Optimal Dispatch Of Hydrogen-Electric Coupled System With High-Penetration Distributed Photovoltaic Generations

With the rapid development of new energy,the proportion of distributed photovoltaics in the power system is getting higher and higher,and it has become an important part of China’s future new power system.However,the problem of distributed photovoltaic integration and consumption is very prominent,and the problem of abandoned light needs to be solved urgently.The hydrogen-electric coupled system realizes the interconnection of electric energy and hydrogen energy,and is an effective way to consume intermittent renewable energy,and is one of the core technologies to achieve carbon neutrality and carbon emission reduction goals.Therefore,the hydrogen-electric coupled system with high-penetration distributed photovoltaic generations has become a hot topic in the current research on distributed photovoltaic consumption.However,on the one hand,traditional methods are difficult to guarantee the accuracy of predicting high penetration distributed photovoltaic power;on the other hand,the internal equipment of hydrogen-electric coupled systems is complex,involving multi-energy flow modeling analysis and application.Traditional modeling and calculation methods neglect the characteristics of hydrogen energy and cannot meet the economy and robustness of actual system operation.Therefore,this paper focuses on the optimal dispatch of hydrogen-electric coupled systems with high-penetration distributed photovoltaic generations,aiming to improve the level of consumption of renewable energy and the economic operation of the system.It targets two key points in the hydrogen-electric coupled system: single subject and multiple subjects,and designs a research framework of "prediction"-"internal system coherence"-"inter-system interaction",constructs an optimal dispatch theoretical system for the efficient and economic operation of hydrogen-electric coupled systems.The proposed theories are verified combined with actual hydrogen-electric coupled systems.Firstly,the power generation power of high-penetration distributed photovoltaic generations is easily affected by weather factors.Traditional prediction methods are difficult to analyze and predict their frequent fluctuation characteristics.To solve the problem,a mixed neural network photovoltaic power prediction method based on multi-exposure high-resolution ground cloud image is proposed.The image data and textual data are efficiently fused.A multi-time scale decoupled method based on guide operator is introduced to fuse high-resolution cloud maps at multiple exposures.Furthermore,the sliding window cutting method is used to generate multiple subgraphs to reduce the information loss caused by image sampling in traditional methods.The mixed neural network model based on CNN-LSTM maps the fused image sequence and photovoltaic power prediction value through data-driven methods.Based on the original image and the fused image separately,it is proved that the image fusion technology can increase the cloud map feature performance.Compared with other photovoltaic power prediction algorithms,the proposed algorithm in this paper has obvious advantages in various scenarios.Secondly,the hydrogen-electric coupled system includes multiple types of energy,and its operating mechanism has complex characteristics.Traditional optimal dispatch methods neglect the characteristics of hydrogen energy,which leads to the inability to achieve optimized operation of the system.To solve this problem,a fine-grained modeling method based on multiple time scales is proposed to improve the modeling accuracy of system optimization and scheduling.It can avoid infeasible problems during optimization and scheduling.On the other hand,there is uncertainty in the photovoltaic power prediction error.In order to ensure the robustness of the system optimization and control,the prediction uncertainty is considered in the fine-grained scheduling,and the original optimization constraints are analyzed by convex relaxation based on distributionally robust optimization.Compared with traditional optimal dispatch methods,the advantages of the proposed in scheduling accuracy and reliability are revealed.The distributionally robust optimization algorithm makes better use of the statistical distribution information of photovoltaic power prediction errors,balances the economic performance and reliability of system operation compared with robust optimization and random optimization methods.Finally,a single system cannot form stable consumption capacity for distributed photovoltaics due to its capacity constraints.It is necessary to study the joint body of multiple hydrogen-electric coupled systems and improve the consumption capacity of new energy through energy interaction.However,traditional methods mostly focus on the interconnection of electric energy and neglect the role of hydrogen energy interaction.To solve the above problems,this paper first deepens the analysis of coarse hydrogen energy interaction based on mathematical statistics and the potential for cost reduction in the system.An example of the construction for composite material hydrogen gas pipelines is under hydrogen energy interaction.Then,a targeted optimal dispatch strategy based on distributed peer interaction for multiple-subject hydrogenelectric coupled systems is proposed.The feasibility of multi-energy flow peer interaction based on Nash equilibrium game theory is constructed and proved through mathematical derivation.Considering the convergence problem of traditional distributed algorithms,an improved alternating direction method of multipliers algorithm is designed to accelerate the calculation speed of optimization.The simulation comparison proves that the inclusion of coarse hydrogen interaction in the energy trading of multiple-subject hydrogen-electric coupled systems can further reduce the operating costs of each hydrogen-electric coupled system.