Research On Operation And Control Method Of Wind And Solar Energy Storage System Based On Deep Reinforcement Learning

China is rich in renewable energy resources.The power system with wind energy,solar energy and hydropower as the main energy is the development form of energy and electricity in China in the future.However,renewable energy has intermittent and volatile shortcomings,and the high proportion of renewable energy penetration poses new challenges to the safe and stable operation of the power system.Pumped storage power station is the most mature,reliable,safe and large-scale energy storage technology at present.It plays an important supporting role in maintaining the safe and stable operation of power grid and building a new power system with an increasing proportion of new energy.It is of great significance to deepen the reform of power system to promote the complementarity of wind,water and other multi energy.Therefore,this paper introduces the wind light pumped storage hybrid energy system to improve the consumption capacity of renewable energy and ensure the safe,economic and stable operation of the hybrid energy system.In recent years,with the rise and rapid development of artificial intelligence,deep reinforcement learning algorithm also provides a new solution for intelligent operation scheduling of power system.As a data-driven method,when facing the solution of high-dimensional nonlinear complex problems,the agent is trained to make the optimal decision through the deep reinforcement learning algorithm to obtain the optimal scheduling strategy of the hybrid energy system.Supported by the national key R ＆ D Program "Research on Key Technologies of planning and stability control of weakly interconnected hybrid renewable energy system"（2018YFE0127600）,this paper carries out relevant research on the operation and control of wind and solar energy storage system.The main research contents of this subject include:（1）Intelligent scheduling strategy of wind light pumped storage hybrid energy system.In the environment of wind power generation,photovoltaic power generation and random fluctuation of load demand,the voltage control problem of hybrid energy system is solved on the premise of 100% consumption of renewable energy.The goal is to minimize the voltage deviation.The depth deterministic strategy gradient algorithm is used to solve the real-time scheduling of wind light pumped storage in uncertain environment,and the optimal scheduling strategy of reactive power of pumped storage power station is obtained.The actual data of a power plant for 365 days and 24 hours are used as the training data set to verify the feasibility and effectiveness of the model and method proposed in this paper.The comparative experimental analysis is carried out by using the traditional algorithm under the same conditions,which proves the superiority of the operation and control method of wind and solar energy storage system based on deep reinforcement learning.（2）Research on scheduling of wind light pumped storage carbon capture hybrid energy system.On the basis of wind light pumped storage hybrid energy system,carbon capture system is added as load to reduce carbon emission.Aiming at the goal of maximum carbon capture,an off-line learning strategy sac algorithm based on maximum entropy reinforcement learning framework is adopted.The agent is trained to obtain the optimal scheduling strategy through sac algorithm to achieve the purpose of maximum carbon capture.（3）Study on the operation economy of wind light pumped storage.At the same time,considering the impact of electricity price,wind power and photovoltaic power generation,the goal is to maximize the power sales revenue of hybrid energy system.Td3 algorithm can suppress continuous overestimation,and has a clear target strategy for smoothing.The agent is trained by td3 algorithm to obtain the intelligent dispatching strategy of pumped storage power station and solve the problem of system operation economy.