Application Of Consensus Of Multi-Agent System In Microgrids

In recent years,the problem of energy shortage is becoming increasingly serious,which hindered social and economic development.It is urgent to establish a clean and sustainable energy system and realize the energy structure represented by renewable energy.The microgrid integrates wind,solar and other renewable energies,which is not only clean and environmentally friendly,but also flexible in distribution.To further improve the operation efficiency and stability of the microgrid,this paper uses the theory of multi-agent consensus and inverse system to research the power distribution between wind turbine generators and batteries in microgrid.In addition,the power and torsional vibration controller of doubly-fed induction motor are designed.The main researches in this paper are as follows.Firstly,a power hierarchical control method is proposed to solve the problem of budget and control of island microgrid.The power control process is divided two layers in microgrid:The upper layer is the communication network between wind turbine generators and batteries,which can exchange local information with each other and realize power distribution;The bottom layer controls the power according to the power value allocated by the upper layer.Among them,the upper generators and batteries are modeled as a multiagent system,which used the consistency algorithm to allocate generators and batteries power according to the proportion of installed capacity and batteries state of charge respectively.The inverse system method is used in the wind turbine generators power control layer,and the neural network is used to approximate the inverse system.The wind turbine generator nonlinear system is combined into a pseudo-linear system,which simplifies the design of the controller,and the problem of the model inaccuracy is avoided using the neural network inverse system.The simulation results show the effectiveness of the proposed hierarchical control method.Then,a torsional vibration controller is proposed in this paper based on deep reinforcement learning for a wind turbine generator to improve the stability of the wind turbine without drivetrain mathematical model.The inputs of the controller are set as the speed difference of the blade and the rotor.The deep deterministic policy gradient algorithm is used to study the electromagnetic torque compensation strategy by design a reward function to suppress the wind turbine generator torsional vibration when the speed difference and phase angle are detected.The simulation results shows that the wind turbine generator torsional vibration controller presents a superior dynamic performance and effectively suppresses the oscillation caused by grid disturbance or indeterminacy in wind speed without an accurate system model.