Research Of Energy Dispatching Algorithm For Microgrid Based On Reinforcement Learning

Energy is an important guarantee for human survival.Energy conservation,emission reduction and clean energy have become an important direction for the development of the world’s energy economy.With the continuous development of information technology,microgrid as a green、sustainable and efficient power system,is widely regarded as an emerging direction with open technology platform and broad market prospect.In view of the continuous development of micro-grid technology,scholars from all over the world have also put forward many theories on its optimization and control strategy.Reinforcement learning is a popular optimization algorithm in machine learning,and it can also be applied to microgrid.For different types of microgrids,different mathematical models are constructed,and different optimization objectives are proposed.For multiple microgrids,a dynamic game model of electric energy is proposed.According to the above model,the continuous scheduling process is discretized and the optimal scheduling scheme is solved by combining the reinforcement learning algorithm.The contributions of this paper are as follows:1.On the basis of the reinforcement learning,specifically into the concept of asynchronous.Aiming at residential park micro power grid through analyze the model,and to determine the control method,constitutes the residents of electricity conductive zone scheduling policy,according to its load and the client at the same time,established the three optimization goal,and then through the theoretical analysis to determine the model of low dimension matrix.And at the same time the theory is validated by simulation.2.This paper puts forward the genetic algorithm with reinforcement learning,the implementation of the division of space to improve convergence speed and convergence precision.The Model is divided into two parts,one of them is the micro grid for single industrial park,and the its different with the residents of the distributed energy.For the optimization goal,general industrial park into consideration the safety performance.The second part describes the dynamic game problem among the microgrid systems in multiple industrial parks,and maximizes its own and total economic benefits through the transaction of electric energy.This chapter theoretically analyzes and proves the existence of the Nash equilibrium of the benefit function.,Because of the wind energyinto consideration,as well as V2 G and P2 G,so the model is of a higher dimension matrix.According to the mathematical model,in the simulation verified the idea,at the same time to improve the idea of parameter seriatim simulation test,and get the optimal parameters.