Research On Operation Control Method Of Distributed Generation Based On Cloud Edge Collaboration

The goal of “emission peak” and “carbon neutrality” in China has gradually transformed the power system into a new power system dominated by new energy.But distributed photovoltaic,wind power,electric vehicles and other distributed power and load penetration to the safe,reliable and efficient operation of the grid has brought new challenges.The unadjustability of new energy sources such as wind and photoelectric and the uncertainty of their output greatly reduce the flexibility of power grid,and its distributed attribute also increases the complexity of power grid dispatching.The development of advanced information and communication technologies provides support for the flexible management and control of distributed power sources.However,the optimal control of massive distributed power sources requires a large amount of computing power,and edge servers are difficult to meet the computing needs of optimization algorithms.In view of the above problems,this dissertation studies the whole process of distributed power operation management and control,constructs a "cloud-edge" collaborative optimization architecture that couples the information layer and the energy layer,and studies the allocation strategy of computing tasks in the cloud and edge.On this basis,the model parameter identification,load prediction and optimal control strategy for distributed power generation are studied,and a complete system for distributed power operation management and control is formed,and the research results of this paper are practiced in demonstration projects.The main research contents of this dissertation are as follows:(1)Aiming at the integration of the information layer and the energy layer in the operation and control of distributed power generation,an information-energy coupling architecture is proposed,and the implementation of the architecture is discussed in depth from the perspective of engineering practice.According to the characteristics of distributed power,the key equipment of distributed power connected to the power grid is analyzed,and the application in the demonstration project is shown.On this basis,a "cloud-edge" collaborative computing task optimization strategy for distributed power operation management and control is proposed,which fully considers factors such as computing delay and network communication delay,and maximizes the use of computing resources of cloud servers and edge servers.Compared with individual cloud computing,the computing efficiency is effectively improved.(2)Aiming at the characteristics of distributed power output characteristics that are time-varying and easily affected by the environment,a neural network-based distributed power model parameter identification method is proposed,which comprehensively considers the timing characteristics of distributed power output and the influence of other external factors.The self-adaptive adjustment of the model output is realized through the neural network.Compared with the traditional model identification method or the heuristic model identification method,the output accuracy of the method proposed in this dissertation is significantly improved.(3)Aiming at the problem that different flexible resources have different adjustable time scales in microgrid with distributed generation,a multi-time scale optimization operation strategy of the microgrid with distributed power generation is proposed.The proposed algorithm fully considers the actual business logic of microgrid optimization and regulation,and makes full use of different characteristics of adjustable resources through the coordinated optimization of day-ahead demand response,hour-level distributed power generation scheduling,and minute-level energy storage device control.Compared with the adjustment strategy that only considers a single type of resource,it effectively improves the operating efficiency of the system.(4)Aiming at the privacy information of multiple microgrids with distributed generation is not interoperability and needs collaborative optimization,a collaborative optimization strategy of multiple microgrids based on gradient interaction is proposed.The sub-gradient push algorithm is introduced to solve the economic collaborative optimization problem of multiple microgrids with distributed power sources in an environment where private information is not interoperable.Furthermore,the safety of the interconnected system of multiple microgrids is considered.The ADMM algorithm is introduced,and through the interaction of hyperparameters,the rapid coordination of multiple microgrids is achieved while ensuring economic efficiency,so as to maintain the frequency stability of the system.The research results solve the collaborative optimization problem of multiple distributed power microgrids in an environment where private information is not interoperable,make full use of the adjustment capabilities of distributed power generation belonging to different microgrids,and achieve comprehensive optimization of economy and security.