Research On Hierarchical Intelligent Control Of Hybrid Microgrid Cluster Based On Deep Reinforcement Learning

With the large-scale popularization of distributed clean energy,renewable energy penetrates into the distribution network in the form of microgrid,and a large number of hybrid microgrids are connected to the distribution network feeders to form a hybrid microgrid cluster.The clustering of microgrids can realize power complementarity and coordinated control among multiple microgrids,which solves the problem of small volume and poor regulation ability of a single microgrid.However,the microgrid cluster has problems such as complex energy transmission,inter-network communication delay,high power flow coupling and data dimension explosion,which makes it difficult to establish and solve the control and optimization model of the hybrid microgrid cluster.Aiming at the stability control and economic optimization of hybrid microgrid cluster,this paper proposes a multi-layer collaborative intelligent control system for microgrid cluster.Based on the hierarchical collaborative control framework,innovative achievements have been made in energy management and voltage stability control of microgrid cluster.This paper has carried out the following specific work on the above problems:(1)Based on the topology of hybrid microgrid cluster control system,the idea of hierarchical collaborative control of microgrid is extended to the level of hybrid microgrid cluster,and a hierarchical intelligent collaborative control system based on AC / DC hybrid microgrid cluster is proposed.The hierarchical control architecture of hybrid microgrid cluster is described in detail.Starting from the problems of high data dimension and high uncertainty in centralized control of microgrid,the intelligent control method based on neural network model training and its participation in microgrid control process are introduced.(2)Based on the intelligent control process of hybrid microgrid,aiming at the energy management problem of AC / DC hybrid microgrid in the local centralized control layer,the realtime energy management cost optimization model of AC / DC hybrid microgrid is constructed from the operation and maintenance cost and balance stability support cost of hybrid microgrid.Considering the multiple uncertainties of the distributed power load and the high dimension of information data,the DDPG algorithm is used to solve the energy optimization model,and the algorithm is improved by experience action guidance and exploration strategy optimization.Finally,through the training of the improved algorithm,a neural network model that can quickly realize the energy management decision of the hybrid microgrid in real time is obtained,which improves the economic benefits of the hybrid microgrid.(3)On the basis of hierarchical control of hybrid microgrid group,aiming at the problem of voltage fluctuation of interconnected nodes caused by source-load uncertainty in the global centralized control layer,a voltage stability control method of hybrid microgrid grid-connected nodes based on MADDPG is proposed.It alleviates power congestion and improves voltage quality by controlling the wind turbines of the microgrid cluster and the inverters of the energy storage system.The optimization model of node voltage stability control is constructed.Under the framework of ’centralized training and decentralized execution ’ of MARL,the neural network parameters are continuously updated through the continuous interaction between the agent and the environment to train the optimal strategy network.Finally,the effectiveness of the proposed method is verified by numerical simulation,which improves the voltage stability of the hybrid microgrid cluster and reduces the network loss.In this paper,a hierarchical cooperative control system of hybrid microgrid cluster is constructed.Based on the hierarchical cooperative control framework and deep reinforcement learning algorithm,the practical problems of energy management optimization and voltage stability control of microgrid cluster are solved,which provides a new method for the safe and stable operation and economic optimization of hybrid microgrid cluster.