Research On Optimal Dispatching Strategy Of Integrated Energy System Based On DQN Algorithm

As a large number of distributed renewable energy sources are injected into the grid,the traditional single energy optimization utilization model is undergoing changes.The diversity and coupling of energy structure will have a great impact on the safety and stability of system operation.This promotes the mixed transmission of energy in the network in the form of multi-energy flow,forming a comprehensive energy system coupled by multiple energy sources.Moreover,the energy flow calculation of the integrated energy system is also a multivariate nonlinear equation,which is difficult to describe with specific mathematical equations.Therefore,in order to realize the dynamic optimization and adaptive operation of the integrated energy system and fully absorb and utilize renewable energy,it is particularly important to conduct energy optimization scheduling strategy research.Based on the basic framework of the integrated energy system,this paper proposes a simplified system model using graph theory,and uses the hierarchical control method to establish an integrated energy system optimization model based on reinforcement learning,and proposes to use DQN algorithm to solve the dynamic optimization problem of multi-energy systems.Combined with example simulation,corresponding strategies are proposed.The core work is summarized as follows:First,the development status of the integrated energy system and its optimization problems are proposed,the current mainstream optimization technologies are clarified,and the shortcomings are identified;the application of reinforcement learning algorithms in the energy field is described,and the work arrangement of this article is introduced;secondly,Analyze the topological structure of the integrated energy system,introduce the energy constraint model of its key components,and establish an integrated energy system model with an energy hub as the core based on the coupling characteristics of multi-energy flow,use graph theory to simplify the model and analyze multi-energy The transmission mode coupled in the energy hub forms a balance equation of energy flow,which lays the foundation for the following optimization analysis and simulation experiments;on this basis,a comprehensive energy system layering strategy is proposed,which divides the system into three layers and uses The Q learning algorithm calculates the power distribution of each layer to lay a good foundation for system optimization;finally,the Markov decision and reinforcement learning process is introduced,and a detailed application of DQN algorithm in the hierarchical integrated energy system optimization control model is proposed.Establish an example simulation model,apply the DQN distributed reinforcement learning algorithm to realize the parallel optimization of the integrated energy system,and propose a series of strategies including the conversion and grid connection of the distributed power generation system,the charging and discharging of the energy storage system,and the cutting-out of gas generating units.Under the premise of ensuring system operating costs,fully consume and utilize renewable energy.Finally,the full text is summarized and further prospected.