are analyzed. On this basis, the prediction error model is established, and the white noise in the model is stratified by the Latin hypercube sampling method, which ensures the uniformity of the sample in the original distribution space. Then a large number of repeated or similar scenes in the original scenes are removed by the backward reduction technique based on scene distance and then generates a set of pre-set number of classic scenes.Secondly, scheduling model based on the maximum profit in the virtual power plant is established and the "component" constraint is proposed for the fair principle of the power grid and virtual power plant which ensure the interests of them. Subsequently, the problem that the intelligent optimization algorithm takes time and easy to fall into local optimal solution in solving the problem of high dimensional multiple optimization problems is improved by improving the generation of initial population and the weight coefficient in the process of particle evolution of the method of particle swarm optimization. Finally, the simulation analysis is carried out in a virtual power plant including wind power and the effectiveness and efficiency of the proposed method based on multi scene scheduling model is verified.